aiLL June 12, 2023 - 4 Hours of Rabbit Holes, Epiphanies & Twerking Down the Generative AI Path

Description

Chapters

Transcript

0:02 um
0:19 hello hello it's Kyle Shannon with the
0:21 AI learning lab I think I'm gonna have a
0:24 hair situation tonight
0:26 let's see if we can put a little water
0:29 in there see if we can get it to behave
0:30 a little bit uh what's up what's
0:34 happening
0:36 what's your name
0:37 benviton that's a cool name do you use
0:41 tick tock streaming platform no because
0:44 I'm on a Mac and they made it for
0:45 Windows and they're like if you're on a
0:47 Mac go [ __ ] yourself
0:49 yeah
0:50 what are you gonna do uh I'm just using
0:53 the tick tock this is uh there's a dual
0:56 camera
0:57 option when you go live and that's
0:59 that's what this is uh welcome everyone
1:02 this is the AI learning lab if you have
1:04 questions about AI pop them in the
1:05 comments if you're new to AI I put these
1:08 URLs up here just as kind of a public
1:10 service announcement the top one is chat
1:12 GPT the next one is Bing which is gpt4
1:15 for free the next one is poe which is
1:18 you can play with a bunch of them
1:20 um
1:21 what do you use to get the picture in
1:23 picture
1:23 when you go live there's an option that
1:25 says dual camera and it just turns on
1:27 both cameras this is just my laptop
1:29 in my back facing camera
1:31 all right
1:33 see hand and same hand now it could be
1:37 AI this could all be recorded
1:41 um welcome everybody if you have
1:44 questions about AI pop them in the
1:45 comments I'll do my best to answer them
1:47 my name is Kyle Shannon I'm a long time
1:49 entrepreneur and a uh I'm the founder
1:52 and CEO of a company called storyvine
1:54 which is an automated video storytelling
1:56 platform does not use AI although we're
1:58 starting to roll a bunch of it in there
2:00 as you can imagine and
2:03 um
2:04 and I've been been an entrepreneur
2:06 entrepreneur for yeah getting my lawn
2:09 mode someone's getting their lawn mode I
2:11 am my own lawnmower I have my own lawn
2:14 service
2:16 um the the Nuggets game just started
2:18 they're here in Denver I'm in Denver
2:20 they're down two to zero
2:24 oh what a mess
2:26 all right
2:29 um yeah I guess I can close my window
2:31 but I like the air blowing so they
2:33 should be done soon enough it's it's
2:34 almost seven they start at seven in the
2:36 morning and they they go until seven at
2:38 night I don't know I don't know
2:41 is it always like this
2:43 um anyway uh I've been been an
2:45 entrepreneur for a while the the reason
2:47 the thing that got me so interested in
2:49 this generative AI stuff is it's got
2:51 very strong parallels to the early days
2:53 of the World Wide Web
2:56 um and
2:58 I started one of the first digital
3:00 agencies in 1994 a company called
3:02 agency.com and I kind of had a big
3:04 Epiphany about oh this world wide web
3:06 thing this is going to change things and
3:09 I haven't had an epiphany like that
3:10 until last year December of 2022 shortly
3:14 after Chad GPT came out I was like oh
3:17 this is one of those uh this is a big
3:20 big deal and that's what this channel is
3:21 about it's really about exploring what's
3:24 going on with all these tools and and
3:26 looking at what are the implications of
3:29 them how do you roll them into your
3:31 business how do you deal with the fact
3:33 that your job might
3:35 um very quickly be disrupted and how do
3:39 you how do you deal with that so that's
3:41 what this is all about so if you have
3:42 any questions pop them in the comments
3:44 below looks like there's a bunch of
3:45 people here my name is Kyle Shannon this
3:47 is the AI learning lab
3:49 these URLs here are really just there
3:51 for your convenience more than anything
3:53 that top one is chat GPT if you have not
3:55 played with chat GPT go play with it now
3:57 while you listen to me ramble
4:00 um go do that Bing is also chat GPT but
4:03 it's just Microsoft's version flavor of
4:05 it and then poe.com gives gives you
4:08 access to a bunch of other things I
4:11 strongly strongly strongly strongly
4:13 strongly strongly encourage you to go do
4:15 that
4:16 okay
4:18 leaving my experiences for future
4:20 through my avatar for my kids to ask
4:22 questions is yeah that's gonna that's
4:23 gonna be a thing and I think that's a
4:25 really interesting idea
4:27 um
4:28 I would say it's probably not going to
4:29 be that long
4:31 probably less than a year may definitely
4:34 less than two years where there will be
4:37 tools that make it very very easy to
4:39 kind of Point AI at your digital
4:42 footprint and say go make a you know an
4:45 avatar that is basically like me and
4:48 people can start interacting with it
4:51 um yeah it's a really good one hello
4:53 hello so let's see
4:56 um gcg creative what can you do with
4:59 chat GPT
5:01 um so start exploring it you can do lots
5:04 I'll go I'll go
5:06 um let's go play mailman
5:13 bad dog champ
5:17 this is this is what we deal with hey
5:19 Liz
5:20 uh yeah there's the mail lady
5:23 evil male lady she's out to get us I
5:26 don't know if you knew that
5:28 champ will tell you she's out to get us
5:30 what was I gonna do oh what what can you
5:32 do with chat gbt so let me go let me go
5:34 play a little bit I'm just gonna go to
5:36 GPT 3.5 which is the lyrics so so
5:41 um Chachi PT is
5:43 is good at a lot it can write really
5:45 well it can program it can code
5:49 it can create tables it can
5:53 um help you write marketing plans
5:55 business plans it can write content for
5:57 you
5:58 so so very often where I'll start is is
6:02 something that when you first see it
6:05 you're just like if you haven't seen it
6:06 before you're like oh that's pretty
6:08 amazing and it feels kind of cute but
6:11 then the deeper you dig the more
6:12 interesting it gets so so we'll just do
6:13 this
6:15 um okay right
6:17 um
6:18 write me so the the interface is
6:21 basically just like this box at the
6:22 bottom it's just kind of like a Google
6:23 box right write me a country
6:27 song about a fat black and
6:32 white dog that thinks
6:37 the male lady
6:40 is
6:42 out to get us
6:45 and then you just watch it right very
6:48 very quickly a whole country song now
6:50 you'll notice there's choruses and
6:52 verses there let's let's see what it did
6:55 in a small town There's a dog named Jack
6:58 oh wait
6:59 um so now I'm just gonna say
7:01 um the dogs
7:04 the dog's name is Champ
7:09 see we'll we'll have it so so now it's
7:11 rewriting the song for champ in a small
7:14 town There's a dog named champ black and
7:16 white fur with a belly so plump he's got
7:18 a heart as big as the sky but when the
7:20 mail lady comes by oh my oh champ the
7:23 fat black and white dog so so without
7:26 really giving it anything it wrote me
7:28 you know a clever little country song
7:31 um but then I could do something like
7:33 this I could say
7:35 make
7:36 make let's let's say I wanted to do
7:39 something for marketing make me a table
7:44 comparing
7:46 the
7:48 um
7:50 attributes of
7:53 major social let's see social media
7:59 platforms
8:01 for my small
8:04 I don't know gardening business
8:08 and again I'm not really giving it a
8:10 whole lot of context here I'm just
8:12 saying make me a table comparing some
8:13 attributes
8:15 of these things and boom there it is
8:18 so
8:20 you know it it very very quickly like
8:24 you can see that you can see very very
8:26 quickly that what this thing is doing is
8:29 different than other things the question
8:31 can it design a logo it can't but you
8:33 can use things like mid-journey and
8:35 stable diffusion that that can do that
8:38 what Chachi PT can do is describe it
8:41 um so so what what I can say is
8:44 um I have a
8:48 um business
8:51 um
8:52 that makes
8:55 Gardens
8:57 for really busy
9:00 people
9:03 can you describe
9:07 what
9:10 my logo might look like question mark
9:18 um
9:19 and actually before that I should
9:21 probably have it give us a name but
9:22 that's okay
9:24 it's thinking it's thinking
9:26 it looks like it it has hung when it
9:29 when it when it goes this long it
9:31 generally means it's hung
9:32 it just basically means their servers
9:34 are busy
9:36 um but what I would do is I would take
9:39 anything that you're interested in and
9:41 go just go to chatgpt go to oops not
9:45 that go to chat.openai.com
9:49 sign in get yourself an account or go to
9:52 bing.com and click the chat button and
9:54 start asking it to do things export the
9:58 table in CSV yep you can do that or you
10:00 can just copy and paste it over there
10:05 um uh
10:08 [Music]
10:11 let me see did it did it yeah network
10:13 error regenerate response here we go
10:16 so it's going to describe the logo
10:18 certainly here's a description logos
10:20 clean and modern design then
10:23 embodies the essence of a busy lifestyle
10:25 blah blah the central element is a
10:27 garden icon
10:29 okay so this is a big long explanation
10:31 I'm going to say
10:33 um
10:35 rewrite into a single image prompt
10:40 for a an AI to generate a logo
10:52 and so there's your image prompt a clean
10:54 modern design blah blah blah so now I
10:57 can just copy that I can go over to
11:00 this is a whole different can of worms
11:02 here but I'm going to go over to Discord
11:04 where I can make mid-journey images I'm
11:06 going to go slash imagine
11:09 and then I'm going to paste in that
11:11 description that it just gave me and
11:13 we'll see what it gives us
11:15 so this is now going to go off to their
11:17 servers to Mid Journey servers and it's
11:19 going to make some stuff so so I hope
11:21 that helps so so what I would encourage
11:23 you to do is just go start playing with
11:24 these things start asking it questions
11:26 start thinking about things that you
11:28 know or or problems that you need to
11:30 solve and have it start
11:33 think of it less like a Google Search
11:35 and more like you've got a remote
11:37 co-worker that's smarter than any other
11:40 co-worker you've ever worked with in
11:42 your life
11:43 that's kind of what it's like working
11:45 with chat gbt all right let me go see
11:48 some of the questions here
11:54 wording in chat GPT is difficult at
11:56 times can't find what I'm looking for I
11:59 don't know what you mean wording
12:01 can't find what you're looking for
12:03 um it it might have to so the the big
12:06 trick with with chat GPT is getting what
12:09 what's called prompt engineering or
12:11 prompt crafting down starting to get it
12:13 to
12:14 um
12:15 operate on your behalf now here's some
12:17 of the logos these are a little busy for
12:19 my taste but this gives you an idea so
12:21 we had chat GPT describe a logo of our
12:24 gardening business
12:27 um that I just you know quickly made up
12:29 on the spot without very many details
12:30 and it described a logo and now
12:33 mid-journey took that description
12:35 without any editing like I didn't tweak
12:37 it at all and it it generated some
12:39 things this is a gardening service for
12:41 busy people these are way too busy for
12:44 Logos but they're stylistically pretty
12:46 interesting so and the colors are kind
12:48 of interesting
12:50 so there you go
12:52 hello Anne how are you major gift
12:55 fundraising I'm at a conference where no
12:57 one has used chat gbt they think I'm
12:59 magic it's that is so fun when you when
13:02 you show someone chat GPT that's never
13:04 seen it before and they're like oh
13:08 like it still blows my mind when it does
13:11 like when it makes the table like if
13:12 you've ever tried to do any level of
13:15 research and create like a Google Sheets
13:17 table and fill in the data doing all the
13:20 research it takes you forever
13:23 um
13:24 the other thing the the the the this
13:28 lower chat prompts.chat that's a
13:30 document that will teach you how to
13:32 prompt one of the things that you can do
13:34 with chat GPT that may help the folks
13:37 that are saying they're not getting good
13:38 results with it you tell it to act like
13:41 something act as an accountant act as a
13:43 mathematician act as an expert marketer
13:45 act as a brander
13:47 um and in doing that you kind of limit
13:50 it's the world that it's looking at and
13:53 and then when you interact with it it
13:54 can it can often dramatically improve
13:57 the results
13:59 um what did you do with your evenings
14:00 before you started showing up here every
14:02 night um well so this it's fun what I've
14:05 actually done so for about the last year
14:07 and a half I'd get home after work I'd
14:09 have dinner and
14:11 um then I do Sim racing
14:13 for like a year and a half and so my sim
14:15 rig is over there and so I just decided
14:18 I'm I need to
14:21 figure this AI stuff out and so I just
14:23 thought why don't I just replace the
14:24 hour and a half I do racing with an hour
14:26 and a half doing this and maybe I can
14:29 help some people and you know
14:33 there's that
14:35 um okay what are you putting
14:38 wait what are you putting that under not
14:41 images
14:42 I don't understand what you mean I don't
14:44 understand the question there
14:46 all right
14:47 um
14:49 hey everybody if you're new here my name
14:51 is Kyle Shannon this is the AI learning
14:52 lab uh these URLs here are all different
14:55 flavors of chat GPT the top one is the
14:57 official chat website Bing is
15:00 Microsoft's flavor of that and then pose
15:03 another option that where you can do
15:05 chat GPT plus some other large language
15:08 models so go play with those and then
15:10 prompts.chat will teach you how to
15:11 prompt and future pedia is just a giant
15:13 directory of sites I'm not affiliated
15:15 with this [ __ ] in any way but it's there
15:17 as a service to get you started
15:21 okay let's see I think my shirt is bad
15:26 for this I can't read the text over my
15:27 shirt so I think I might go change into
15:29 a t-shirt
15:31 um I think they made an app for it that
15:33 can generate images on app I'm not sure
15:36 what you're talking about
15:38 does mid-journey
15:40 have the prompt to generate fill in the
15:43 background images like Adobe Photoshop
15:45 so does does mid-journey have out
15:48 painting
15:50 not really stable diffusion does
15:54 um Photoshop does with the generative
15:56 fill now and then there's a new thing
15:57 that is it open AI put out
16:02 what's it called it's called
16:05 um
16:06 uncrop
16:08 we can go find this sorry my brain I'm
16:11 just getting started on this tonight so
16:13 my brain is a little work was busy today
16:15 so I didn't do a lot of AI stuff
16:18 um uncrop
16:21 tool
16:24 stable diffusion so stable diffusion
16:27 okay so if you go to clipdrop.co
16:32 this is pretty cool
16:35 clipdrop.co
16:37 it does there's a bunch of tools here
16:39 like background remover and you can play
16:42 with stable diffusion but in the bottom
16:44 right hand corner there's this new thing
16:46 called uncrop
16:48 which you can try an example or you can
16:51 upload something so I'll I'll upload
16:54 something let me go to I'll go to my
16:55 finder
16:56 here on Mac
16:59 and
17:00 let's go to
17:04 caution and dreams Kyle Shannon
17:09 dreams
17:15 [Music]
17:20 it's gonna be so worth it it's gonna be
17:22 so awesome you're not even gonna believe
17:24 it
17:25 how cool this is all right
17:28 so
17:33 are these here are they up on iCloud I
17:35 have a feeling these are up on iCloud
17:40 let me go find Genesis hold please I
17:44 know what what did you do Monday night I
17:46 watched some guy try to find files on
17:48 his computer how long was he doing it
17:50 for about an hour
17:58 see I think I have too much crap on my
18:00 hard drive so let me see if this
18:02 actually works
18:03 it may not okay there it is yeah okay so
18:06 Jesus that took forever and and we lost
18:09 like you know 70 people that were like
18:11 screw this I'm not watching this so you
18:13 so you have an image here and then you
18:15 have this blue boundary
18:17 and you basically just go next
18:21 right
18:23 and now it's going to create four
18:25 different options to kind of fill in the
18:27 background
18:29 boom
18:30 there's one like friggin Magic
18:33 I mean literally like magic there's
18:35 another one that one's really cool right
18:37 with a little doorway
18:39 um there's one with sort of these plant
18:41 looking things plants or oil refineries
18:44 of the future I think I like the door
18:46 one that's pretty cool
18:48 um
18:49 so so uh this is
18:52 clipdrop.co uncrop it's free you can
18:56 just
18:56 dump any old image up there and uh it
19:00 will do that
19:02 it's amazing isn't it
19:04 all right let's see do they have Adobe
19:07 Photoshop with the generative fill do do
19:09 I have Adobe Photoshop with generative
19:11 fill in the background yeah I'll hop
19:12 over there so if you want to do this
19:15 inside Photoshop if you've got access to
19:17 Creative Cloud you go to the Creative
19:20 Cloud app and then you go to
19:23 um
19:25 you go to
19:27 beta apps
19:29 and you
19:32 um
19:33 download the Photoshop beta
19:36 so I'm going to kill some selected
19:39 layers there
19:40 hold okay so now I'm inside Photoshop
19:44 and this is pretty slick so what I'm
19:45 going to do is this this is a a
19:48 mid-journey image I made of myself in it
19:51 was a mid-journey image I I uploaded a
19:54 couple of photos of myself and told it
19:56 to make a Pixar character and this is
19:58 what it did and so now I just selected
20:00 this picture and the canvas is obviously
20:02 wider than the picture so now I'm going
20:04 to invert my selection
20:08 so now I've selected the two white areas
20:10 and then there's this new context
20:13 sensitive menu here or context
20:18 I don't know I don't know what the term
20:20 is my brain's not working right tonight
20:22 so I hit generate generate fill now I
20:24 can put in a description here so I could
20:26 put blurred
20:29 um background
20:31 image
20:33 um cartoon home office
20:38 or I could just put nothing it'll figure
20:40 it out but I I want there to be a
20:41 blurred home office there and so it's
20:45 thinking it's thinking it's thinking
20:47 it's thinking it's gonna be so good
20:50 you're gonna be like wow or it'll suck
20:52 completely oh that's not bad so there's
20:55 a it's not really blurred but that's a
20:57 home office but they give you three
20:58 options there's a second one that's
20:59 pretty good it's blurred but it took
21:01 some of these images that were above my
21:03 head and added weird stuff to them
21:06 and then that one added some sort of
21:09 that's actually not bad it's kind of
21:10 like a white wall with a TV on it and
21:12 then I can do things like I can go grab
21:14 like these two frames behind my head
21:17 and I can kind of highlight those
21:20 and say let's say generate fill let me
21:24 type in remove
21:26 frames
21:30 or remove pictures
21:34 and we'll see if it does that
21:37 it'll be interesting if it does it
21:42 so this is generative fill so again you
21:44 get to this by going to your Creative
21:45 Cloud go to the beta apps and then
21:47 download the beta version of Photoshop
21:50 now it turned the the images into some
21:53 sort of weird hook thing
21:55 they're made them square and not in
21:57 perspective and there it made kind of a
21:59 little mini putt-putt game behind my
22:01 head anyway so you mess around with it
22:03 but that's generative fill it's it's
22:04 pretty pretty remarkable
22:06 um as are a lot of these tools
22:09 um so if you have questions about AI pop
22:12 them down below if you're new here
22:15 welcome my name is Kyle Shannon this is
22:17 the AI learning lab I am learning about
22:18 Ai and I'm in a lab that looks a lot
22:21 like a home office cheers
22:26 foreign
22:29 okay let's go answer some other
22:32 questions oh by the way if you haven't
22:34 been here before I try to answer all the
22:36 questions so what ends up happening is I
22:38 get really far behind
22:40 there are 29 messages ahead of where I
22:42 am right now so if it looks like I'm a
22:44 recording I'm not I'm just slowly
22:46 getting through the questions
22:47 can you show how
22:50 you would look
22:52 can you show how it looked how I would
22:54 do t-shirt designs well it depends what
22:57 you want the T-shirt design to be
23:00 I think what I would probably
23:02 so when you have a problem you want to
23:05 solve like one of the things to do is
23:06 understand okay what tools are available
23:08 for you well we obviously want to go
23:12 um
23:13 we want to go to a tool like mid-journey
23:14 to create images but we might want ideas
23:18 for like what would be some cool t-shirt
23:19 designs so
23:21 let's go to chat GPT
23:25 and again I'm just going to use GPT 3.5
23:28 which is the free version of it because
23:30 it's super fast
23:32 um and I'm gonna say
23:34 um
23:35 you are an expert t-shirt
23:41 designer and writer
23:46 um
23:47 please
23:49 help me
23:51 come up with some t-shirt
23:56 designs and sayings
24:01 that will be funny
24:05 and memorable and sell Lots
24:11 because I want to make money
24:16 by the way any of the tick tock channels
24:18 that say that ignore them they're all
24:20 full of [ __ ] okay
24:21 that will sell Lots
24:24 um
24:26 I want to make t-shirts
24:33 for Jen
24:35 X moms and dads
24:39 who are sarcastic
24:43 and bitter
24:46 but loving
24:48 can you give me
24:52 a table with
24:55 10 mom
24:58 t-shirt
25:00 sayings wait sayings and and a company
25:04 accompanying
25:10 um image description
25:13 image description and 10
25:17 for dads okay that's really horrible
25:21 English but I just wrote a bunch of
25:23 stuff and
25:25 so it's now making me a table
25:28 and let's go see if these are any good
25:29 okay sure here's a table with 10 t-shirt
25:31 designs for Gen X moms and accompanying
25:34 descriptions followed by another table
25:36 of 10 for Gen X dads sarcas sarcasm
25:39 level Mom that's pretty good that's
25:41 pretty good actually a speech bubble
25:44 with the word uh with the text Mom
25:47 surrounded by a lightning bolt
25:48 representing sarcasm that's not as good
25:51 um
25:52 I'm not a regular Mom I'm a cool Mom
25:54 just another day of raising Geniuses
25:56 that's pretty good master of
25:58 multitasking and sarcasm
26:01 warning approach with coffee not bad
26:04 my patience is on coffee break sarcasm
26:07 is my love language motherhood the
26:09 toughest jobs let's go look at the dads
26:12 um I'm the dad in
26:15 sarcastic dad
26:17 what that doesn't make any sense at all
26:20 dad jokes are my superpower official
26:22 taxi driver for the family that would be
26:24 Uber driver we would update that for for
26:26 for the modern era you know what I'm
26:28 going to do this this should be fun I'm
26:30 gonna I'm gonna copy this prompt exactly
26:33 and I'm going to go over to gpt4 and see
26:36 if they're any better I have a feeling
26:37 gpt4 is going to be better because gpt4
26:41 is better
26:42 it just is it's just way better so take
26:44 the exact same prompt
26:47 it'll be slower
26:49 but here we go
26:55 and it did the table differently it did
26:56 moms and dads as two different columns
26:59 mom mom because therapy is expensive mom
27:03 fuel with a picture of a coffee mug
27:05 that's pretty good my nickname is Mom my
27:08 full name is Mom Mom I don't know what
27:10 that means raising kids my cardio let's
27:13 see the dad ones Dad aka at home I.T
27:16 support Grill Sergeant Grill Sergeant is
27:19 a good t-shirt
27:21 world's okay it's Dad I've seen that one
27:23 before the datalorian the best in the
27:26 parsec sarcasm one of the services I
27:29 offer that's good so these are these are
27:31 a bit better
27:33 um image of a man surrounded by toys so
27:35 so then you can do things like okay
27:38 we'll say
27:39 let's see
27:43 uh
27:46 Work Hard Play Hard nap harder picture
27:49 of a father sleeping on a sofa so I'm
27:51 going to take a picture of a father
27:53 sleeping on a sofa
27:55 and then I'm going to go over to Discord
27:59 for Mid Journey I'm going to go slash
28:01 imagine
28:04 and I'm going to go t-shirt
28:07 design let's let's say
28:11 um
28:12 80s t-shirt design
28:15 of and then I'm going to paste that in
28:17 picture of a father sleeping on a sofa
28:20 um
28:21 Pixar Style
28:24 just because I know what that is and
28:26 then I'm going to go
28:29 um I'll just leave it blank I think
28:31 it'll give us a square image
28:33 now let's just see what that does
28:36 so if I wanted to do t-shirts I would
28:38 you know probably do some version of
28:40 what I just did I come up with a bunch
28:41 of t-shirt
28:43 writing ideas come up with the images
28:45 and then I would come to something like
28:47 mid journey and start playing around
28:50 um you know ideating basically
28:53 so I'm just I'm doing kind of a single
28:55 pass here without really thinking about
28:57 it so
29:01 we'll see if these look like 80 t-shirts
29:03 the one on the upper right starting to
29:04 look like
29:05 some 80s t-shirt kind of thing
29:19 looks like there's some new people here
29:21 while these images build my name is Kyle
29:23 Shannon this is the AI learning lab if
29:24 you have questions about AI pop them
29:26 below
29:27 if you are new to AI altogether
29:31 go to these places go to chat.openai.com
29:34 that's the official chatgpt website
29:37 and bing.com is chatgpt4 for free you
29:41 can pay for it at openai or you can get
29:43 it for free at bing.com and then poe.com
29:46 is another place where you can play with
29:48 chat GPT and some other things so those
29:50 are just places to play with chat gbt
29:52 prompts.chat will teach you how to
29:54 prompt those tools
29:57 um really well there's just some good
29:59 resources there and then futurepedia.io
30:01 will give you
30:04 um access to a bunch of different tools
30:06 so let's go look at these that kind of
30:08 looks like Jerry Garcia that's pretty
30:10 funny
30:12 this old dude
30:14 and what was the
30:19 this was
30:21 Work Hard Play Hard nap harder
30:24 and it looks like that that's pretty
30:28 good
30:29 that's a dad with an iPad that's okay
30:31 that's not great I don't know why they
30:33 all look like Jerry Garcia this guy
30:34 looks like a hoarder like you could
30:36 probably do something around hoarding
30:38 like you know
30:39 now that the kids have left the house
30:41 you know I don't know whatever
30:42 but but you know that's that's a
30:45 starting point for you hopefully
30:47 all right
30:51 all right we're
30:55 let it know where you will tell sell the
30:57 t-shirts oh that's a good point because
30:58 it'll get target audience and stuff like
31:00 that
31:02 um
31:03 yeah one of the things about working
31:05 with chat EBT in particular is
31:09 don't think of it so much as you give it
31:11 one prompt and you get back an answer
31:12 it's more like this collaborative back
31:14 and forth with the tool it's almost like
31:16 you've got the world's smartest intern
31:18 working for you
31:19 and you can just tell the intern to go
31:21 do a bunch of [ __ ] go research some
31:23 stuff go make me a table go write this
31:26 code oh that's the other thing if you
31:28 haven't seen chat gbt before here's
31:30 something that'll blow your mind or blew
31:31 my mind
31:32 so go new chat and I'll just say
31:36 um write me a C sharp application
31:41 that is a clone
31:45 of Flappy Birds
31:49 let's see but
31:51 um
31:53 uses Space Invaders
31:58 game mechanics
32:04 certainly
32:06 there you go did you
32:09 and that will probably run
32:14 so it says this c-sharp application
32:16 creates a console based game that
32:18 combines the mechanics of Flappy Bird
32:19 and Space Invaders the player's
32:22 objective is to navigate a spaceship
32:23 bracket through a series of obstacles
32:26 represented by X characters
32:28 the player can use the space bar okay
32:30 let's say
32:32 make sure this can run in unity and can
32:37 you add
32:40 um character sprites
32:43 and explain how I should use them
32:51 all right so there it's saying
32:54 what to do in unity
32:56 there it's giving me my code
33:02 save the script create an empty
33:06 game object create a new c-sharp script
33:09 name it obstacles attach the script
33:12 continue generating
33:15 save the script so so like
33:19 what did I type in there two sentences
33:21 and it gave me a whole bunch of code so
33:23 if you want to learn coding
33:25 or if you want to learn game development
33:27 or if you want to just go into I don't
33:31 know Unreal Engine 5 and create a
33:33 particle system you can just have it
33:35 write this and then go start
33:36 experimenting and if you get lost and
33:38 don't know how to do something ask it
33:40 and it will tell you it'll create little
33:41 tutorials for you so it can do a lot it
33:45 can do a lot and
33:48 all right so getting back here looks
33:50 like we got some new folks just joined
33:52 welcome everybody my name is Kyle
33:54 Shannon this is the AI learning lab I'm
33:57 learning about Ai and doing what I can
33:58 to pass along
34:00 my knowledge and do it in an
34:02 entertaining way which often looks like
34:05 ridiculous
34:07 because because I just am dumb that way
34:10 I think they made an effort oh I fell
34:12 way behind in the comments hold please
34:14 let me catch you up to where I was
34:15 um
34:17 I did t-shirt designs
34:19 peeps hit the hearts thank you Anne yeah
34:22 feel free to hit the hearts I think
34:23 that'll get people to come here and go
34:25 why is this old guy on tick tock
34:29 is that Rosie O'Donnell
34:32 uh
34:33 I signed for chat gbt a couple of months
34:36 ago and I have and haven't gotten it yet
34:40 so how can I get it ASAP okay there's
34:43 nothing to get so you put your web
34:45 browser to chat.openai.com if you went
34:48 anywhere else
34:50 you were
34:52 um
34:53 you were not at chat you were not at the
34:55 official chat gbt website
34:58 um so go that's why I have these here
35:01 chat.openai.com now if you're in a
35:04 country that blocks it it might not be
35:06 there but then I would try bing.com so
35:09 bing.com if I'll just go and
35:12 when you first get to Bing you'll see a
35:16 search window just like a Google search
35:17 window but this in the upper corner next
35:19 to the Bing logo is this chat button
35:21 click that chat button
35:24 and that will pop you here and this is
35:26 chat GPT it's gpt4 connected to the
35:29 internet you can you can choose more
35:32 creative more balanced or more precise
35:35 depending on what you're trying to do
35:36 and then you just treat it just like
35:38 chat gbt right right me
35:43 a poem about a jolly
35:51 chubby
35:53 black
35:56 and white doggy
36:01 and it's a little slower than uh chat
36:05 CPT but
36:08 there's your poem there once was a
36:10 doggie so jolly and chubby he loved to
36:12 play fetch and snuggle with his buddy
36:14 his fur was black and white like a panda
36:16 bear he had a big smile and a wagging
36:19 tail to share this is more like a
36:20 children's book but it's a pretty good
36:22 one I didn't really give it any context
36:24 so I did use words like Jolly and chubby
36:27 and doggy so yeah it makes sense that it
36:29 would sound like that so that's Bing
36:32 so if if for whatever reason chat.openai
36:35 is not working then go to Bing
36:38 um and yeah I don't know where you
36:40 signed up for it but if it wasn't at
36:42 chat.openai.com it was probably at some
36:44 ripoff site there's a lot of them
36:46 there's a lot of ripoff apps as well
36:49 I'm on
36:51 Chatster
36:53 all right I don't know what Chatster is
36:54 I signed up for chat GPT a couple of
36:57 months ago and I haven't got it yet okay
36:59 I already asked that can you ask chat
37:01 GPT to act as a social media manager
37:05 and I will behave wait and oh and it
37:10 will be the behave the same way as a
37:11 prompt yeah so if you go to prompts.chat
37:14 what that does I'll go show you that
37:15 really quick
37:20 it's gonna be anything interesting to
37:22 look at but it's called awesome
37:24 GPT prompts let me make the text a
37:26 little bigger here
37:28 um the top of this document sort of
37:30 teaches you about chat GPT and prompting
37:33 and and what this all is about how to
37:35 want to write effective prompts there's
37:37 a free ebook
37:39 um there's other resources there they
37:42 they talk about the the chat GPT
37:45 um app and how it works and then this is
37:49 just kind of this never-ending list of
37:51 act as an English translator act as a
37:55 JavaScript console act as an Excel sheet
37:57 act as a travel guide act as a character
38:01 from a movie act as an Advertiser right
38:03 so this just goes it literally just goes
38:05 on and on and on so the these are all
38:08 different ways to get chat chat GPT to
38:12 act like a particular specialist
38:15 um and then you can you can start to
38:18 you'll start to understand from this
38:20 site a lot of the things it makes
38:21 possible
38:24 so hopefully that helps feel free to if
38:27 uh tap the screen share this live do
38:29 that sort of stuff and uh
38:33 and then we can do that impressive damn
38:37 what's the best AI tool to create
38:39 Instagram posts for my business chat GPS
38:44 you go to
38:47 chat GPT
38:49 we'll just do 3.5 since that's the free
38:52 one
38:53 Instagram posts I don't know what your
38:55 business is but we're going to make it
38:56 up let's say your business is
39:00 um you make pizza out of organic
39:03 eggplant I
39:06 uh have a pizza business
39:11 that
39:13 only uses organic eggplants God that
39:19 sounds disgusting
39:21 um
39:23 um give me 10 names
39:28 let's get some names
39:30 um Farm to Pizza no naturally slice
39:32 Earthly garden delights eggplantia okay
39:35 this is good
39:37 eggplantia it it is please act as my
39:44 social media
39:46 marketing
39:51 typing like write me 10 Instagram posts
40:00 um
40:01 appropriate for Denver this time of year
40:08 um I will be
40:12 doing a BOGO
40:15 buy one get one
40:17 but let's see if it knows BOGO I will be
40:19 doing a BOGO special
40:22 next weekend
40:25 Bang
40:29 all right
40:31 so it's writing Instagram posts with
40:34 little
40:34 emojis in them
40:37 Savor the Summers Savor the summer with
40:39 our delicious organic eggplant pizzas
40:41 get ready for our exciting BOGO special
40:43 coming up next weekend stay tuned for
40:45 more details eggplantia organic Pizza
40:48 Denver eats like it got the Denver thing
40:51 in there hot summer days call for
40:53 refreshing mouth-watering pizzas now
40:55 that doesn't seem right to me hot summer
40:57 days call for like lemonade but whatever
40:59 okay enjoy the vibrant Taste of organic
41:02 eggplant made with love and care I would
41:05 think it would be celebrate the season
41:07 of sun and great food indulge your
41:09 scrumptious and so and so here's an
41:11 interesting thing
41:13 um
41:14 if you hate what it wrote it wrote there
41:16 you can say just like you have you know
41:18 someone you work with you could say
41:21 um no this misses the mark
41:25 I hate emojis
41:28 and
41:30 I want the posts to seem like or uh like
41:37 organic content
41:41 versus sales messages
41:46 um
41:47 what if this were more about my
41:52 employees
41:53 give me
41:55 10 posts
41:59 for that
42:00 right again not very well written
42:02 apologies for the misunderstanding so
42:04 one of the things Chachi BT does look
42:06 how fast it's writing I mean it's
42:08 mind-blowing whether it's good or bad
42:10 you you can get to like 80 percent
42:13 completion of you know you're right
42:16 wanting to write Instagram posts in like
42:18 a minute and then you can spend the next
42:20 hour refining them and or or just doing
42:22 10 different iterations of this so let's
42:24 go see if these got better but but
42:26 notice what it does apologies for my
42:28 misunderstanding here 10 Instagram posts
42:30 highlighting your employees creating
42:32 organic content
42:34 creating an organic content feel for
42:37 your eggplantia pizza business meet the
42:39 incredible team behind eggplantia's
42:41 organic Pizza magic it still gave me an
42:43 emoji but whatever from our talented
42:46 chefs to our friendly servers each
42:48 member plays a vital role
42:50 uh attic planty are our employees are
42:53 the heart and soul of our business
42:56 um so I don't like those either so let's
42:58 go
42:59 um no not right oops wait no not
43:05 right please make
43:10 each post and intro
43:15 production to one of the employees
43:25 apologies from a misunderstanding
43:29 meet Emily our Master doe Whisperer say
43:32 hello to Mark our passionate sauce
43:33 Maestro meet Sarah our veggie virtuoso
43:36 like these are these are cute right say
43:39 hi to Alex our cheese connoisseur meet
43:41 Javier our girl Master extraordinaire
43:43 meet Maria our topping artist
43:46 so so yeah so so chat GPT can do like I
43:50 would say anything word related
43:53 start with chat GPT there might be more
43:55 specialized tools but
43:57 um but very often you don't need them uh
44:00 so if you're new here and and you just
44:02 came in on me babbling on about writing
44:04 Instagram posts in chat gbt that's what
44:07 I was doing
44:09 um my name is Kyle Shannon this is the
44:10 AI learning lab these are just some
44:12 places to go play with chat GPT I'm not
44:15 affiliated with affiliated with any of
44:17 those I'm not making any money there
44:19 those are just the the top ones the
44:20 official chat gbt website
44:22 okay
44:24 um
44:26 what site is this please the one that I
44:29 was I don't know what you're asking
44:30 about
44:32 um
44:33 address agai
44:36 oh A.G I'll talk about AGI in a second
44:39 what was that clip drop something
44:41 clipdrop.co
44:44 so if you go to where you put in a small
44:46 image and then you can make it like
44:47 landscape and just tell it to uncrop it
44:49 and it'll fill in all the details
44:51 it's pretty remarkable
44:55 website again clipdrop.co
44:58 slash
45:00 uncrop
45:02 for that specific tool
45:06 clip drop says this is awesome great
45:09 Professor Sonia thank you I appreciate
45:11 that clip drop says something about an
45:13 API key do I understand
45:16 wait I don't understand API it shouldn't
45:21 it shouldn't need wait
45:23 it shouldn't need a
45:29 let me see if let me see if I logged in
45:31 maybe I'm logged in and I give it some
45:34 sort of API key but this is stable
45:36 diffusion I don't think
45:38 sign in sign up no
45:40 maybe you went to clipdrop.com that's
45:43 not the site clipdrop.co
45:46 here wait hold
45:50 there's more than one way to skin this
45:51 cat we're gonna go this we're gonna go
45:54 this
45:55 we're gonna go this
45:59 we're gonna go
46:01 where's my
46:04 view where's my tools
46:07 there we go
46:08 clip
46:13 drop
46:15 dot Co CL
46:17 so that's where you should go and then
46:19 slash uncrop for that tool it shouldn't
46:22 be asking you for an API
46:25 that is probably just the wrong website
46:29 all right is this mid-journey
46:32 clipdrop.co is not mid-journey it's
46:34 stable diffusion I saw Tick Tock that
46:36 Matt that matched the AI
46:38 to a Pink Floyd song it was surreal how
46:41 is that done
46:42 um
46:44 they I don't I don't know I've haven't
46:47 seen it but there's there's a thing
46:48 called Runway ml that you can do you can
46:52 take a video and then
46:54 turn that video into something else like
46:57 you can make it like you walking down
47:00 the street and then you can turn it into
47:01 a claymation model of you walking down
47:04 the street is is Gen one of Runway ml so
47:08 if you go to the go to the App Store and
47:11 get the runway ml app that does some of
47:14 that video stuff I assume that's
47:16 probably what they used
47:17 but there's lots of tools now the the
47:20 the the reason that I put up this other
47:25 hang on
47:31 here I come to save the day
47:34 the reason that I put up
47:36 futurepedia.io is that's a big directory
47:40 of
47:41 um AI sites so just go there
47:46 and look up animation film movies things
47:49 like that that'll help you find that
47:51 stuff
47:55 why does it look like AI why does what
47:57 look like AI not sure what that means
48:00 what is a good AI tool to upload and
48:03 simplify legal documents well
48:07 so you can go to futurepedia.io
48:11 um there's there's a problem there's a
48:15 problem right now with chat GPT so you
48:17 can just copy and paste
48:20 um documents into chat GPT if you pay
48:24 for the
48:25 um
48:27 if you pay for the chat GPT plus so the
48:30 20 bucks a month version you get so some
48:34 of the plugins here there's there's one
48:36 called
48:37 um
48:38 link reader and there's one called ask
48:41 your PDF
48:42 and so generally what I'll do is I'll
48:45 turn on link reader and ask your PDF
48:48 and then if I give it
48:51 a URL of a PDF file it'll read it and I
48:54 can start interacting with that I don't
48:55 know what the limit is and on how big
48:57 they are but you can just do it right
48:59 within chat EBT using plugins
49:02 and I can I can talk more about them if
49:04 it if it makes sense
49:07 um let it know where you'll sell the
49:09 t-shirts okay I'm still back at the
49:11 T-shirt questions are you are you on
49:14 open AI right now I am
49:17 yes
49:18 yes
49:19 um how can I get chat CBD to create
49:21 illustrations for a children's book
49:23 well you can get chat gbt to write the
49:25 children's book so let's go ahead and do
49:27 that
49:28 um let's say write a
49:32 children's
49:33 book for
49:36 kids aged I don't know four to eight
49:40 about
49:44 a
49:46 chubby
49:50 dog named champ
49:54 that barks
49:57 at everyone but really just wants to
50:02 play
50:05 bang it should just write it we should
50:08 have a solid book here although I'm in
50:10 the plugins thing so it might be trying
50:11 to figure out if it needs to use a
50:13 plug-in
50:14 it didn't once upon a time in a sunny
50:17 little town there was a chubby little
50:18 dog named champ champ was not just any
50:20 dog he was dog with a big heart and even
50:22 bigger bark his bark echoed through the
50:24 town from the Red Roof Bakery to The
50:27 Blue Door post office okay so that's
50:29 perfect
50:30 so
50:32 what I would do in this case so I'd have
50:34 it I'd have it write the book and then
50:36 you need to go to some sort of image
50:38 creation tool so there's Adobe Firefly
50:42 which is if you've got access to the
50:45 Adobe suite I I don't know if you can
50:47 use it for free but we could we could
50:49 you can try that there's mid-journey
50:53 there's
50:55 um
50:56 Bing you can actually generate images on
50:59 Bing but what I would do I'll just do
51:01 this I'll do it on Firefly because it's
51:04 pretty fast
51:05 so let's see
51:07 a chubby black and white dog
51:11 the red roofed Bakery Okay so
51:15 so let me go here and go to
51:21 um firefly.adobe.com
51:24 and then they've got text to images
51:28 and down at the bottom we're going to
51:30 say
51:31 chubby
51:33 black and white
51:36 dog
51:38 um walking past a red roofed
51:42 Bakery
51:44 in a small town
51:48 and what I'm going to do at the
51:49 beginning of this I'm going to do
51:52 um
51:52 children's book illustration
52:00 of a chubby black dog so let's see what
52:03 that does
52:06 so you don't have to put in much and it
52:08 like if I had more time with this I
52:09 could spend more time okay I don't like
52:11 that illustration style but
52:13 like I can do
52:15 over in the right hand side with Adobe
52:17 like right now it shows the art style
52:20 but I can I can turn this into a photo
52:23 and it'll sort of recreate these as
52:25 photographs rather than illustrations
52:29 oh they're not the oh because I said a
52:31 children's book illustration now I'm
52:33 going to say graphic
52:35 we'll see if how this changes it
52:40 that's okay and then I'm going to do
52:42 none for the content type see what this
52:45 does so I'm really just experimenting
52:47 here
52:49 those are okay let's do layered paper
52:52 layered paper is one of these so I'm
52:55 going to now regenerate this and
52:57 hopefully it'll look like layered paper
52:58 that could be a cool style so you need
53:01 to find a style that you like
53:04 and then you need to get your prompts
53:06 yeah that's kind of a cool little style
53:07 so so you know there's the beginning of
53:10 you know illustrations for a children's
53:11 book about my fat dog walking around
53:14 town there's also a site called
53:16 lexicon.ai
53:20 Lexica
53:22 or no sorry lexicon.art
53:25 and what this is is a bunch of stable
53:28 diffusion images so you can just scroll
53:31 forever but I can do
53:33 um
53:35 children's book
53:38 as the search term
53:40 and let's see children's book
53:42 illustration
53:45 dog
53:49 all right and so now I can just start to
53:51 scroll through here and see if I find
53:53 any styles that I like
53:55 and okay here's kind of a cool one this
53:58 is sort of a wolf and a little boy and
54:00 then what it says up here is here's the
54:03 prompt that made this up and I can even
54:05 open it in the editor this this site's
54:08 not free you eventually have to pay for
54:09 this I think you can do some of them for
54:11 free but you can just come here and sort
54:13 of learn different styles that people
54:15 are using and and the prompts that
54:17 they're using to get those Styles so
54:19 that's another way to to get that so I
54:21 hope that helps
54:26 um
54:28 all right
54:30 what's the image one
54:33 you're using called I don't know which I
54:35 think it was mid-journey which which was
54:37 the other one all right it just jumped
54:38 me to the bottom of the so so I'm trying
54:41 to go through the the comments here one
54:43 by one I I I think I'm far enough behind
54:46 I'm just going to skip ahead for a
54:48 second here so if I missed your comment
54:50 sorry about that all right
54:53 um is there a way to make AI generated
54:55 videos text to vid yes so there's a
54:58 there's a
55:00 a company called Runway ml
55:03 and they've got these new text-to-video
55:06 things called gen 1 and gen 2. well gen
55:09 1 is video to video so you can take the
55:12 the problem with them right now is
55:14 they're limited to three seconds or four
55:16 seconds
55:17 so gen one is you take a video and you
55:20 upload it and then you can change the
55:22 look and feel of it and then Gen 2 is
55:24 text to video so
55:27 um they do have an app but I don't think
55:29 Gen 2 is part of all the apps yet they
55:31 haven't rolled it out to everyone yet
55:32 but you can go to their Discord and you
55:35 can play with it so I will go to the
55:37 Discord and play with it we'll do this
55:39 real time we'll do this live these
55:42 things down the left are servers so
55:44 these are like different communities
55:46 that I'm a part of
55:47 um discord's very much like slack it
55:49 started in the gamer Community slack
55:51 stole it so it's very much like slack
55:53 and then within one of these
55:56 um servers there are different channels
55:58 so this third one I have down here is
56:00 Runway
56:01 and then
56:03 um
56:04 Gen 2 Beta research so I you may have to
56:08 sign up for it I think they made this
56:09 available for everyone though but in
56:11 this Gen 2 area there's these generate
56:14 channels right so I'm going to go into
56:16 one of these generate channels and then
56:18 I'm going to put
56:20 at
56:22 Gen 2 gen Dash 2
56:27 and I'm going to hit return
56:29 and then now I'm going to do a prompt
56:31 I'm going to do
56:33 a
56:34 chubby
56:37 black and white dog and by the way I'm
56:40 not fat shaming my dog he's chubby all
56:42 right he's he's a stocky fella would
56:44 stocky make you feel better a stocky
56:48 black and white dog
56:52 um
56:54 with
56:56 black ears and white
56:59 face walks by a red
57:04 roofed
57:06 Bakery
57:09 in a small town
57:13 and so now
57:15 it goes off somewhere
57:17 and what what happens here in Discord is
57:20 so I was in the generate two
57:22 um
57:24 Channel
57:25 I think that's called a channel and then
57:27 it created this thread now this thread
57:29 that's my video
57:31 and so there's the prompt that it made
57:33 so it took my prompt and made that as
57:35 the text prompt and then it used some of
57:37 these default you know technical things
57:39 that it uses
57:42 and if what you're asking right now is
57:44 do you really have to do this in Discord
57:45 with mid-journey and with this thing
57:47 right now yes and if your next question
57:50 is isn't that a disaster from an
57:52 interface standpoint yes but we're very
57:55 early with this AI stuff so that that's
57:57 part of the part of the trade-off with
57:59 this AI stuff is
58:02 um we're very very early it's exciting
58:04 but the tools as a result are kind of
58:08 awful so we'll come back oh there it is
58:10 so so there
58:12 is let's see what the video looks like
58:16 he's not really walking he looks like
58:18 he's paralyzed but but a car walk a car
58:21 drove by watch the car in the window
58:23 there's a car in the window
58:25 that's really funny
58:28 those are weird dogs
58:34 very strange yeah there's three dogs and
58:36 they're very weird
58:38 um but they don't they don't look unlike
58:40 my dog
58:42 um anyway it looks like we've got some
58:43 new folks here my name is Kyle Shannon
58:44 this is the AI learning lab if you have
58:46 questions about generative AI pop it
58:48 down below these URLs right here are
58:51 chat GPT and different flavors of chat
58:53 at the top and then a couple of
58:54 resources at the bottom so I would
58:57 strongly encourage you to
58:59 um go check those out also tap the heart
59:02 screen I think that'll get more people
59:03 to to show up I I don't really know how
59:06 to do this Tick Tock [ __ ] I'm 58 years
59:08 old what do you want from an old man
59:10 aren't you Rosie O'Donnell okay
59:14 I was accused of looking like Rosie
59:16 O'Donnell the other day I'm stuck in a
59:18 code can you bring me up so I can show
59:21 oh yeah
59:23 um I'm gonna lose my
59:26 my double uh
59:28 camera thingy here if I do that let me
59:30 come back to you Mexico
59:34 entertaining Yeah question mark is the
59:37 appropriate punctuation punctuation mark
59:39 for entertaining I just opened a small
59:41 business can chat GPT help with
59:43 marketing yes in the dramatic fashion
59:49 so so like start playing again I'm just
59:52 using chat gbt 3.5 which is not
59:55 connected to the internet you don't have
59:57 to pay for it it's free for everyone
59:58 it's wicked fast act as
1:00:02 a marketing VP
1:00:06 and I want you to ask me five questions
1:00:12 about my business
1:00:16 one at a time
1:00:19 and don't move on to the next question
1:00:27 until I say ready
1:00:32 do you understand
1:00:35 wait understand stand question mark uh
1:00:38 before that I'm going to say
1:00:40 um
1:00:42 you should learn enough to help me write
1:00:48 a marketing plan
1:00:50 period do you understand
1:00:53 yes I understand as the marketing VP I
1:00:55 will ask you five questions when you're
1:00:57 ready let me know and we can begin
1:00:59 ready
1:01:02 so so now we have like a consultant
1:01:05 right so who's your target audience
1:01:08 um wait this is for my eggplantia pizza
1:01:10 business
1:01:11 um Pizza Lovers who also love organic
1:01:18 eggplant
1:01:20 it's a very specific audience thank you
1:01:23 for providing that next what sets your
1:01:25 pizza business apart from the
1:01:27 competitors so notice it's not just
1:01:29 asking me generic questions it's taking
1:01:31 my answers into context
1:01:33 um
1:01:34 my pizza
1:01:37 is delicious
1:01:41 and has a slimy texture
1:01:48 thank you for sharing that unique aspect
1:01:51 of your pizza no judgment
1:01:53 um what are your current marketing
1:01:55 efforts
1:01:57 I have a sign saying when we're
1:02:04 when we're open
1:02:07 and that's it
1:02:12 all right now let's proceed what are
1:02:14 your business goals I want to be
1:02:18 bigger than McDonald's
1:02:22 I'm also delusional what's your budget
1:02:25 for marketing initiatives 100 a week
1:02:30 thank you
1:02:31 all right
1:02:33 please give me some time to create the
1:02:35 plan and I will present it to you
1:02:36 shortly that's interesting that it said
1:02:38 that
1:02:39 um okay what's the plan
1:02:45 based on the information you gave me
1:02:48 Define your brand identity enhance your
1:02:50 online presence
1:02:51 user-friendly website optimize your
1:02:53 website and great engaging shareable
1:02:56 content
1:02:58 leverage social media platforms
1:02:59 Implement targeted advertising allocate
1:03:02 a portion of your budget
1:03:03 encourage customer reviews and referrals
1:03:06 collaborate with local Partners Monitor
1:03:09 and analyze your results right this is
1:03:11 if if you were to hire a marketing firm
1:03:13 this is what they give you after having
1:03:15 a conversation like that
1:03:17 and then you can dig into each one of
1:03:19 those you can say
1:03:21 um
1:03:22 engaging shareable content I like the
1:03:26 shareable
1:03:29 content idea can you give me a table of
1:03:35 20
1:03:37 social tactics
1:03:42 the Channel
1:03:46 and and
1:03:48 an example
1:03:51 um
1:03:53 piece of content I don't even know if
1:03:56 that makes sense but let's see what it
1:03:57 gives us
1:04:01 so Pizza trivia on Facebook did you know
1:04:04 pizza margarita was named after Queen
1:04:06 Margherita of Italy learn more pizza
1:04:08 trivia here recipe sharing Instagram
1:04:11 mouth-watering image of your customer
1:04:13 testimonials Twitter tweet featuring a
1:04:16 customer review along with a thank you
1:04:17 message with a special offer for their
1:04:19 next visit Pizza poles interactive Pizza
1:04:22 poll behind the scenes videos so yes it
1:04:25 can help you with marketing I haven't
1:04:27 really dug too much into marketing but
1:04:29 it's it's almost like you know anything
1:04:30 that you you it shape it shape shifts
1:04:35 and it kind of does what you want it to
1:04:37 do so again uh if you're new to this
1:04:39 stuff
1:04:41 um
1:04:42 go to these sites
1:04:44 go play
1:04:46 don't hesitate don't just be like yeah
1:04:48 Kyle I'll go there tomorrow no go there
1:04:52 while I'm babbling
1:04:53 babbling rambling
1:04:56 I've been watching your lives quite a
1:04:57 bit sorry about that thank you
1:05:00 agent GPT is useful agent GPT if you
1:05:04 don't know is something called an
1:05:05 autonomous agent and agent gbt is a web
1:05:09 a web version of it like a web front end
1:05:12 so you don't have to install it but you
1:05:14 can just give it a go and it will sort
1:05:15 of write its own prompts and go try to
1:05:18 do stuff on your behalf which is a
1:05:20 little bit terrifying but it's also
1:05:21 super cool
1:05:23 um and they're also brand new so they're
1:05:24 a little wonky they don't work super
1:05:26 well just yet um I noticed chat GPT has
1:05:28 a 4K limit how can I get over that okay
1:05:30 Bill Smith great question
1:05:33 that's the reason I have poe.com up here
1:05:35 so let me go show you that so for those
1:05:38 of you that don't know there's this
1:05:39 thing called a context window which
1:05:43 basically means you can't have more
1:05:45 words than that
1:05:47 um it's four thousand tokens which
1:05:50 translates into three thousand words if
1:05:52 your question is why don't they just say
1:05:54 three thousand words instead of four
1:05:56 thousand tokens they're Engineers they
1:05:59 think in tokens they programmed it in
1:06:01 tokens tokens was the solutions tokens
1:06:03 it is
1:06:04 it's about three quarters the amount of
1:06:06 words
1:06:08 to tokens so
1:06:10 Chachi BT has a 4 000 token limit and it
1:06:13 has a three thousand word limit on
1:06:15 average so basically what that means is
1:06:17 if you take a really long document and
1:06:19 copy and paste it in there it will break
1:06:21 it it'll just be like I don't wanna I
1:06:23 don't want fail I can't do it's too much
1:06:26 it's too much but
1:06:29 if you go to poe.com
1:06:33 um
1:06:35 so poda comes just like you know chat
1:06:38 GPT there's a little thing at the bottom
1:06:40 you type [ __ ] in there and it makes
1:06:41 stuff but down the left hand side you
1:06:44 have some other options here one of them
1:06:46 this fourth one down Claude instant 100k
1:06:51 is a hundred thousand token limit
1:06:54 chat like it's like chat GPT 3.5 but you
1:06:57 can put 75 000 words in it so let's go
1:07:00 do something fun let's go to
1:07:03 I wonder if I still have these
1:07:05 screenplays I think I do is this it
1:07:08 does
1:07:10 [Music]
1:07:13 what's this okay
1:07:16 so I went out on the old internet
1:07:20 and I did me a Google search for
1:07:23 screenplays and I found this is the
1:07:25 screenplay of The Godfather so I'm going
1:07:28 to copy this and it's long right it's
1:07:30 it's a complete feature-length
1:07:31 screenplay
1:07:32 so I'm going to go back to poe.com and
1:07:35 I'm going to go to Claude instant 100k
1:07:38 I'm going to sort of clean clean
1:07:41 why is it not working why is that not
1:07:44 working why are you not working
1:07:46 context cleared okay because it's ready
1:07:48 okay so
1:07:50 the context is cleared now so so now I'm
1:07:53 in
1:07:54 Claude instant 100k
1:07:57 come on Focus
1:07:59 these stupid cameras
1:08:01 all right whatever
1:08:02 okay I'm gonna go into the little prompt
1:08:05 window here I'm going to paste and so
1:08:08 in the prompt is the entire screenplay
1:08:10 of The Godfather and then I'm going to
1:08:12 say
1:08:13 who let's see
1:08:15 make me a table
1:08:18 of the main
1:08:21 characters and what happens
1:08:25 to them
1:08:27 in this screenplay
1:08:34 and so
1:08:36 did it do it already it's not a table
1:08:39 wait
1:08:41 oh it's waiting so it's doing a thing
1:08:43 all right
1:08:45 still waiting so so right now I guess
1:08:47 it's reading reading all that it's
1:08:48 interpreting all that
1:08:52 all right so here's a table
1:08:54 Don Corleone is shot by solazzo
1:08:57 Marco Michael Corleone starts as an
1:09:00 outsider in the family business but
1:09:02 after Sonny's death he takes over
1:09:03 spoiler alert Sonny Corleone takes over
1:09:06 running the Corleone family business
1:09:08 after Daddy's shot Fredo Fredo
1:09:12 Connie Corleone Tom Hagan key Adams
1:09:15 right so
1:09:18 um now I could say
1:09:20 um
1:09:21 what is the inciting incident
1:09:26 incident in this screenplay
1:09:36 [Laughter]
1:09:39 we're waiting we're waiting we're
1:09:41 waiting for miraculous things to happen
1:09:43 with technology that just seems like
1:09:46 magic yeah
1:09:50 who said that what's his name
1:09:53 um
1:09:53 science fiction writer any sufficiently
1:09:56 advanced technology is indistinguishable
1:09:58 from Magic that's what this is
1:10:02 the inciting incident of The Godfather
1:10:04 screenplay is when Vito Corleone the Don
1:10:06 is shot by solazzo
1:10:08 um and the tataglias
1:10:11 the spark this Sparks a chain of events
1:10:14 that follow sunny retaliates
1:10:16 Sonny's eventually assassinated Michael
1:10:18 Corleone has stayed out of the family
1:10:20 business is drawn in and plots the
1:10:22 murder of salazzo McCluskey and the head
1:10:25 of their rival families eventually Carlo
1:10:28 Rizzi Connie's husband
1:10:30 through these action Michael Rises to
1:10:33 become the new head of the Harley own
1:10:35 family and solidifies his family's power
1:10:38 so the shooting of Vito Corleone is so
1:10:41 like remarkable so you can paste in up
1:10:44 to 75 000 words that's like a whole
1:10:46 novel or a whole screenplay or a whole
1:10:49 legal document and you can start playing
1:10:50 with it you do have to pay for so the
1:10:53 these gpt4 Claude plus and Claude
1:10:56 instant 100 000 are paid subscriptions
1:10:59 um but I think it's 20 bucks a month
1:11:01 so that's how I would do that right now
1:11:04 there's going to be lots of other
1:11:05 options for that moving forward I think
1:11:07 I just saw this last week someone is
1:11:11 working on a three million token context
1:11:14 window
1:11:15 which is just silly
1:11:19 um but you can have really long
1:11:20 conversations and it won't kind of lose
1:11:22 its mind
1:11:24 all right love this [ __ ] thank you I'm
1:11:26 over 50 and I don't usually get into
1:11:28 this techie stuff appreciate you Gen X
1:11:30 baby I'm 58 58 and twerking all night
1:11:34 long
1:11:36 old guys twerking on Tick Tock yeah
1:11:39 party gym
1:11:41 or what's your name it ain't Jim
1:11:43 B party B
1:11:46 all right
1:11:49 someone's coming in here is is this guy
1:11:52 mentally challenged
1:11:56 can you show me how to add a photo in
1:11:58 mid Journey so I can search for more
1:12:00 photos like it
1:12:03 you can't really do that that I know of
1:12:08 [Laughter]
1:12:11 what you can do is you can put in a
1:12:14 photo into mid journey and then you can
1:12:15 generate you can generate images that
1:12:18 are based on it like I could upload a
1:12:20 photo of my my face and then say you
1:12:24 know I want a gangster in a 30s movie
1:12:27 and it'll look like me some version of
1:12:29 me
1:12:30 what site is this again I don't know
1:12:32 which site it is I don't see I'm not
1:12:33 seeing these comments in real time you
1:12:35 just ruined pizza for an entire
1:12:37 generation thank you that's why I'm here
1:12:40 eggplantia it's gonna be the new thing
1:12:42 slimy Pizza
1:12:44 um
1:12:45 all right Bard sounds
1:12:47 like sales messages just isn't the same
1:12:49 Bard sucks which is ironic because it's
1:12:53 from Google that invented the
1:12:54 Transformer that's the thing powering
1:12:57 all of these super cool tools but then
1:12:59 they kind of stuck it in the back room
1:13:01 because it was going to threaten their
1:13:03 business and then open AI came along and
1:13:05 said we'll take it from here and then
1:13:09 Microsoft came along and said we'll give
1:13:10 them 10 billion dollars and we'll both
1:13:12 take it from here and Google have fun
1:13:15 catching up and that's what's happening
1:13:17 right now
1:13:18 there's some things that Bard does
1:13:20 apparently that are better than uh
1:13:22 better than chat CBT but nothing that
1:13:25 I've used it sucks at writing
1:13:30 can I do an example for creating
1:13:31 illustrations for children's books I did
1:13:33 that
1:13:35 can you do an example creating
1:13:36 illustrations for children's books if it
1:13:38 seems like I'm ignoring you it's just
1:13:39 because I'm really behind on these you
1:13:41 don't need to ask questions over and
1:13:42 over again
1:13:43 love this Luke the X fighter camera
1:13:45 angle
1:13:47 is that better that better I don't know
1:13:50 whatever my camera angle is never right
1:13:52 I've adjusted my camera angle 16 times
1:13:54 at least I'm not getting the could you
1:13:57 go trim your nose hairs comments anymore
1:13:59 so I think I've improved it
1:14:01 what's all this hubbub about Al who's Al
1:14:04 exactly
1:14:06 it's funny that we call it AI because
1:14:08 it's really machine learning but machine
1:14:09 learning sounds like something you have
1:14:11 to work at right oh it's a machine and I
1:14:13 have to learn it that's exhausting let's
1:14:16 say I oh that sounds good I'll take an
1:14:18 AI
1:14:20 and also the movies have told us about
1:14:22 AI or Al as you call him good question
1:14:25 who's Al
1:14:27 um there's more than one way to skin
1:14:29 this cat and AI
1:14:32 wait an AI to the need of humans
1:14:35 existing I'm not quite sure what you
1:14:37 mean there but
1:14:38 sure I think what you said yes or not
1:14:42 fun fact when I added my plugins link
1:14:45 reader is not in the store any chance
1:14:47 this is a glitch
1:14:49 so so Mark there's a weird thing and
1:14:53 which is like I'll see you know people
1:14:57 will be on Tick Tock and they're like oh
1:14:59 you know go check out this plug-in on
1:15:01 the plugin store and then I go to try to
1:15:02 find it and it's not there so I have I I
1:15:05 I don't know
1:15:07 I don't know how they're rolling these
1:15:09 features out because these are beta
1:15:11 features so if you if you pay for chat
1:15:13 GPT plus and you don't see plugins at
1:15:16 all
1:15:17 you have to go enable them so you go to
1:15:19 your settings and then there's this beta
1:15:21 features thing and you have to enable
1:15:22 them you have to turn them on
1:15:24 so so one issue could be if you don't
1:15:26 see them at all that's what's going on
1:15:27 there but then when you go in here
1:15:30 and you go to this little pull down menu
1:15:31 you go to the bottom you go to plug in
1:15:33 store
1:15:35 it usually takes forever to load because
1:15:36 it's probably on some shitty piece of
1:15:38 azure software because you know
1:15:40 Microsoft
1:15:42 and then if I say all how many are there
1:15:45 there's 97 Pages times four so there's
1:15:48 almost 400 plugins
1:15:50 but like link reader let me let me type
1:15:53 in link
1:15:56 Access Link
1:16:05 yeah that's weird
1:16:07 I wonder if they changed the name of it
1:16:10 um
1:16:11 let me see what I've got installed oh
1:16:12 yeah see
1:16:14 see this little
1:16:16 um
1:16:16 oh you can't see it hang on let me dim
1:16:18 my screen a little bit there's a there's
1:16:21 a little thing it looks like link reader
1:16:23 is a dead plug-in
1:16:26 so I'm gonna uninstall it I'm gonna
1:16:28 uninstall it and then I'm going to go
1:16:30 back to all
1:16:32 and I'm going to type in reader
1:16:37 yeah I have I have a feeling that that
1:16:40 link reader
1:16:41 um
1:16:44 what it what it probably was this is my
1:16:47 guess so so these the the plugins
1:16:51 this is a really new infrastructure like
1:16:54 they just kind of launched them quickly
1:16:57 it's very very simple to make them I
1:16:59 have a feeling that what happened with
1:17:01 link reader was it was probably like
1:17:03 some college student or some you know
1:17:05 just some sort of hacker type
1:17:08 um that
1:17:09 quickly threw together a plug-in
1:17:12 it probably had an API that was doing
1:17:14 whatever it was doing like reading the
1:17:16 link and they were probably being
1:17:17 charged two pennies per call and like
1:17:21 people started using it they're like oh
1:17:22 [ __ ] this is going to be a big Bill let
1:17:24 me turn this thing off so so a bunch of
1:17:27 the plugins right now are just janky
1:17:29 Little Fly by Night operations so I so I
1:17:32 think that's what happened there
1:17:34 um interesting though thanks for
1:17:35 bringing that up because I've been I've
1:17:37 been using link reader in some of the
1:17:39 demos that I do so now I know that it's
1:17:42 not there which website is the one with
1:17:45 the illustrations there's a bunch of
1:17:46 them so there's
1:17:48 um
1:17:50 there's
1:17:53 hold please hold please
1:17:57 there's Lexica
1:18:00 lexicon.art is is basically like a
1:18:03 Google a Google kind of thing for stable
1:18:07 diffusion images and you can subscribe
1:18:10 to it and generate your own images here
1:18:12 as well but you can just go copy
1:18:13 someone's prompt and then iterate on it
1:18:16 there's mid-journey which in my opinion
1:18:18 is the best quality
1:18:21 there's Adobe Firefly so go look up
1:18:23 adobe Firefly I think that's in a free
1:18:26 beta but you might need an adobe account
1:18:30 but Firefly is
1:18:33 um the advantage of firefly other than
1:18:36 the fact that it's like some of the
1:18:38 interface elements in it are pretty good
1:18:39 is it's trained it's not trained on any
1:18:43 proprietary data it's trained on public
1:18:45 domain data and Adobe stock images that
1:18:47 adobe owns all the rights to so Adobe is
1:18:51 making these generative images safe for
1:18:54 commercial use right so if you're if
1:18:56 you're doing something in a business
1:18:57 especially if you're in a big business
1:18:59 mid-journey and stable diffusion
1:19:02 could potentially put you in legal weird
1:19:05 Waters depending on what you do so if
1:19:07 you just want to be safe safe safe and
1:19:09 you're in a commercial environment use
1:19:11 Adobe Firefly the other thing that
1:19:13 they're going to be doing which I think
1:19:15 is is really brilliant is they're going
1:19:16 to allow artists to opt out of having
1:19:18 their images trained or trained on and
1:19:22 they're going to allow them to opt in
1:19:23 and they're going to create an
1:19:24 attribution engine and they're going to
1:19:26 create a compensation engine for this
1:19:29 generative AI so the fact that adobe's
1:19:31 in this game is a really big deal
1:19:34 can it sequence various images to
1:19:36 compile a gif
1:19:38 um
1:19:39 sort of not really but with Jenna with
1:19:42 with uh with runways Gen 2 you can do
1:19:46 text to images or with Gen 1 you could
1:19:50 shoot a video of like a dog walking and
1:19:52 then change the style of that
1:19:54 so you could probably generate that on
1:19:57 your own and then I think within stable
1:19:59 diffusion here's the chat here's the the
1:20:02 good news about stable diffusion is the
1:20:04 developer communities creating all sorts
1:20:06 of stuff on it the bad news is you kind
1:20:09 of need to be a geek to use it right now
1:20:10 like you need to install virtual
1:20:12 machines and
1:20:14 um you know there's a version you can
1:20:16 install on your Mac but it's got to be
1:20:17 an N1 Mac and then it doesn't work and
1:20:19 then you got to update it
1:20:23 but there are some animation things
1:20:25 within stable diffusion that you could
1:20:26 potentially do what you're talking about
1:20:28 all right let me all right it just
1:20:30 jumped me to the bottom so I'm just
1:20:32 going to go up a little bit from the
1:20:33 bottom here and start answering some
1:20:34 questions so I probably just skipped
1:20:36 your things
1:20:37 what do I recommend for Designing and
1:20:40 formatting an e-book well I would just
1:20:43 use canva like go into canva and select
1:20:45 one of their ebook formats and then use
1:20:49 Chachi chat chat EBT to write this [ __ ]
1:20:53 I know I know a handful of people that
1:20:55 are doing books like that this one guy
1:20:57 went to some party
1:20:58 and this someone's like oh you got to
1:21:00 talk to this guy he's doing AI too I'm
1:21:02 like hey how's it going he's like
1:21:04 I've published 14 books in the last
1:21:06 three days I'm like dude dude you're
1:21:09 exhausting me no so anyway
1:21:14 deep fakes lately reading that ai's
1:21:17 impact is overstated hyped by tech
1:21:20 companies looking to pump their stock
1:21:22 well Gregory
1:21:25 um I wouldn't listen to that and I would
1:21:27 go start using it so I don't know what
1:21:29 you've used with chat GPT that's why I
1:21:32 have these things here is
1:21:35 this is really hard to understand from
1:21:37 the outside it like it's not until you
1:21:39 actually use this stuff and and
1:21:42 particularly when you use it in there in
1:21:44 an area where you've got deep expertise
1:21:47 like if you're really good at marketing
1:21:48 have it write you a marketing plan or
1:21:50 like work with it to do some marketing
1:21:52 stuff if you're good at organizing stuff
1:21:54 have it help you come up with a project
1:21:57 management plan
1:21:59 um
1:22:01 I would strongly encourage you to get
1:22:03 curious about it learn enough about it
1:22:05 to know what it makes possible and then
1:22:08 if you still feel that way cool but I
1:22:10 don't think you will
1:22:12 um so so just to give you a little
1:22:13 context
1:22:15 um I've been an entrepreneur for I don't
1:22:17 know 35 or 40 years and
1:22:20 um
1:22:21 in the mid 90s I started one of the
1:22:23 first digital agencies and and in 1994 I
1:22:26 had I had a big Epiphany about the World
1:22:28 Wide Web I started an online magazine
1:22:30 three weeks after I launched it in 1994
1:22:34 there was a full page article about this
1:22:38 magazine that I published in Paris
1:22:41 I publish it from my apartment in
1:22:42 Brooklyn I uploaded these files to a
1:22:44 computer in California three weeks later
1:22:47 it was written about in Paris so I had
1:22:50 this Epiphany that holy [ __ ] the world
1:22:51 just changed and and you know I created
1:22:54 this company called agency.com which was
1:22:56 one of the early digital agencies and
1:22:57 built a lot of the websites for the
1:22:59 Fortune 500 we built British Airways
1:23:01 website it was intense we grew to like 2
1:23:03 200 people it was just insanity
1:23:06 I had an almost identical Epiphany in
1:23:09 December of 22 you know six months ago
1:23:12 with chat gbt where holy [ __ ] the world
1:23:15 just changed and no one knows it yet
1:23:18 um this stuff is profound so it
1:23:22 certainly isn't under hyped right lots
1:23:24 of people are talking about it when they
1:23:25 start covering [ __ ] on 60 Minutes in the
1:23:27 morning shows you know there's some
1:23:29 stuff happening so it's definitely over
1:23:31 hyped
1:23:33 um
1:23:34 but for example I spent about a year two
1:23:37 years ago digging deep into blockchain
1:23:39 and nfts I got involved with a project
1:23:41 called Neo Tokyo which was really
1:23:44 intense and cool and far out
1:23:47 and I was like ah this I this I think
1:23:49 this could be the next big thing and
1:23:50 then the more I dug into it the more it
1:23:52 kind of vaporized and it was just all a
1:23:54 shell game
1:23:56 and then I started playing around with
1:23:58 generative image stuff and as I got more
1:24:02 and more into that the more I got into
1:24:03 it the more I realized this is this
1:24:05 feels like a bigger deal and then Chachi
1:24:08 PT came out and
1:24:10 the adoption of chat GPT so here this
1:24:14 may help put it in context as well
1:24:16 the World Wide Web so we know the impact
1:24:18 that the World Wide Web had right like
1:24:20 it completely transformed how we
1:24:22 communicate how we publish how we
1:24:24 entertain how we connect social media
1:24:26 all that sort of stuff it was the
1:24:28 foundation of that it took six years to
1:24:31 get from zero users to 100 million users
1:24:33 of the World Wide Web Chat GPT to get to
1:24:37 100 million users it was six weeks
1:24:40 six years for the World Wide Web six
1:24:42 weeks for chat GPT
1:24:44 so there's something different here and
1:24:47 and it's hard to explain it from the
1:24:49 outside because it kind of looks like oh
1:24:51 it's like Fancy Google
1:24:53 it's not fancy Google
1:24:55 is
1:24:57 shaking in their boots right now because
1:24:59 it's not fancy Google it threatens fancy
1:25:02 Google or I mean it threatens Google ask
1:25:05 it how it can survive the next two weeks
1:25:07 with a dollar one in my bank account I'm
1:25:09 right there with you my bank account's
1:25:11 [ __ ] right now
1:25:14 you know I think dollar one at this
1:25:17 point I don't even know if you can buy a
1:25:18 box of macaroni and cheese for a dollar
1:25:20 one when I was an actor you get 10 for
1:25:23 10 for a dollar ten boxes of mac and
1:25:25 cheese for a dollar
1:25:27 um what can help with prompts for
1:25:29 mid-journey something in po oh so who's
1:25:33 what's your name lock luck lock in 20
1:25:37 luck 120 I don't know
1:25:39 um okay one of the one of the options to
1:25:42 go play with
1:25:44 so I showed you the Claude 100 000 thing
1:25:47 that's you can put in a whole screenplay
1:25:48 or just long documents that's super cool
1:25:51 but if you go to the bottom here below
1:25:54 below chat GPT are Bots and like here's
1:25:57 one called I am Yoda that I made and
1:26:00 then if I go to all your Bots there's
1:26:03 there's a bunch of them here I am Yoda
1:26:06 Emily Dickinson I am Steve Jobs I am
1:26:08 Jay-Z so these are all a bunch of bots
1:26:10 that I made so you can make your own bot
1:26:13 so
1:26:14 and then and then you can store them if
1:26:16 you use them and like them you can store
1:26:18 them so there's two here one's called MJ
1:26:20 captain and one's called mid Journey
1:26:22 both of these are mid-journey prompt
1:26:25 amplifying Bots for lack of a better
1:26:28 term
1:26:29 so if I go to MJ captain
1:26:32 I'm going to clear my context
1:26:35 and so okay we're ready to go and so
1:26:37 basically it says welcome I'm a prompt
1:26:39 building bot that creates prompts for
1:26:42 you to use with mid Journey you can ask
1:26:43 me to create prompts for paintings
1:26:45 photographs or designs simply say I need
1:26:48 a photo of X what kind of prompt do you
1:26:50 need I need
1:26:53 a design of a
1:26:57 website for
1:27:01 a pizza
1:27:04 company that uses
1:27:08 organic
1:27:10 eggplant
1:27:12 [Music]
1:27:14 as the main ingredient
1:27:20 and so then it rewrites what I wrote
1:27:23 into a mid-journey friendly prompt so a
1:27:27 website designed for a pizza company
1:27:28 that uses organic eggplant as the main
1:27:31 ingredient with a modern aesthetic
1:27:33 bright green and purple palette so it
1:27:36 pulled that from there
1:27:39 champ
1:27:41 champ
1:27:42 no one cares no one cares
1:27:58 you good you good you good hey buddy
1:28:01 champ you good no
1:28:06 all right now thank you for saving us
1:28:11 those bombs with strollers they're so
1:28:13 they're so dangerous they're so
1:28:15 dangerous
1:28:16 um anyway so so
1:28:18 uh inspired by the art of Henry Matisse
1:28:20 using Parallax scrolling technique so
1:28:23 this is going to create a website so now
1:28:24 I'm just going to copy this prompt
1:28:27 and I'm going to go over to Discord and
1:28:29 you might ask why would you go to
1:28:31 Discord to make an image because that's
1:28:33 how this [ __ ] is right now and just
1:28:35 life is a poo poo head
1:28:39 with this technology being so early
1:28:41 that's also like something very
1:28:43 reminiscent of the early days of the web
1:28:44 is there were just all these janky
1:28:46 little tools some of which were really
1:28:48 cool and became major companies and some
1:28:51 of which were just pieces of [ __ ] and
1:28:53 that's we're right in the middle of that
1:28:54 right now so I'm going to do slash
1:28:56 imagine
1:28:57 and then I'm going to paste in that
1:28:59 prompt that it gave me
1:29:02 and it should give us a website design
1:29:04 for our silly little Pizza Company we'll
1:29:06 see what it does
1:29:09 um so that's that hopefully that answers
1:29:10 your question if you have questions
1:29:12 about uh AI generative AI pop them into
1:29:15 the comments below
1:29:17 um if you like barking dogs great if you
1:29:22 like
1:29:23 um
1:29:25 silly men that wax poetic about
1:29:28 technology you've come to the right
1:29:30 place I do not twerk there have been
1:29:32 requests
1:29:34 for me not to twerk
1:29:36 I can imitate Brando
1:29:38 Stella all right does AI save those
1:29:42 results to be more efficient if I asked
1:29:45 again does AI save those results so
1:29:49 is it
1:29:51 is it training so not really not really
1:29:54 it kind of It kind of does improve
1:29:57 within a conversation and it saves your
1:30:00 prompts for you to go back to and they
1:30:03 are training
1:30:05 chat GPT to get better on all of our
1:30:08 prompts but they're not doing it like
1:30:09 your prompt will get better over time
1:30:11 it's not there's not quite a one-to-one
1:30:13 relationship like that
1:30:15 I'm 59 I love this [ __ ] yeah there's a a
1:30:18 friend of mine Jessica Birdman Bergman
1:30:21 Bergman Bergman I think it's Bergman
1:30:24 um just wrote a book about Gen X and how
1:30:26 generative AI might be the new sort of
1:30:29 you know savior of Gen X because we
1:30:31 bring a lot to the table that are our
1:30:34 appropriate skills for this generative
1:30:36 AI stuff
1:30:37 gonna be 50 in August do you think
1:30:39 machine learning will be the Spades to
1:30:42 Ai and the gold gold rush
1:30:45 um
1:30:47 well I think the large language models
1:30:49 the the big base large language models
1:30:54 I don't I don't quite know which piece
1:30:57 of the metaphor is there but it but it's
1:30:59 this is definitely this generative AI
1:31:01 stuff is definitely in my opinion this
1:31:04 is like Industrial Revolution scale
1:31:07 transformation
1:31:09 it's big and it's coming fast
1:31:13 um so whether you call it Gold Rush or
1:31:16 whether you call it I don't know
1:31:19 it's coming fast it's going to be big
1:31:22 challenge make an AI vid or Kyle
1:31:24 twerking
1:31:25 I made an AI vid and it kind of sucked
1:31:27 make an eggplant
1:31:29 um T-shirt design with AI okay I'll do
1:31:31 that in a second but why is this not
1:31:33 going
1:31:34 come on Mid Journey the hell's up all
1:31:37 right let's go back to we'll go back to
1:31:39 PO for a second we'll come up with a an
1:31:42 eggplant t-shirt design
1:31:45 um
1:31:46 uh
1:31:49 funny
1:31:52 eggplant Pizza
1:31:56 t-shirt
1:31:58 image
1:32:01 a funny t-shirt design featuring an
1:32:03 eggplant Pizza
1:32:05 with a cartoonish style a bright and
1:32:07 bold color palette inspired by pop art
1:32:09 using halftone that's interesting
1:32:12 typography Mastery by Jessica heesh
1:32:17 with a humorous catchphrase excellent
1:32:21 pizza that's funny
1:32:24 um no dull serious realistic I don't
1:32:26 know let's go take a look and see if
1:32:28 that does anything interesting
1:32:31 there's some things in there that I
1:32:32 think are going to [ __ ] it up but we'll
1:32:33 see
1:32:35 all right slash imagine
1:32:39 imagine all the people all right so we
1:32:43 got those going so I got two going there
1:32:45 we'll come back to those
1:32:47 um welcome everybody looks like there's
1:32:48 some new folks here if you're new here
1:32:50 this is the AI learning lab my name is
1:32:52 Kyle Shannon
1:32:54 um these websites that I put in this
1:32:55 magic little window here are
1:32:58 um just there for your viewing pleasure
1:33:00 those are those are like a public
1:33:02 service announcement the the top one is
1:33:05 the official chat thank you for the rose
1:33:06 chat GPT website you got bing.com is uh
1:33:10 Microsoft's implementation of chat gbt
1:33:13 it's gpt4 connected to the internet and
1:33:16 then poe.com there's lots of different
1:33:17 models you can play with so that's those
1:33:19 are there and then the other two are
1:33:20 just resources that I think are worth if
1:33:23 you're just getting started they're
1:33:24 worth you playing with
1:33:26 um lots of rabbit holes to dive down how
1:33:29 do you do the camera effect
1:33:31 I tried I couldn't do it on live you
1:33:33 mean with the with this thing and this
1:33:35 thing
1:33:35 with the cameras doing this
1:33:38 um it's called dual camera I don't know
1:33:40 when I go live there's an option there
1:33:42 for dual camera tick tocks of
1:33:46 I was gonna say a piece of [ __ ] but tick
1:33:48 tocks
1:33:50 it's a mess of an interface and they
1:33:52 just randomly change [ __ ] and take [ __ ]
1:33:54 away and like not everyone has access to
1:33:56 the same tools as others but it's called
1:33:59 dual camera and it that it could be it
1:34:02 could be restricted to certain phones
1:34:04 like certain devices I don't know
1:34:07 I don't know
1:34:08 I wish I could help you
1:34:11 all right with the square okay thanks
1:34:13 for the link reader explanation I found
1:34:15 some Alternatives but wasn't sure if I
1:34:16 was missing out and quite frankly link
1:34:18 reader I think with just with um
1:34:24 with GPT connected to the internet with
1:34:26 Bing I I think you could probably just I
1:34:28 haven't tried it you probably just
1:34:30 Post in a URL there and ask it to go
1:34:33 read it and see if it see if it did it
1:34:35 hello Kyle oh [ __ ] Tick Tock just
1:34:38 bounce me to the bottom hold please like
1:34:41 it's like it's usually when I'm in the
1:34:42 middle of reading something and I'm
1:34:43 trying to slow the scrolling down
1:34:47 um I'm gonna sit quietly quiet quietly
1:34:49 and listen in okay
1:34:52 won't be quiet I'm a noisy pain in the
1:34:55 ass AI is the equivalent to the
1:34:57 innovation of the calculator calculator
1:34:59 maximizes time and effort it is it's
1:35:01 very much like the calculator but it's
1:35:03 like you know it's like the calculator
1:35:04 on steroids it's like
1:35:07 the Lance Armstrong of calculators
1:35:11 right it up guy beat your right it up
1:35:13 guy you ever hear Bill Burr talk about
1:35:16 the Lance Armstrong thing hilarious
1:35:20 you answer questions or no I don't know
1:35:22 do I
1:35:25 I might just Ramble On and never answer
1:35:28 them I think that's probably more likely
1:35:30 but yeah I answer questions
1:35:33 Trey Moore Trey Moore
1:35:36 ask me a question I'll answer it
1:35:38 everyone knows that so what all right
1:35:41 listen don't troll if you if you're
1:35:43 gonna troll don't hang out not
1:35:44 interested not interested who's behind
1:35:46 chat gbt who owns it okay interesting
1:35:49 question
1:35:51 um
1:35:52 a number of investors started it
1:35:54 including Elon Musk it was started as a
1:35:57 non-profit company
1:35:59 um Sam Altman is the guy that runs it he
1:36:01 used to run y combinator
1:36:03 um I like Sam Altman a lot
1:36:06 one of the things that happened was
1:36:09 when Google came out with the
1:36:11 Transformer model and openai started
1:36:14 experimenting with
1:36:17 the Transformer with these tools and
1:36:19 they created gpt1 and then gpt2 what
1:36:22 they learned is that the more compute
1:36:24 power and the more you know gpus
1:36:29 you threw at it and the bigger the data
1:36:31 set that you created the better it got
1:36:33 and so what they realized is
1:36:36 they needed way more money than they
1:36:39 were going to realistically be able to
1:36:40 raise as a non-profit
1:36:42 this is Sam Altman's story right so so
1:36:45 you can like I don't know I don't know
1:36:48 where the like you know the facts are
1:36:49 always somewhere in the middle right
1:36:50 this is Sam Altman's story
1:36:53 so what they realized was they needed a
1:36:55 hell of a lot more money
1:36:57 um to really do this right and and Sam
1:36:59 Altman has since come out and said he
1:37:01 thinks that it's going to take about a
1:37:02 hundred billion dollars to get to AGI
1:37:05 which stands for artificial general
1:37:06 intelligence
1:37:08 um so they flipped open AI from an
1:37:12 it was supposed to be an open source
1:37:14 not-for-profit company they flipped it
1:37:16 to a closed Source for-profit company
1:37:19 but with a capped profit meaning that no
1:37:23 investors could make more than a certain
1:37:25 uh it's like 100 X or something like
1:37:27 that they couldn't make more than that
1:37:30 um and then Microsoft came on and first
1:37:32 invested a billion dollars and then
1:37:34 invested 10 billion dollars and I I
1:37:36 assume most of that was in compute time
1:37:38 right so they created you know AI super
1:37:41 computers on Azure cloud services and
1:37:44 I'm sure that's what's running all this
1:37:45 [ __ ]
1:37:46 um so that's what that's where the
1:37:47 money's going
1:37:48 so at the end how does it work
1:37:53 Microsoft
1:37:56 gets
1:37:59 70 of their revenue initially
1:38:03 and then until they're paid back and and
1:38:05 at whatever Point Microsoft is paid back
1:38:08 then they'll own 49 of open AI so open
1:38:12 AI still maintains control I think
1:38:15 um Microsoft will own a minority share a
1:38:18 more you know a non-controlling stake
1:38:22 um apparently open AI is free to go
1:38:24 compete and do all the [ __ ] they want to
1:38:27 do they don't they're they're not
1:38:28 beholden to Microsoft but Microsoft has
1:38:31 gone all in so Microsoft owns GitHub and
1:38:35 so github's co-pilot is gpt4 with with
1:38:38 you know code specialization
1:38:41 um they announced they're rolling gpt4
1:38:43 into windows at the operating system
1:38:45 level they've announced that they're
1:38:47 rolling it into all of Office 365 into
1:38:49 Dynamics 365 into all the Azure services
1:38:54 Google has also announced they're doing
1:38:55 the same with bard
1:38:57 um but Microsoft is making a big bet on
1:39:00 this and and you know they've completely
1:39:02 transformed Bing which was their search
1:39:04 engine which was I think they had two
1:39:06 percent market share and Google had 92
1:39:08 or 94 market share
1:39:11 um so Microsoft's going in big
1:39:15 um I don't know who the owners are I
1:39:17 think it's private owners one really
1:39:18 interesting fact is Sam Altman doesn't
1:39:20 own any open AI stock apparently he said
1:39:24 that I just heard that in an interview
1:39:25 I'm like what
1:39:27 why would he not own a part of that
1:39:29 maybe it was just the way they set up
1:39:30 the non-profit
1:39:32 um
1:39:33 so that's the
1:39:35 all right
1:39:38 what can't so do I don't know what that
1:39:40 means he talks a lot but doesn't say
1:39:42 much stick with tutorials yeah I would
1:39:45 absolutely leave Cadillac Sev
1:39:47 champ barking is The Terminator at your
1:39:50 door
1:39:50 champ said I care great job you've been
1:39:53 doing this for 90 minutes I've been
1:39:54 doing this for 90 minutes holy crap
1:39:57 uh welcome if you're new here my name is
1:39:59 Kyle Shannon this is the AI learning lab
1:40:01 some of the haters in here are hating me
1:40:02 tonight but whatever
1:40:04 best way to use AI
1:40:06 um
1:40:07 just start getting curious with it take
1:40:10 what you know and start playing within
1:40:13 that so if you're good at marketing have
1:40:15 it do marketing stuff for you if you
1:40:17 have an idea for a novel or a recipe or
1:40:21 your dinner plan for the week have it
1:40:23 make your dinner plan for the week this
1:40:24 is a fun one
1:40:26 come up with like stuff you want to eat
1:40:29 for the week and tell it to make you a
1:40:30 dinner plan and then tell it to make you
1:40:32 a grocery list and then tell it to
1:40:34 organize the grocery list by section of
1:40:35 the grocery store you can knock that out
1:40:37 in a minute and a half so just go to
1:40:40 chat.openai.com and start playing
1:40:43 uh how about a floss
1:40:45 um do I food my teeth
1:40:49 boy the trollers are out tonight bitter
1:40:52 you betcha
1:40:54 you know the other option is you could
1:40:56 just leave here and go do something with
1:40:58 your life
1:41:00 um even bluesy's Grandma can can floss
1:41:03 um if I want to copy the design of an
1:41:05 existing website if I use word copy chat
1:41:09 EPT doesn't like it
1:41:12 I don't quite know what that means
1:41:17 um can it make a Father's Day card yes
1:41:20 you can so so
1:41:23 you can write a Father's Day card like
1:41:25 what I would do if if you wanted to do a
1:41:28 Father's Day card I'd go to
1:41:30 chat gbt
1:41:32 and I'd go
1:41:34 um
1:41:35 write me
1:41:37 um 20
1:41:39 Father's Day cards
1:41:42 in a table
1:41:44 with the
1:41:47 um
1:41:48 covered described
1:41:51 and the quote
1:41:54 um
1:41:55 on the
1:41:58 inside
1:42:01 for a father
1:42:03 that
1:42:05 is a used car dealer
1:42:10 here sure here's 20 cards right
1:42:13 and then it describes
1:42:17 picture of a classic car Dad you're a
1:42:19 classic thanks for steering me in the
1:42:21 right direction
1:42:22 multiple car keys the father who knows
1:42:25 the key to happiness get it keys and key
1:42:27 so yeah it can do that and then go over
1:42:29 to canva find a bunch of cards there
1:42:31 find a bunch of images and just knock it
1:42:34 out canva open AI 20 minutes Dad will be
1:42:39 happy
1:42:41 all right I just signed up for Chachi T
1:42:44 chat chat GPT plus turned on beta
1:42:48 plug-in features still no plugins
1:42:50 they'll probably get there I know
1:42:53 they're rolling them out they're rolling
1:42:55 them out
1:42:56 um
1:42:57 over time so what I would do is I'd go
1:43:00 to open Ai and see if they have a
1:43:02 plug-in wait list so just go to open and
1:43:05 search for it or just search on Google
1:43:07 open AI plug-in wait list you might need
1:43:10 to get on the wait list but but they're
1:43:11 rolling it out wide so I'm pretty sure
1:43:13 if you have plus and you sign did you
1:43:15 you said you turned on the beta features
1:43:18 yeah
1:43:19 I turned on the beta plug-in feature if
1:43:22 you turned on plugins they should be
1:43:23 there so
1:43:24 [Applause]
1:43:26 let me let me just show you what it
1:43:28 should look like and if it doesn't then
1:43:30 you probably just don't have access yet
1:43:32 gpt4 you've got default browse with Bing
1:43:36 and plugins so when you roll over GP T4
1:43:39 if you don't see these three things
1:43:42 um then either you don't have them
1:43:44 enabled in settings or you might not
1:43:46 have access to plugins that's a
1:43:48 possibility again it's very very early
1:43:51 and they're rolling this stuff out slow
1:43:53 and there's there's no Rhyme or Reason
1:43:55 to it like I got access to plugins
1:43:58 um super early
1:44:01 but then a bunch of people were getting
1:44:04 access to other things and and plugins I
1:44:06 didn't have access to so it just stopped
1:44:08 I don't have Dev access to gpt4
1:44:12 um
1:44:12 so I don't know what their what their
1:44:15 Rhyme or Reason is GPT writes my grocery
1:44:19 list now that's what I was just is
1:44:20 talking about new recipes with our
1:44:23 preferences this is bananas yeah it's
1:44:24 pretty amazing
1:44:26 um
1:44:28 up it just jumped me to the bottom again
1:44:34 all right chat gbt writes my grocery
1:44:37 lists thank you for the heart
1:44:39 I appreciate that do we need the chat
1:44:42 GPT for subscription or can I use it via
1:44:45 Bing and watching from Melrose Spain so
1:44:48 you don't need it
1:44:50 certain countries seem to be blocking it
1:44:52 the reason I put all of these up here
1:44:55 is for exactly your question
1:44:59 um it's a champ
1:45:01 no one cares just stop
1:45:05 what's going on here ah nothing's
1:45:08 working right I'm getting cranky
1:45:11 um
1:45:12 so
1:45:14 if you pay for chat CPT plus you get
1:45:18 access to gpt4 if you just go to
1:45:20 bing.com that is chatgpt4 connected to
1:45:24 the internet so so a big part of what
1:45:26 you pay for at open AI is free at Bing
1:45:29 the
1:45:31 the downside to it is
1:45:34 Microsoft had to kind of lobotomize Bing
1:45:37 because of a bad PR thing that happened
1:45:39 with the New York Times where Bing told
1:45:41 one of their reporters it should leave
1:45:43 his wife because it loved him uh that
1:45:45 made the cover of the New York Times so
1:45:47 Bing had to shut that [ __ ] down
1:45:50 um and they're kind of slowly opening
1:45:51 Bing back up so it I don't think Bing is
1:45:55 quite as good as chat's EPT but it is
1:45:57 gpt4 it is pretty good
1:45:59 um it's it's just it's just not quite
1:46:01 there yet
1:46:02 um so that's why I have those up there
1:46:03 so yes you yes you can is the is the
1:46:06 very short answer that I somehow turned
1:46:08 into a four-hour speech
1:46:10 um when do you start dancing on this
1:46:12 dance app well you're gonna have to
1:46:14 stick around I think the longer I Don't
1:46:16 Twerk the more the tension builds don't
1:46:19 you think
1:46:21 [Music]
1:46:25 and the new currency
1:46:28 PT can create programs for you how do I
1:46:31 actually create things like games and
1:46:33 trading Bots
1:46:34 you go to chat gbt
1:46:37 so so here's here's kind of a trick like
1:46:40 if you
1:46:42 if you don't know how to do something
1:46:44 tell chat GPT you don't know how to do
1:46:47 it and have it teach you so
1:46:50 let's go do a trading bot right now
1:46:54 um
1:46:55 we'll we'll do gpt4 connected to the
1:46:57 internet so I'm going to do browse with
1:46:58 Bing so I'm going to go
1:47:01 um
1:47:02 act as a
1:47:05 trading bot
1:47:07 specialist
1:47:10 and I want you
1:47:13 to teach me about trading Bots
1:47:19 and make me one even though I am totally
1:47:26 Clueless
1:47:28 can you help question mark
1:47:35 let's see what she says sure I can
1:47:38 certainly help you to understand trading
1:47:39 Bots and guide you through the process
1:47:42 a trading Bots essentially a software
1:47:45 program that interacts with a okay I
1:47:47 don't really care about any of this so
1:47:49 I'm going to say stop generating
1:47:52 you know
1:47:54 I don't want to learn I just
1:47:59 want to get rich
1:48:04 um what are some sites that have trading
1:48:09 Bots you can help me with
1:48:17 I understand you're looking for a
1:48:19 ready-to-use trading bot platforms
1:48:20 however I must remind you that while
1:48:22 trading Bots can potentially be
1:48:23 profitable so it gives you the
1:48:25 obligatory you know this is not
1:48:27 Financial advice
1:48:29 it's pretty funny
1:48:31 three commas crypto Hopper
1:48:34 Haas online
1:48:37 GunBot
1:48:39 trade Santa so anyways let's give me a
1:48:42 bunch of sites
1:48:44 um
1:48:46 what are some of the mower pop Euler
1:48:53 trading platforms that also have Bots
1:49:02 binance
1:49:03 can't do that in the U.S anymore
1:49:05 etoro
1:49:08 interactive brokers
1:49:15 this seems to be more crypto
1:49:18 coinbase Pro stop
1:49:21 these seem more
1:49:24 crypto
1:49:27 crypto what about stocks
1:49:35 if you're more interested in trading box
1:49:37 for traditional markets interactive
1:49:39 brokers
1:49:41 metatrader
1:49:51 all right and then so it'll give you
1:49:53 some of those so you can figure that out
1:49:54 then you can say things like
1:49:56 um can you show me the code for a stop
1:50:02 loss
1:50:04 bot
1:50:06 um with triggers I don't even know what
1:50:08 kind of triggers
1:50:09 and explain
1:50:13 how the code
1:50:15 works
1:50:20 sure I can give you a simple example
1:50:23 written in Python before diving into the
1:50:26 code you'll need to have python
1:50:27 installed on your thing
1:50:29 so I'm going to stop it I'm going to go
1:50:32 um I don't want
1:50:35 python but the code
1:50:40 for a major trading
1:50:43 platform
1:50:45 which one should I use and which
1:50:50 code is it
1:50:54 and I'm kind of purposely writing shitty
1:50:56 questions here
1:50:58 so this is going to do something for
1:51:00 metatrader
1:51:02 they have their own built-in programming
1:51:05 language called them ql4 and 5
1:51:08 respectively
1:51:12 so it's starting to write me code
1:51:17 so
1:51:19 you know so so what I would say is just
1:51:21 start just go to chat GPT and start
1:51:23 asking a question start interacting with
1:51:25 it like it's an expert and and
1:51:28 just tell it what you need to figure out
1:51:32 by the way the Nuggets game during the
1:51:34 fourth quarter Denver's up 75 71. I'm in
1:51:38 Denver by the way
1:51:40 all right
1:51:45 we love you thank you for sharing your
1:51:47 AI journey I appreciate that unlike
1:51:50 these other haters in here going let's
1:51:52 obvious everyone knows that Kyle you're
1:51:54 a stupid dumb dumb
1:51:56 we don't like stupid Dum Dums on Tick
1:51:59 Tock we want hot chicks that twerk
1:52:02 [Laughter]
1:52:08 oh it'd make us such a good troller
1:52:10 someday
1:52:11 haters
1:52:14 making it actually more effective than
1:52:16 positive feedback in this effed up world
1:52:18 yeah I know whatever
1:52:21 whatever
1:52:22 trolls are boring I think trolls are
1:52:25 boring
1:52:26 it's kind of like this AI stuff like
1:52:28 it's really easy to go to this negative
1:52:29 place this you know the future where the
1:52:31 robots kill us I feel I feel like that's
1:52:34 like the lazy way out like isn't it more
1:52:36 interesting like let's get curious about
1:52:38 it and figure out what this [ __ ] can do
1:52:40 and see if it can be a superpower for us
1:52:42 I mean it does two things it makes it
1:52:45 less boring right
1:52:47 it gives you hope but also it like puts
1:52:50 you a bit more in control because this
1:52:51 is going to [ __ ] up a lot of jobs so why
1:52:54 not learn it
1:52:55 and be one of the ones that knows how to
1:52:57 use it when the tidal wave hits
1:53:01 um say I wanted to copy the look of the
1:53:03 Nike website how would I do that
1:53:06 um there are
1:53:08 there are now ai sites that do that so
1:53:12 um
1:53:13 I don't know what they're called I know
1:53:15 they exist though go to futurepedia.io
1:53:17 that's why I did that I'm sure there's a
1:53:19 web design section in there I'm pretty
1:53:22 sure you can just point there's a couple
1:53:24 of tools that you can just put in the
1:53:26 URL of a website and it will it will pop
1:53:28 it into figma it'll it'll just literally
1:53:31 pull everything and pop it into layered
1:53:33 figma designs I'm pretty sure that
1:53:36 exists so
1:53:37 go to futurepedia.io
1:53:41 um that's why I have that there because
1:53:43 there's I mean there's there's thousands
1:53:45 of these tools now so it's really hard
1:53:47 to know what's what writing a novel has
1:53:50 never been easier any Bots or plugins
1:53:52 that can help chat GPT keep it organized
1:53:55 I the the plugins right now are are
1:53:58 really kind of janky and weird but I'd
1:54:00 go go search for novel or writing in uh
1:54:05 in the plug-in store and see if there's
1:54:07 anything there
1:54:08 um I'm organizing all my [ __ ] outside
1:54:10 well I'm really not organizing [ __ ] to
1:54:14 be quite honest I have 80d so my
1:54:17 executive function the whole frontal
1:54:19 lobe here is essentially a piece of
1:54:21 Play-Doh it's a non-functional frontal
1:54:23 lobe so so my executive functions pretty
1:54:26 low
1:54:27 um if I had one I would be organizing
1:54:30 this somewhere outside of chat GPT
1:54:32 because it's Chachi PT is such a
1:54:36 shitty
1:54:37 um thin piece of software like it was
1:54:40 built in two weeks and they've been
1:54:42 improving it and adding things to it but
1:54:44 it's still a very incomplete piece of
1:54:46 software so I would I would be
1:54:48 organizing stuff over in notion or
1:54:51 whatever your note-taking you know tool
1:54:54 of choices Google Sheets something like
1:54:56 that
1:54:58 thanks for being brave for all of us GI
1:55:01 Joe Transformers Thundercats generation
1:55:03 roll on
1:55:05 how about designing a house holy [ __ ]
1:55:08 some of the uh some of the house designs
1:55:10 in mid-journey oh oh there's my website
1:55:14 designs those are not bad
1:55:19 it still doesn't do text well drama faga
1:55:23 pazama
1:55:24 [Laughter]
1:55:26 but those actually look pretty good the
1:55:27 one with the those look more like grapes
1:55:29 than eggplants in the upper left but the
1:55:31 colors are kind of cool
1:55:33 so those are all not bad and then what
1:55:36 else did I do oh these were the t-shirts
1:55:37 these are the eggplant t-shirts
1:55:40 those are pretty good
1:55:42 and notice it put them on models and on
1:55:44 t-shirts so this was I used that Poe
1:55:47 mid-journey writing prompt writing thing
1:55:50 I don't know what the hell that is but
1:55:52 it's cool looking that's definitely
1:55:54 eggplant like but it's like it's an
1:55:56 eggplant that's having grape babies
1:56:00 um that does not look like eggplants at
1:56:01 all and that's sort of pizza-ish that's
1:56:04 at least a design style that you could
1:56:06 you could you could use so that's
1:56:09 interesting
1:56:12 um
1:56:13 let me go to I'm gonna go to
1:56:18 I'm gonna go to Facebook
1:56:23 and then I'm gonna go to
1:56:32 the
1:56:37 mid-journey official
1:56:41 and then within here I'm going to search
1:56:44 for
1:56:45 architecture
1:56:51 so so I'm on
1:56:54 um mid-journey mid-journey's official
1:56:57 channel on
1:56:59 um Facebook
1:57:01 and let me see if I can find some
1:57:03 architecture
1:57:05 um yeah there's a cool one
1:57:07 like like some of the some of the things
1:57:10 Architects are doing with interior and
1:57:12 exterior spaces are just insane like
1:57:15 look at these those are super cool
1:57:19 so I think one of the things we're going
1:57:20 to start seeing is we're going to start
1:57:21 seeing Architects and and designers
1:57:24 using mid-journey for inspiration and
1:57:26 then rolling that stuff into the real
1:57:29 world look at those designs I mean these
1:57:30 are just [ __ ] insane
1:57:33 look at this
1:57:35 and it's like if I'm an architect and I
1:57:38 see that like you know I either love it
1:57:40 or I hate it but like that's gonna this
1:57:42 is gonna Inspire like whole
1:57:45 like a whole generation of Architects to
1:57:47 just
1:57:49 start thinking in completely different
1:57:51 ways and bringing these designs into the
1:57:54 world
1:57:57 um yeah these are some more cool ones
1:57:59 like you know just sort of
1:58:00 weird glass
1:58:03 modern
1:58:06 you know George Jetson like this George
1:58:08 Jetson [ __ ]
1:58:09 speaking of gen xers
1:58:12 that's [ __ ] Jetson's territory right
1:58:14 there
1:58:17 um
1:58:19 that's cool look at this thing this is
1:58:21 it's not Art Deco what is this style not
1:58:23 Baroque what is this Art Nouveau look at
1:58:26 this thing
1:58:28 [ __ ] gorgeous
1:58:31 I mean come on
1:58:34 tell me you don't want to like see some
1:58:36 of these things in the world
1:58:38 look how beautiful that is look at the
1:58:40 colors
1:58:45 there's a boudoir
1:58:48 that is a boudoir
1:58:52 I mean look at these things crazy crazy
1:58:55 crazy [ __ ]
1:58:56 look at that
1:58:59 I mean that looks like where you know
1:59:03 I don't know 20 000 Leagues Under the
1:59:06 Sea was written
1:59:10 what else we got in here wow that's
1:59:12 beautiful
1:59:17 oh that one's beautiful look at that
1:59:25 nuts huh
1:59:30 all right
1:59:32 so hey everybody my name is Kyle Shannon
1:59:34 this is the AI learning lab
1:59:36 boom
1:59:38 hold
1:59:39 boom if you're new here
1:59:42 these URLs are chat GPT is the top one
1:59:46 and then bing.com is free chatgpt well
1:59:50 the top chat cbd's free but gpt4 is not
1:59:53 but if you want to use gpt4 for free go
1:59:56 to bing.com and use their chat that's
1:59:58 that and then Poe is another alternative
2:00:00 where you can use lots of different
2:00:02 models
2:00:03 how about writing oh I just lost my
2:00:05 place dang it
2:00:07 um
2:00:08 [Music]
2:00:10 wait why Bing again I'm in China and you
2:00:13 know
2:00:14 that's what we use daily I'll come back
2:00:16 to that well
2:00:18 why Bing again
2:00:20 so I don't know what you're seeing in
2:00:23 China but if you go to Bing
2:00:28 that's Poe that's not being hold that's
2:00:30 Bing
2:00:32 if you go to Bing
2:00:35 you'll see this and then you have your
2:00:38 menu across the top you should see
2:00:41 a chat button and if you click on that
2:00:43 chat button
2:00:45 that's chatgpt but I think open Ai and
2:00:50 chat gbt is blocked in China so I assume
2:00:52 you won't see that
2:00:55 but I don't know but you tell me that
2:00:56 would be interesting to know
2:00:59 um
2:01:02 all right sorry I got lost in my
2:01:04 questions I'm going back to try to grab
2:01:05 some questions all right
2:01:08 um how about writing your Memoirs could
2:01:10 it tell me
2:01:12 wait
2:01:14 could I tell it my memories and it
2:01:18 writes them down Hemingway style yeah
2:01:20 absolutely
2:01:22 um so here check this out this will be a
2:01:24 fun exercise
2:01:25 let's go let's go play let's go play
2:01:30 all right we're gonna go to chat GPT you
2:01:33 can make money
2:01:36 all right we're gonna go to chatbt we're
2:01:38 gonna go to gpt4
2:01:41 and I am gonna pop on over to LinkedIn
2:01:47 link linked
2:01:49 in wait
2:01:52 Linked In
2:01:57 and I'm going to go find my profile
2:02:00 this is just an experiment this is a
2:02:02 dumb idea I have and I'm going to copy
2:02:05 my entire LinkedIn profile
2:02:15 down to there all right then I go back
2:02:17 to chat EBT and I'm going to say
2:02:21 um based on this
2:02:24 LinkedIn profile
2:02:26 colon
2:02:31 paste
2:02:34 um write me
2:02:37 a memoir
2:02:42 uh
2:02:44 memoir
2:02:49 in
2:02:51 the style
2:02:54 of
2:02:57 I don't know Hemingway
2:03:01 which is what you said in the style of
2:03:03 Hemingway
2:03:05 um
2:03:06 for my
2:03:09 early
2:03:11 internet experience
2:03:17 make it
2:03:19 five paragraphs
2:03:21 just to keep it a little
2:03:23 faster
2:03:25 all right let's see what it does so it
2:03:28 read my LinkedIn profile
2:03:30 and
2:03:33 is it writing
2:03:35 in the early dawn of the internet there
2:03:37 was a strange sense of anticipation and
2:03:39 certain uncertainty it was like standing
2:03:42 on the precipice of a cliff peering into
2:03:45 the abyss of the New Frontier waiting
2:03:47 for the first Spark to ignite the
2:03:48 darkness the world was digital but not
2:03:51 yet connected isolated in pockets of
2:03:54 binary Solitude and then the first rays
2:03:58 of sunlight breaking Over the Horizon
2:03:59 the connection came not in a sudden
2:04:02 burst of light but in the slow steady
2:04:04 growth of a new day I found myself at
2:04:06 the heart of this digital Evolution a
2:04:08 Pioneer in a Land of endless
2:04:10 possibilities
2:04:11 the early days were curious blend of
2:04:15 exhilaration and frustration the
2:04:17 frontier was wild and Untamed the
2:04:19 terrain unknown the possibilities were
2:04:21 boundless yet the tools were crude the
2:04:24 protocols undefined we were explorers
2:04:27 charting an Unseen World building our
2:04:29 own maps and ventured further into the
2:04:31 unknown this is good I mean you know
2:04:35 it's not anything I would write but
2:04:37 in this new world I found a canvas for
2:04:40 my creativity a place where worlds and
2:04:42 images and ideas could be woven together
2:04:43 to connect people to create experiences
2:04:45 the internet was not just a network of
2:04:47 computers it was a network of minds and
2:04:50 in this network I saw the potential to
2:04:52 create something truly remarkable with a
2:04:54 team of like-minded pioneers I
2:04:56 co-founded agency.com one of the first
2:04:58 interactive marketing firms we were the
2:05:00 blacksmiths of the digital age forging
2:05:02 the tools and the strategies that would
2:05:04 shape the landscape of the online World
2:05:06 which is yeah I mean it's a little
2:05:09 blowing Sunshine up my ass but but yeah
2:05:12 so yes it can hope that helps uh and and
2:05:16 uh the uh the interpretive reading huh
2:05:19 how about that bringing dusting off them
2:05:22 old acting skills I don't know if that
2:05:23 was acting that was just thank you huh
2:05:27 it's like it's like winning an Oscar
2:05:29 getting that little heart on the head
2:05:31 I wonder if chat GPT understands food
2:05:33 science enough to help develop or refine
2:05:35 recipes holy [ __ ] yes it does so I'll
2:05:38 tell you a cool story I did um
2:05:40 I've got in my family I've got people
2:05:42 with chronic pain
2:05:44 and there's this pain juice that we use
2:05:46 called rightful and it's it's got I
2:05:49 don't know it's got [ __ ] in it that
2:05:50 makes the pain better
2:05:52 and the company went out of business so
2:05:54 we bought a bunch of bottles and we're
2:05:56 winding down on the bottles so but on
2:05:58 the bottles they have the percentages of
2:06:00 the active and inactive ingredients
2:06:03 and so I went into chat GPT pasted in
2:06:07 all the ingredients I actually had two
2:06:08 different versions of the the recipes
2:06:10 and I had to compare why are these
2:06:12 different what's different about them
2:06:14 then I had it tell me what do each of
2:06:16 these active ingredients do what do the
2:06:18 inert ingredients do
2:06:21 um are there any of the inert
2:06:22 ingredients that I actually don't need
2:06:26 um and then I had it calculate up a
2:06:29 recipe it basically built me a recipe
2:06:31 for doing it so yeah for things like
2:06:34 like you can give it three three
2:06:36 ingredients and tell it to make you a
2:06:37 meal plan for the week focusing on those
2:06:39 ingredients it'll do it it's remarkable
2:06:43 can we stop pretending AI isn't very
2:06:46 dangerous well I'm not pretending it's
2:06:48 not very dangerous I'm just not focusing
2:06:50 on the robots will kill us future I'm
2:06:52 trying to focus on the what can we do
2:06:54 with it right now future
2:06:56 um
2:06:57 if if the implications in
2:07:00 can we stop pretending it's not very
2:07:03 dangerous I don't know what you mean I I
2:07:06 assume what you're doing then is working
2:07:08 really hard Building Technology to keep
2:07:11 us safe
2:07:12 yes or no you're just on a soapbox going
2:07:15 the robots are evil and you should
2:07:17 ignore them I'm not in that place
2:07:20 I'm like the robots are incredibly
2:07:22 powerful and we're lucky enough to be
2:07:24 able to play with them and if we can get
2:07:26 curious enough and start exploring them
2:07:28 and do positive things with them we will
2:07:31 have a disproportionately positive
2:07:32 impact on the future of the industry
2:07:35 which might prevent the future that
2:07:38 you've bought into
2:07:40 I'm not saying they're not dangerous
2:07:42 this is not where I'm focusing
2:07:45 there's a lot of people focusing on that
2:07:47 including Hollywood
2:07:48 let the tension build into a new
2:07:50 universe how accurate
2:07:53 is it providing code I have no Tech
2:07:55 skills and messing around with it
2:07:57 um it makes a lot of mistakes so the
2:07:59 thing about chat EPT is that
2:08:03 it seems very much like it's aware of
2:08:06 what it's doing it's actually not it's
2:08:08 just a probability engine
2:08:10 so I like to describe it as mansplaining
2:08:14 as a service it will confidently answer
2:08:16 you even if it's wrong
2:08:18 so
2:08:19 the way the nice thing about code is
2:08:22 that it either works or it doesn't so if
2:08:24 you don't know anything about code so
2:08:25 this was my big Epiphany with chat EBT I
2:08:29 was sitting watching TV with my wife in
2:08:31 December and
2:08:33 I had heard that it could write code and
2:08:36 so I said write me a piece of python
2:08:38 code that takes an input or something
2:08:40 like that
2:08:41 and it wrote a piece of code and then
2:08:43 and then I realized I actually don't
2:08:46 know how to run python code so I said
2:08:48 hey chatgpt can you help me run python
2:08:50 code and it said sure go to your
2:08:52 terminal program and do a pip install
2:08:54 and it was like all this command line
2:08:55 [ __ ] and I'm like no I don't want to
2:08:57 do command line isn't there a website
2:08:59 where I can just paste this python code
2:09:01 and see if it works it was like oh sure
2:09:02 here's three of them
2:09:04 so I picked one of them and then it said
2:09:06 okay you're going to take this code
2:09:08 you're going to paste it into the left
2:09:09 window you're going to hit the play
2:09:10 button and then it's going to run the
2:09:12 code I'm like cool so I paste the code I
2:09:14 hit the play button and it crashes and
2:09:16 and it and I I copied the error code and
2:09:18 I go back to chat CPT and I'm like you
2:09:20 gave me this error code it's like oh
2:09:22 well you need to install this Library
2:09:23 clearly and I'm like how do you install
2:09:25 this Library
2:09:27 and it said okay on that page in the
2:09:29 left hand corner you click on this
2:09:31 button and then the upper right hand
2:09:32 corner you search for this library and
2:09:34 then you hit install and I did that and
2:09:36 then the code ran within 90 minutes I
2:09:40 had a fully functional python
2:09:42 application that took some inputs about
2:09:45 a topic and generated five pieces of
2:09:47 social media content in 90 minutes never
2:09:50 having written a line of python in my
2:09:52 life
2:09:53 so the cool thing about code is it
2:09:55 either works or it doesn't so have it
2:09:57 teach you how to run the code how to
2:09:59 install the code whatever system you're
2:10:01 on and then just go start testing the
2:10:03 code is it going to be easy no is it
2:10:06 going to break some [ __ ] and not work
2:10:08 and have you be frustrated for weeks on
2:10:10 end yes
2:10:11 when I tried to get stable diffusion
2:10:13 running I failed nine times to get
2:10:16 Google collab like like an instant of
2:10:19 stable diffusion up and running nine
2:10:21 times it took me like two months not
2:10:23 because it actually took me two months
2:10:24 it was like I would try it I would spend
2:10:26 like an hour hour and a half and then it
2:10:28 would fail and I'd just be like ah [ __ ]
2:10:30 it and then I I'm like oh no I should
2:10:33 learn it and I come back three days
2:10:35 later and I tried again it failed I did
2:10:36 that nine times
2:10:38 when I finally got it to work I was like
2:10:43 that's where it is right now I mean the
2:10:45 good news about these tools is we we you
2:10:47 know we are lucky enough to be playing
2:10:50 with some of the most powerful tools
2:10:52 ever put in the hands of mankind that's
2:10:54 the good news the bad news is it's early
2:10:56 and all this shit's janky as [ __ ] and
2:10:59 you got to be a little geeky to figure
2:11:00 it out
2:11:02 um if you're new here my name is Kyle
2:11:03 Shannon this is the AI learning lab I'm
2:11:04 an entrepreneur I'm a founder and a CEO
2:11:07 of a company called storyvine which is
2:11:09 an automated video platform has nothing
2:11:11 to do with AI although it does now
2:11:14 um
2:11:14 I've been kind of deep diving into the
2:11:16 say I stuff for for about a year now uh
2:11:19 I am not a machine learning expert or a
2:11:21 data scientist I'm an entrepreneur and I
2:11:24 was there for the early days of the web
2:11:26 and this feels very much like the early
2:11:28 days of the web but like on steroids and
2:11:31 coming way faster and way more profound
2:11:33 so that's why we're here I'm here to
2:11:35 demystify what I can about AI ask
2:11:39 questions below tap the little heart
2:11:41 button because I think that lets people
2:11:42 know we're here I don't know much about
2:11:44 Tick Tock there will be no twerking
2:11:46 you're welcome
2:11:47 and my hair is behaving tonight but I'm
2:11:50 thirsty and my throat's dry hold
2:11:54 um
2:11:58 okay
2:12:03 it sent me to Azure with Azure open AI
2:12:07 what sent you there
2:12:09 I uh be be more
2:12:12 um
2:12:12 detailed Terror I'm not quite sure
2:12:14 you're talking about
2:12:15 wait wiping again oh you're the China
2:12:18 person
2:12:19 um Brown Knight
2:12:21 but Google search is better than Bing
2:12:23 Okay so
2:12:25 I don't think you have access to chat
2:12:27 GPT in bing Google search is better than
2:12:31 Bing but Bing chat kicks the [ __ ] out of
2:12:34 Google Search
2:12:37 um and and Bing Bing chat also kicks the
2:12:40 [ __ ] out of Google bard
2:12:43 um I'm pretty sure China's got all this
2:12:44 stuff blocked I don't think they want
2:12:46 you getting answers from a large
2:12:48 language model
2:12:50 I don't mean to laugh but like
2:12:52 like it's a tool that just gives you
2:12:54 answers that there's probably a lot of
2:12:56 answers in there they don't want you to
2:12:57 have is my guess do you ever feel bad
2:13:00 for chat EBT when you ask it hard
2:13:02 questions it's really funny you say that
2:13:04 Gregory Gregory Murphy um
2:13:07 so so I absolutely find myself thanking
2:13:11 it and saying please and hey good job
2:13:14 you know now could you do this I have
2:13:17 absolutely felt pangs of guilt
2:13:21 like the sixth time I ask it to rewrite
2:13:23 something
2:13:25 like if I were dealing with a human
2:13:27 being and I asked them to write me
2:13:28 something and what they gave me back was
2:13:29 shitty and then I said okay go rewrite
2:13:31 it and then what they gave me back was
2:13:32 shitty and then I said okay go rewrite
2:13:34 it I don't think I would make it to the
2:13:36 third iteration right at some point I'd
2:13:38 just go oh [ __ ] I'll just do it myself
2:13:39 right it's just too uncomfortable I
2:13:41 don't feel like explaining it
2:13:43 so I've absolutely had pangs of guilt
2:13:46 and then I realized oh it's a [ __ ]
2:13:47 machine it's going to take it 10 seconds
2:13:49 like it doesn't care
2:13:51 and so and and then I ask it and I feel
2:13:54 like this giddy like oh like I got one
2:13:57 over on it so yeah it's like my
2:14:00 experience working with chat EPT is very
2:14:03 much like you're working with a remote
2:14:04 co-worker that happens to be way smarter
2:14:07 than you'll ever be like that's what
2:14:08 it's like and so yeah so I find myself
2:14:11 what like whatever my you know
2:14:13 psychological trappings are I find
2:14:15 myself falling into it because I'm like
2:14:17 I'm up here in the work I'm trying to
2:14:19 figure out the work and then I'm like oh
2:14:20 I gotta ask this thing this thing and
2:14:22 yeah it's very much like that so
2:14:25 C plus plus
2:14:27 whatever
2:14:28 chat EBT is a language model not very
2:14:31 smart sir I understand that well it it's
2:14:33 an incredible
2:14:35 um probability engine it's not aware of
2:14:38 what it does so I agree with you there
2:14:39 but you know again A lot of times
2:14:41 technologists will come in here and and
2:14:44 you make a big deal out of making
2:14:46 distinctions like is it smart or not I
2:14:49 don't know it's a probability engine so
2:14:50 you're right it is it is as smart as a
2:14:53 calculator right but what it does is
2:14:56 [ __ ] Magic
2:14:57 when it writes you a screenplay when it
2:15:00 writes you a table comparing five
2:15:02 different obscure Concepts and gets it
2:15:04 right
2:15:05 that's [ __ ] Magic
2:15:06 so sure technically it's not smart but
2:15:09 it's way smarter than I'll ever be on
2:15:11 all of these different topics so again
2:15:14 like I
2:15:17 I don't know I would just caution
2:15:19 against doing things like it like trying
2:15:22 to label it try to put it in a box
2:15:23 trying to judge it and and and dismiss
2:15:26 it right because if you dismiss it it
2:15:29 kind of discounts the fact that this is
2:15:30 coming and the changes that this stuff's
2:15:33 gonna cause are going to be profound and
2:15:35 they're going to be Swift
2:15:37 they're going to be Swift like you know
2:15:40 agencies copywriters
2:15:43 um
2:15:43 like this is this is gonna change some
2:15:46 businesses and it's going to change them
2:15:47 quickly the legal profession
2:15:50 how are people feeding info into chat
2:15:52 GPT so
2:15:54 the initial model if so sometimes when
2:15:57 you use chat gbt it's like hey just so
2:16:00 you know I only know up to September
2:16:02 2021 anything after that I don't know
2:16:04 the new version of it that's connected
2:16:07 to the internet kind of solves for that
2:16:08 but basically what that's indicating is
2:16:11 there's this base large language model
2:16:13 so they took a bunch of years to find as
2:16:16 many data sources as they could so
2:16:17 public internet sites and libraries and
2:16:21 blogs and Wikipedia and just all these
2:16:23 public sites and they they got them into
2:16:25 the however you embed these things right
2:16:29 um
2:16:30 and and sort of dumped it all in and
2:16:33 then on in September 2021 I guess they
2:16:35 were done enough that they were like
2:16:37 let's lock down this model and then
2:16:39 we'll start fine-tuning it and the
2:16:41 fine-tuning there was kind of this kind
2:16:43 of the base model and then three
2:16:45 additional fine-tuning layers on top of
2:16:47 it to get to GPT 3.5 and gpt4 and that's
2:16:52 what chat gbt is built on
2:16:55 um so so that's what it's that's what
2:16:58 the base layer is trained on then all
2:17:00 the fine tuning is human beings using it
2:17:03 and going that was a good answer that
2:17:04 was a bad answer that was a good answer
2:17:05 that was a bad answer and and that's
2:17:08 sort of making it smarter and smarter
2:17:09 and smarter and the reason that
2:17:12 um open AI put chat gbt out for free is
2:17:16 generally If the product is free you're
2:17:19 the product
2:17:20 so
2:17:22 chat GPD is free why because they need
2:17:24 as many people as possible using it in
2:17:27 as many ways as possible so they can
2:17:29 continue to fine-tune the model so it
2:17:31 gets better and better and better over
2:17:33 time so we are all as we use it
2:17:36 contributing to it getting better but
2:17:38 we're not contributing to that base
2:17:40 model that will be in gpt5 there'll be
2:17:42 an updated base model
2:17:44 so I hope that helps
2:17:47 built in two weeks what Okay so
2:17:50 this is a fascinating story this is in
2:17:52 one of the interviews with Sam Altman
2:17:55 um gpt3 was was released in
2:18:00 I think fall late summer or fall of
2:18:03 2020.
2:18:04 and in the documentation of gpt3 were
2:18:08 the instructions for how to build chat
2:18:10 GPT
2:18:12 and Sam Altman assumed that someone
2:18:15 would build it and nobody built it and
2:18:18 then in October of 2022
2:18:21 Microsoft was going to announce that
2:18:24 they made this billion dollar investment
2:18:26 in open AI
2:18:27 and I guess you know Sam Altman wanted
2:18:30 something that they could show publicly
2:18:32 so he said to his Tech Team go build
2:18:34 chat GPT they built it in two weeks
2:18:37 they launched it and six weeks later
2:18:39 they had a hundred million users
2:18:41 it's insane it's insane
2:18:45 and there's interviews with Sam Altman
2:18:47 after
2:18:48 um
2:18:49 you know after this happened he's like
2:18:52 we we didn't think anyone was going to
2:18:54 use this like like I mean at that scale
2:18:56 like we just put this thing out there as
2:18:58 kind of like a proof of concept he's
2:19:00 also said since they're not doing any
2:19:02 other consumer facing apps like they're
2:19:04 they're about the underlying technology
2:19:06 chechi PT was really just a proof of
2:19:08 concept that Microsoft could point to it
2:19:10 and say look how cool that thing is
2:19:11 without having to go use the back end
2:19:13 tools so yeah yeah built in two weeks
2:19:16 100 million users in six weeks and it
2:19:20 completely changed everything
2:19:21 everything's different because of that
2:19:24 product that choice that announcement
2:19:26 it's amazing
2:19:28 all right disagree that exploring the
2:19:30 negative implications of AI is lazy so
2:19:33 much is at stake it's important I
2:19:34 totally agree that it's important it's
2:19:38 if you're talking about the negative
2:19:40 implications in an active sense and
2:19:43 you're actually working on it cool like
2:19:45 absolutely we need to do it we need
2:19:47 smart people attacking these problems
2:19:49 the fact that Jeffrey Hinton left Google
2:19:51 so that he could address this stuff it's
2:19:53 great
2:19:54 my the thing I'm combating against right
2:19:57 now is there are a lot of people that
2:20:00 are just defaulting to the robots will
2:20:02 kill us without actually understanding
2:20:04 what the robots are and what they do and
2:20:06 that's what I'm trying to do is let's at
2:20:08 a minimum get curious about what they
2:20:10 are have chatgpt write you a poem or a
2:20:14 table or a piece of code it doesn't
2:20:15 [ __ ] matter just go use it so that
2:20:18 you actually understand it from the
2:20:19 inside and at that point you can say oh
2:20:22 this isn't as scary as I thought or this
2:20:25 is scarier than I thought either one of
2:20:27 those is fine because they're coming
2:20:29 from a place of knowledge rather than
2:20:31 ignorance so all I'm trying to battle is
2:20:33 the the ignorance and and blanket
2:20:36 statements like you know AI is evil
2:20:39 tend to be more from people that are not
2:20:41 coming at it from the inside you sound
2:20:43 Gregory like you're coming at it from
2:20:45 the inside so totally respect that point
2:20:46 of view I'm just not there I'm an
2:20:48 entrepreneur so I'm trying to look at
2:20:50 what can these tools for me do for me as
2:20:53 an entrepreneur and as a professional
2:20:55 and I'm trying to head off at the pass
2:20:57 being blindsided by this stuff because I
2:21:00 think it's coming really fast and I
2:21:02 think a lot of people's jobs are going
2:21:04 to be decimated and and they're they're
2:21:07 going to be completely caught off guard
2:21:08 and that's that's why this channel
2:21:10 exists
2:21:11 is to try to I don't know make some
2:21:14 little dent in that problem because it's
2:21:16 I think it's a big big problem
2:21:19 oh
2:21:20 Denver Nuggets just won the NBA finals
2:21:28 people are crying
2:21:31 it's in Denver
2:21:33 I'm hearing fireworks outside
2:21:37 yokich
2:21:41 [Applause]
2:21:44 you're a big goof
2:21:46 [Applause]
2:21:50 go
2:21:52 congrats to the Nuggets I live in Denver
2:21:54 huh oh it's probably it's probably gonna
2:21:56 get noisy gonna be a lot of drunk people
2:21:58 running around here shortly there go the
2:22:01 fireworks uh these d-gens remind me of
2:22:03 Gowdy okay I don't know what that means
2:22:05 Harry Potter
2:22:06 located in Portugal oh it just jumped me
2:22:09 to the bottom again ah dang it
2:22:11 all right ask me some questions I'm
2:22:13 gonna that's not too late I'll keep
2:22:15 hanging out here for a bit
2:22:17 oh it looks like we've got some new
2:22:18 people if you're new here my name is
2:22:19 Kyle Shannon this is the AI learning lab
2:22:22 I put these URLs up here if you've never
2:22:24 tried chat GPT I strongly encourage you
2:22:26 to do it just to see what it is
2:22:30 um and and uh
2:22:32 just to learn what it what it makes
2:22:35 possible
2:22:36 um start out by asking it to write you a
2:22:38 country song or a poem or a short story
2:22:41 or I don't know a grocery list have it
2:22:44 make recipes
2:22:46 um and then if you've got any questions
2:22:47 Denver 88 Miami 89 nope it's the final
2:22:52 well the final is whatever the Nuggets
2:22:55 won it was under both of them were under
2:22:57 100 points
2:23:00 um Google and Bing are Sean X's [ __ ]
2:23:03 um any idea how I can chat with chat GPT
2:23:06 in my car on the move that would be cool
2:23:09 yeah so there's so just go to Twitter
2:23:13 and look up
2:23:15 um
2:23:16 iPhone chat GPT voice you can also
2:23:20 so I've got I've got the the chat
2:23:23 GPT website the official chat CPT
2:23:26 website
2:23:28 um as a as a bookmark on my phone so I
2:23:30 have it on my home screen
2:23:32 and then you can just when you you know
2:23:35 type into text you can just hit the
2:23:36 little microphone icon in the bottom
2:23:38 right corner of your phone and just talk
2:23:39 and you can just ask it questions and
2:23:41 it'll come back
2:23:43 um there are apps out there there's one
2:23:45 called
2:23:47 what the hell is it called ask
2:23:51 there's a lot of boom booms going on in
2:23:53 my neighborhood
2:23:56 there are there are a number of apps in
2:23:58 the App Store that do it I
2:24:00 I don't know enough about them but but
2:24:02 there are there are apps now that some
2:24:04 use Siri there's some guys that'll
2:24:06 figure out how to hook Siri up to chat
2:24:08 gbt
2:24:10 um
2:24:10 there's probably some in the App Store
2:24:12 but I would imagine that they're going
2:24:14 to cost a decent amount of money and
2:24:16 they're going to be janky pieces of [ __ ]
2:24:18 so
2:24:19 um I know it can be done I would just go
2:24:21 on Twitter and look up you know iPhone
2:24:24 chatbot chat GPT or something like that
2:24:27 and you you'll probably see some
2:24:29 projects there I wish I knew more on
2:24:31 that one I'll look that one up for next
2:24:33 time
2:24:33 can you tell us about your experiment
2:24:35 with the internet boom days and what
2:24:37 that was like uh oh my experience with
2:24:39 the internet boom days and what that was
2:24:41 like it was it was really insane so
2:24:44 um
2:24:45 I was desktop publishing in kind of late
2:24:48 93 early 94.
2:24:51 I I was pursuing an acting in a
2:24:53 screenwriting career at the time and I
2:24:55 was doing desktop publishing as my day
2:24:56 job and
2:24:59 I was into Tech like I like technology
2:25:01 and I I remember like I distinctly
2:25:03 remember buses in New York City had like
2:25:05 Wired Magazine ads and mondo 2000 I
2:25:09 don't know if you remember Mondo 2000
2:25:11 but there was clearly like this there
2:25:13 was clearly something buzzing about this
2:25:16 internet thing and
2:25:18 um I was on an online Forum called echo
2:25:21 which was a dial-up bulletin board
2:25:22 system which if you don't know a dial-up
2:25:24 bbs's are your computer would call up
2:25:27 another computer and then you could chat
2:25:29 with people in text only
2:25:32 um and and uh and so it was on I was on
2:25:35 this group called Echo and I ended up
2:25:37 starting the World Wide Web section of
2:25:40 echo
2:25:42 um because I kept hearing about this
2:25:44 thing called the world wide web and I
2:25:45 tried to learn about it and no one on
2:25:47 Echo knew about it and so I'm like well
2:25:49 I'll start a group and then I started
2:25:52 um a group called The World Wide Web
2:25:54 artist Consortium where we got together
2:25:56 in person and talked about this web
2:25:58 stuff and then I had an idea for a
2:26:00 magazine
2:26:00 and I went on vacation with my wife and
2:26:03 I learned enough HTML to put together
2:26:04 the first issue of a internet magazine
2:26:07 and and figured out how to upload it to
2:26:10 some server in California and then three
2:26:13 weeks later someone on Echo said hey did
2:26:15 you know that um Urban desires was
2:26:18 written up in Liberation the Parisian
2:26:20 newspaper
2:26:21 and three weeks after I uploaded these
2:26:24 files to some computer in California
2:26:25 it's written about in Paris
2:26:27 and I just had this Moment of clarity I
2:26:30 went to the international news stand in
2:26:31 Times Square I bought
2:26:33 the paper copy of this newspaper I
2:26:35 flipped it open and sure enough there's
2:26:37 this full page article with screenshots
2:26:39 of my website in French from Paris
2:26:43 and I just had this moment of like holy
2:26:47 [ __ ] the world just changed like the the
2:26:50 world's different and I remember being
2:26:52 in Times Square and
2:26:55 just like the bustle of Times Square
2:26:57 like all these people are walking around
2:26:59 and I'm like nobody knows nobody knows
2:27:03 what's coming nobody knows what's here
2:27:04 and it's like it just [ __ ] lit a fire
2:27:07 under me and that's exactly what's
2:27:08 happening with this AI stuff this is
2:27:10 this is the first time in my life since
2:27:13 that day in 1994 standing in the middle
2:27:16 of Times Square
2:27:18 where I'm like holy [ __ ] this is one of
2:27:22 those
2:27:24 like these don't these are generational
2:27:26 or sometimes once in a century right
2:27:28 like this is like fire language written
2:27:32 language printing press
2:27:35 Industrial Revolution right like these
2:27:37 don't come along that often we've we had
2:27:40 a lot of them in the 20th century right
2:27:43 we had turn of the century you have all
2:27:45 the media Innovation you've got
2:27:47 photography and film and radio and
2:27:50 television you've got all that sort of
2:27:51 stuff going then you have all of the
2:27:54 you know the mechanical right the steam
2:27:56 engine all of the sort of remnants of
2:27:59 the steam engine right so all of that
2:28:00 stuff happens in the 20th century then
2:28:03 in the 50s you get the computer
2:28:05 in the 70s you get the personal computer
2:28:07 I was just too young to be a part of
2:28:09 that one but but I was acutely aware of
2:28:12 it
2:28:13 then you have the world wide web and
2:28:15 then you could argue maybe that the
2:28:18 iPhone was one of them and that social
2:28:20 media was one of them but I think those
2:28:22 are all just extensions of
2:28:25 um
2:28:26 of the World Wide Web right the the
2:28:28 World Wide Web becomes this
2:28:31 kind of seminal moment where real people
2:28:33 get access to this incredibly powerful
2:28:35 infrastructure
2:28:37 same thing happened with chat GPT
2:28:39 machine learning and ai's been around
2:28:41 for decades and all of a sudden this
2:28:44 little piece of software by this company
2:28:47 called openai gets stuck out
2:28:49 tied to some press released about a big
2:28:51 Corporation investing money in them and
2:28:54 a hundred million people start using it
2:28:56 within six weeks
2:28:58 and it does what it does it's [ __ ]
2:29:01 remarkable
2:29:03 right it's so
2:29:06 that's like why am I on fire about this
2:29:10 and why do I get passionate about this
2:29:12 because of
2:29:13 like I've been in this movie before and
2:29:16 this is a way more exciting movie like
2:29:18 what was the output of the World Wide
2:29:20 Web the output of the World Wide Web was
2:29:22 a thing that looked like a page like a
2:29:23 shittily formatted piece of text with
2:29:26 blue underlined words on it and when you
2:29:29 clicked on them it took you to another
2:29:32 page with blue underlying words on it
2:29:34 that's what the internet was when we
2:29:36 started when I started
2:29:38 it was incredibly basic right
2:29:42 this stuff even though the tools are
2:29:45 very very basic and raw the output of
2:29:48 what they do is [ __ ] remarkable
2:29:51 like nothing we've ever seen nothing we
2:29:54 we as human beings have not had tools
2:29:57 this powerful ever
2:30:00 you know you could argue that the early
2:30:02 days of the World Wide Web there was
2:30:03 nothing actually revolutionary about it
2:30:05 wasn't doing anything markedly new other
2:30:08 than the infrastructure of the internet
2:30:11 was there and the underlying protocol
2:30:13 was there and then you know Tim
2:30:15 berners-lee creates this the the you
2:30:17 know HTML language and and that protocol
2:30:21 that that make turns it from a command
2:30:24 line interface to a click interface
2:30:26 right it was all about the interface
2:30:28 what's chat EPT it's all about the
2:30:30 interface you don't have to bring your
2:30:32 own data you don't have to know anything
2:30:34 it's literally just a search box so you
2:30:36 type words in it and then what you get
2:30:38 back is dramatically better than what
2:30:41 you typed in
2:30:43 so so it's an interface thing because
2:30:45 the text been there gpt3 was available
2:30:48 for us to play with for like a year and
2:30:50 a half before chat gbt came out I [ __ ]
2:30:52 around with it I'm like yeah that's kind
2:30:53 of cool but there was something about
2:30:55 the Simplicity of chat EVT that changed
2:30:57 everything
2:30:58 so that's
2:31:02 that's
2:31:03 what it's all about say what I'm saying
2:31:08 just see what I'm saying here it's not
2:31:10 stupid there's a language barrier
2:31:12 between the input and the
2:31:17 cambodia's
2:31:20 ability to pretend I don't quite know
2:31:22 what that means but yeah I agree it's
2:31:23 not it's not stupid but it's not aware
2:31:25 of what it's creating
2:31:27 do you think chat GPT will be close to
2:31:30 AGI in the next year I I don't know I
2:31:32 don't so AGI stands for artificial
2:31:35 general intelligence that's basically
2:31:38 when when the when the Tool is smarter
2:31:41 than the smartest human basically and
2:31:46 it is it is a very imprecisely defined
2:31:49 term like Sam Altman even when he wrote
2:31:51 he wrote that that paper kind of talking
2:31:53 about AGI and and how we need to be
2:31:56 careful
2:31:57 um even within that thing that he wrote
2:31:59 he kind of contradicted what it actually
2:32:01 is and what it means so I don't think
2:32:03 there's a clear definition of it
2:32:05 um there's a paper out by Microsoft
2:32:07 talking about gpt4 that it's showing
2:32:09 signs of AGI so there's some behaviors
2:32:12 that gpt4 does
2:32:14 that you could categorize as as HEI but
2:32:17 I think you start to look at things like
2:32:18 reasoning and awareness and
2:32:21 um
2:32:22 accuracy
2:32:24 um that that the current chat GPT
2:32:26 doesn't have that you know
2:32:29 the a system in the future that's called
2:32:32 AGI that that passes whatever that test
2:32:34 is is going to have to have some more
2:32:36 sophisticated stuff
2:32:38 I I'm personally of the mind that I
2:32:40 don't give a [ __ ]
2:32:42 like I I don't give a [ __ ] if if we call
2:32:46 it AGI or not I think I think the the
2:32:48 current gpt4
2:32:51 is so much more powerful than anything
2:32:53 we've ever had that I think it's
2:32:54 academic at this point whether we've hit
2:32:56 AGI or not you know the fact that this
2:32:59 can truly act as a creative
2:33:00 collaborative partner the fact that this
2:33:03 is gonna completely redefine entire
2:33:07 Industries like illustration and
2:33:09 copywriting and legal document
2:33:11 processing and
2:33:14 just [ __ ] every every aspect of
2:33:16 knowledge work and creative work is
2:33:18 going to come be completely redefined
2:33:21 it's already in a in a class of you know
2:33:25 profoundness that's above
2:33:28 what I ever thought I'd see in my life
2:33:30 it sent me to Azure to feed me the code
2:33:32 it gave me to build my AI model ah I see
2:33:36 I had to get approved for Azure so
2:33:39 there's there's prob so is yours
2:33:40 probably one place that you can run
2:33:42 virtual machines where you have to put
2:33:44 the code in there's Google collab
2:33:46 there's hugging faces got some things
2:33:48 you can run it in
2:33:50 um yeah those are all just places where
2:33:52 you're going to run the code so Azure is
2:33:54 fine
2:33:55 um I don't know anything about Azure
2:33:57 I've never done anything on it I have
2:33:59 done things on Google collab they're all
2:34:01 kind of the same
2:34:03 um they're sort of web interfaces that
2:34:05 are giving you access to really powerful
2:34:07 computers to run the code there are
2:34:09 probably things you could run locally on
2:34:11 your machine so you're not having to do
2:34:12 it on Microsoft's cloud services
2:34:16 um but there you have it
2:34:18 does cheap does chat GPT remember
2:34:21 results
2:34:22 um no it doesn't it will store your
2:34:26 prompt and it it remembers its results
2:34:28 within it within a chat up to a three
2:34:31 thousand
2:34:33 um word context window so there's this
2:34:34 kind of sliding 3000 word window uh that
2:34:38 it will remember within that
2:34:40 if you want to have memory and things
2:34:42 like that you need to start playing with
2:34:43 there's a there's a thing out there
2:34:45 called Lang chain and then there's other
2:34:47 things called Auto GPT that are kind of
2:34:49 automated things built on top of Lang
2:34:51 chain and Lang chain has a database in
2:34:55 it called a vector database so you can
2:34:57 train it on your own documents and then
2:34:59 you can put the results of your chats
2:35:01 into that Vector database so you can
2:35:04 create systems with memory but right now
2:35:06 chatgpt doesn't have it so give it a
2:35:10 year and like there'll be lots of things
2:35:12 with memory but right now you kind of
2:35:14 have to roll your own
2:35:17 any AI to make me a website I can't code
2:35:20 well just you know all the no code
2:35:23 website Builders out there so
2:35:25 um
2:35:26 uh whatever you call it I don't know
2:35:28 webflow is one of them
2:35:31 um just all of those like I would just
2:35:33 like if you can't code
2:35:36 um just go start with a template it
2:35:38 depends what you need it to do and then
2:35:40 once you have one of those up and
2:35:42 running you're gonna have chat gbt help
2:35:44 you add functionality to it
2:35:46 it's good at moderating my email when
2:35:48 I'm too rude which is often that's funny
2:35:51 yeah I mean that's actually a really
2:35:53 good use case of you know if if you
2:35:57 right in a style that that you want to
2:35:59 improve on you can have it basically
2:36:01 audit you I'm a millennial I remember
2:36:04 the dial up days myself but I wasn't an
2:36:06 adult during it at all yeah
2:36:08 yeah it was wild times I mean that
2:36:10 transition from when you had to die when
2:36:12 you when your computer had to call up
2:36:14 other computers and then transitioning
2:36:16 to the internet just kind of being there
2:36:18 that was a really interesting transition
2:36:19 it didn't seem like a big deal at the
2:36:21 time but now when I look back on it the
2:36:24 the fact that you had to think about oh
2:36:27 I want to get online I have to sit at my
2:36:30 computer and I have to figure out which
2:36:32 other computer do I want to call
2:36:35 and then you would call it and then
2:36:36 eventually you got to okay I'm just
2:36:38 going to have it call the internet right
2:36:40 the world wide web and then I'm going to
2:36:42 go to this all these things and then it
2:36:43 just got to now it's just there and then
2:36:46 Wi-Fi came and things like that and now
2:36:48 we have you know essentially super
2:36:50 computers in our pockets
2:36:52 so it's you you very quickly kind of
2:36:54 forget the you know the the big you know
2:36:58 errors of the of the technological
2:37:00 change
2:37:03 um please wear a solid dark colored
2:37:05 shirt I know I'm gonna do I'm gonna go
2:37:06 get a t-shirt
2:37:08 I'll be back
2:38:04 all right oh my t-shirts had white [ __ ]
2:38:07 on them too so I had to turn it inside
2:38:08 out but that's better
2:38:11 all right
2:38:13 um oh there's some new people here hi my
2:38:14 name is Kyle Shannon this is the AI
2:38:16 learning lab I'm learning about AI I'm
2:38:18 taking whatever I learn and trying to
2:38:21 pass that along to anyone who gives a
2:38:23 [ __ ]
2:38:24 um these sites here if you haven't used
2:38:26 chat GPT that top site there is the
2:38:28 official chat CPT website go there now
2:38:34 um bing.com is also chat gbt if you
2:38:36 click on the chat button that's gpt4
2:38:38 that you can use for free and then
2:38:40 poe.com has a bunch of different things
2:38:41 you can play with
2:38:43 and then those other two prompts.chat
2:38:45 will teach you about prompting and
2:38:47 futurepedia.io is a big directory of
2:38:49 lots of AI tools I'm not affiliated with
2:38:51 any of that [ __ ] it's just there if you
2:38:54 haven't played with this stuff that's
2:38:55 why it's there
2:38:57 um I have updated my shirt to dark
2:38:59 colored so you can see the chat I'm way
2:39:01 behind on my chats there's 68 messages
2:39:03 down below where I am so if you're
2:39:06 typing [ __ ] to me right now and I'm not
2:39:08 responding to it it's not because I'm a
2:39:10 robot it's because I'm behind and I
2:39:12 don't know how to use tick tock because
2:39:13 I'm 58 years old and what are you gonna
2:39:15 do sue me it's free
2:39:17 you get the energy thank you for the
2:39:18 rose you get the entertainment of me
2:39:21 sitting here slightly sweaty with
2:39:24 apparently a bad toupee talking with
2:39:27 words
2:39:30 that were composed by chat gbt probably
2:39:33 I made a hypertext transfer protocol
2:39:35 project HTTP before it was called a
2:39:37 website yeah exactly man those early
2:39:40 days I'll tell you we so so I was I was
2:39:44 early enough in in the world wide web
2:39:45 that the center tag didn't apply to
2:39:48 images so if you had images they were
2:39:49 always left Justified
2:39:51 so I remember we invented a series of
2:39:54 pixel High invisible images there was
2:39:57 like a 10 pixel wide one and a five and
2:39:59 a three and a one we did like prime
2:40:01 numbers so that you could scooch images
2:40:03 to the center of a web page that was
2:40:06 that was Innovation back back in my day
2:40:08 you know there's like 12 HTML tags
2:40:10 including a blink tag
2:40:13 it was the the uh the marker of someone
2:40:15 who's bad at design was they used the
2:40:17 blink tag
2:40:21 all right
2:40:24 let's see it also contains all your
2:40:26 genetic info
2:40:28 thanks to okay
2:40:30 don't really care about conspiracy
2:40:32 theories but appreciate the
2:40:34 the effort I think this will be similar
2:40:37 but as he said more profound yeah I
2:40:39 think this is I the parallels to the
2:40:42 early days of the web are are very clear
2:40:45 to me but the holy [ __ ] what this stuff
2:40:48 does and the speed with which it's
2:40:49 coming is just crazy
2:40:52 um
2:40:53 someone called me adorable I was also
2:40:55 called adorably optimistic so some of
2:40:58 the people talking about the uh the dark
2:41:00 future probably think I'm adorably
2:41:03 optimistic too
2:41:05 some would call it naive he just thinks
2:41:08 everything's going to be Shangri-La
2:41:12 I don't
2:41:14 I just I don't know
2:41:17 I'm I'm more a tech Optimist I suppose I
2:41:20 I really just
2:41:21 for me the thing that excites me is what
2:41:24 this stuff makes possible like when you
2:41:26 actually start to get your head around
2:41:28 what it can do
2:41:30 and what so here so here's the thing
2:41:33 Microsoft calls their product their
2:41:36 their chat GPT product co-pilot that
2:41:39 feels like
2:41:40 great branding to me because that's the
2:41:43 experience you're the pilot this tool is
2:41:45 your co-pilot
2:41:47 but it happens to be an incredibly
2:41:49 capable co-pilot like way more capable
2:41:53 than you likely are because
2:41:55 it is
2:41:57 like I I can't imagine that there's a
2:42:00 human being on Earth
2:42:02 that begins to scratch the surface of
2:42:04 the capability of of what's sitting
2:42:07 inside that little
2:42:09 innocent looking search box
2:42:12 or chat box whatever you call it
2:42:16 can AI tell you if the sounds are
2:42:18 fireworks or gunshots actually at some
2:42:21 point sure yeah you could train it on
2:42:23 that
2:42:25 um nuggets fries Coke Sports
2:42:28 um I hope Chachi GPT becomes available
2:42:31 for America Online soon that's good
2:42:33 that's good callback post seems like the
2:42:36 Fisher-Price version of chat gbt that's
2:42:38 a great way to describe it why would I
2:42:40 use it instead of open AI the only
2:42:42 reason you would use Poe instead of open
2:42:44 AI is
2:42:46 for hold
2:42:50 [Music]
2:42:53 one hundred thousand
2:42:55 Claude instant 100 000 that means a
2:42:59 hundred thousand tokens so if you pay
2:43:01 for poe.com
2:43:04 this is a model that you can put a 75
2:43:07 000 Word document into the chat
2:43:09 into the chat block and start talking to
2:43:12 it I just I just did an example with the
2:43:14 Godfather
2:43:15 I put in the entire Godfather screenplay
2:43:19 and then had it create me a table of all
2:43:21 the characters and what happened to them
2:43:23 in the movie and they're right it's
2:43:25 right
2:43:26 so
2:43:28 that's why but yeah it is it's that's a
2:43:30 great description of it it's the Fisher
2:43:32 Price of of GPT models it's pretty good
2:43:36 all right I like my uh I like my crew
2:43:40 here I like you peeps you're good you're
2:43:42 Saucy you're spicy you're keeping me
2:43:44 honest this is good your trolls your
2:43:47 trolls are a little less creative
2:43:48 tonight I've I've had better nights of
2:43:50 trolling
2:43:52 the uh
2:43:53 three times in a night saying you kind
2:43:56 of look like Rosie O'Donnell
2:43:58 that was those were good trolls that's
2:44:00 sort of a compliment right it's no it's
2:44:03 no
2:44:04 yeah
2:44:06 I was scrolling Tick Tock I thought you
2:44:08 were Alec Baldwin but got fat
2:44:10 um
2:44:11 is it a compliment is it a troll like I
2:44:13 like the I like the ambiguous trolls
2:44:17 what's happening and it's happening
2:44:19 twice in your life it's crazy it's it's
2:44:20 absolutely insane I remember so my my
2:44:24 first awareness of these kind of events
2:44:27 happening in my life was the first day
2:44:29 of seventh grade in 1977.
2:44:33 I walked into in fact I did a I did a
2:44:35 slide in a presentation for this I'll
2:44:38 show it to you because I don't know
2:44:40 because we're here
2:44:42 um let me get to the slide
2:44:49 so
2:44:51 on my first day of seventh grade in 1977
2:44:56 I walked into a classroom that looked
2:44:58 like that
2:44:59 and sitting on every desk was a Radio
2:45:01 Shack TRS-80
2:45:03 so that computer right there hang on I'm
2:45:06 trying to make this window a little
2:45:07 smaller
2:45:08 oh I can't because it's running and I
2:45:12 remember
2:45:14 I was in seventh grade I remember being
2:45:16 acutely aware that I was just too young
2:45:18 to be one of these two guys and I was
2:45:21 just too young
2:45:22 to be
2:45:25 on that team
2:45:27 and I remember being really sad because
2:45:30 what I knew was
2:45:32 like I missed it by like 10 years right
2:45:35 or not even yeah probably 10 years who
2:45:38 knows how old Bill Gates was there he's
2:45:39 probably in his teens so he was probably
2:45:41 six or seven years older than I was
2:45:44 and
2:45:46 I just remember like like the per like
2:45:48 as I learned about personal computers
2:45:50 and what they were I'm like
2:45:53 this is like the printing press like
2:45:55 like
2:45:57 I'm not gonna get to experience this
2:45:59 again and I missed it I just missed it
2:46:01 and so when I had the Epiphany about the
2:46:03 World Wide Web
2:46:04 it was it was like this
2:46:08 holy [ __ ]
2:46:10 this is one of those
2:46:12 I get to do it this time and so I
2:46:15 [ __ ] don't do that [ __ ]
2:46:18 and I feel the exact same way about AI I
2:46:20 haven't it's been what's it been 30
2:46:21 years 94
2:46:24 94 204 14 wait 204 14. yeah
2:46:30 yeah we're coming up on 30 years since I
2:46:33 had that Epiphany so the fact that like
2:46:36 within a 30 year period we're having you
2:46:38 know another one of these is a big deal
2:46:40 but yeah that's why I'm so excited about
2:46:42 this stuff
2:46:43 we had a Commodore 64 in the 80s oh yeah
2:46:48 nuggets are NBA champions
2:46:51 jokic
2:46:54 how can a dude that is so clumsy looking
2:46:56 and so like like like his arms his his
2:47:00 arms when he shoots basketball are like
2:47:02 you know fingers in in mid-journey and
2:47:05 stable diffusion we had these hands with
2:47:07 like 17 fingers on them it's like
2:47:09 whatever he does physically is like that
2:47:11 should not a basketball should not fall
2:47:14 in the net based on what he does with
2:47:16 his body but he [ __ ] he's [ __ ]
2:47:18 amazing
2:47:20 um incredible how the web empowered
2:47:22 creators totally
2:47:24 um think what what AI will do that's
2:47:26 that's exactly right I mean that's
2:47:29 I I I truly feel like we're entering a
2:47:32 great Renaissance in fact I I have a
2:47:34 slide in this in this deck
2:47:38 um
2:47:42 this mid-journey rocks
2:47:48 um
2:47:49 and I also noticed that
2:47:52 um there's there's always been like I
2:47:54 always wondered why there was an i in
2:47:56 Renaissance like it never made sense to
2:47:58 me but it's because AI is in Renaissance
2:48:01 so I think I think this is the age we're
2:48:04 entering is the great Renaissance
2:48:06 um oh by the way if you're new here it
2:48:08 looks like some new people joined my
2:48:09 name is Kyle Shannon this is the AI
2:48:10 learning lab
2:48:12 um I'm just sort of waxing poetic about
2:48:14 generative AI answering any questions
2:48:15 that I can
2:48:17 um getting trolled sometimes
2:48:19 appropriately sometimes not but that's
2:48:20 the nature of trolling and
2:48:23 um just generally being a smart ass and
2:48:26 doing what I can to um demystify this
2:48:29 stuff absolutely remarkable yes I think
2:48:31 it is
2:48:32 oh what I don't know what you're asking
2:48:34 absolutely remarkable that the Denver
2:48:36 Nuggets won the NBA Championship yeah
2:48:38 that's pretty [ __ ] remarkable
2:48:41 um I didn't have that on my dance card
2:48:43 for this year I want to train AI on my
2:48:45 own data how do I do that um you can
2:48:47 wait for six months
2:48:49 because
2:48:51 um
2:48:52 um Microsoft and Google are both going
2:48:54 to be rolling all this AI stuff into
2:48:56 their office suite of products
2:48:58 so you're just going to be able to point
2:49:00 your [ __ ] Point AI at your own [ __ ] and
2:49:03 do it
2:49:04 if you're geeky and you want to do
2:49:06 things ahead of that and you kind of
2:49:09 want to learn how this stuff actually
2:49:10 works check out a project called Lang
2:49:13 chain l-a-n-g chain
2:49:16 and it's got like four components to it
2:49:18 there's you you embed your text and then
2:49:21 you get it in a vector database and then
2:49:23 you hook it to Google so you connect
2:49:25 together a bunch of tools but basically
2:49:27 what you're doing is you're you're
2:49:28 creating your own database where you can
2:49:30 store embedded data embedded information
2:49:34 that's your information and then you can
2:49:36 search on it and you can do all sorts of
2:49:39 [ __ ] with that but it requires you to be
2:49:41 a bit geeky now here's the good news you
2:49:44 got chat GP day that can help you learn
2:49:47 how to do that [ __ ]
2:49:50 um fun fun
2:49:52 little micro hack you can take the
2:49:55 readme file from a GitHub repository
2:49:57 paste it into chat GPT and have it
2:50:00 create a step-by-step
2:50:03 um installation instruction for your
2:50:05 specific machine
2:50:06 and it will do that and if you tell it
2:50:08 not to give you the next step until you
2:50:11 say the word okay it'll literally do it
2:50:13 step by step and if you get confused on
2:50:16 a step you can tell it and it will you
2:50:18 know teach you how to do that step and
2:50:20 then you can say okay and it'll move on
2:50:21 to the next step that's kind of fun
2:50:24 hey dude not sure if you remember me and
2:50:27 my animations I got hired to make music
2:50:29 video yesterday cool congratulations oh
2:50:33 that's awesome
2:50:34 wow yeah that's so cool and I was
2:50:37 talking to yesterday
2:50:41 um I was talking about this stuff and
2:50:43 just like the profound impact it might
2:50:45 have on people and and one of the guys
2:50:46 in the chat has a degenerative nerve
2:50:49 disease he was an illustrator and can no
2:50:52 longer illustrate he can still sketch
2:50:53 but he can't he doesn't have fine motor
2:50:55 skills to to finish it and he's using
2:50:57 stable diffusion to do that and then one
2:50:59 of the other people had some other
2:51:01 condition and couldn't write anymore and
2:51:03 now he's able to write again and it's
2:51:05 like yes like these are like these are
2:51:08 Humanity amplifiers these are Talent
2:51:11 amplifiers right you get to still choose
2:51:15 what you want to do in the world and you
2:51:17 just have a really capable assistant to
2:51:19 help you do it that's what's going on
2:51:21 with this stuff and that's why it's so
2:51:22 profound
2:51:24 um and and and listen there's a dark
2:51:27 side to that level of power in that
2:51:30 every single job
2:51:33 that involves
2:51:35 data in any way or words or creativity
2:51:38 are all going to be upended like upended
2:51:42 completely redefined
2:51:45 and and there's not a it's it's not like
2:51:48 you can opt out of this I mean that
2:51:50 that's the problem is that if you opt
2:51:52 out you're like opting out of the
2:51:54 competitive you know job market
2:51:57 for for a lot of these
2:51:59 um kinds of jobs
2:52:00 the only antidote to it that I know is
2:52:03 get curious about it and start learning
2:52:05 it
2:52:06 figure out how to disrupt yourself
2:52:07 figure out how to use these tools to do
2:52:10 what you do
2:52:11 and it will be terrifying
2:52:15 and existential and you'll probably go
2:52:18 through a morning period like I went it
2:52:21 took me four months when I started
2:52:23 learning about this AI stuff I've got a
2:52:25 video technology company of my I got a
2:52:27 company called storyvine I'm the CEO of
2:52:28 it
2:52:29 it's an 11 year old company
2:52:32 and I started looking at this AI [ __ ]
2:52:34 and I realized there's a bunch of stuff
2:52:36 that we do right now that the value of
2:52:38 it a year from now is going to be zero
2:52:40 like a big goose egg
2:52:43 that's [ __ ] depressing
2:52:46 right I've been working at this for a
2:52:47 [ __ ] decade
2:52:49 and I'm like the automation that our
2:52:51 current technology
2:52:52 does a year from now it's gonna be like
2:52:55 yawn whatever everyone's automated now
2:52:57 okay what do I do with that
2:53:00 it took me about four months to to not
2:53:04 have ai be this separate
2:53:06 thing
2:53:07 right I had my company here I had this
2:53:09 AI [ __ ] here and I like it almost felt
2:53:11 like I was like like cheating on the
2:53:14 company
2:53:15 and then slowly as I started to get my
2:53:18 head more around it I'm like okay
2:53:21 here's some of the core value of my
2:53:23 company that remains the automation
2:53:25 stuff that goes away but authenticity
2:53:27 that's that's good narrative that's good
2:53:30 those things remain valuable in this new
2:53:32 future
2:53:33 and then I can then I can start to say
2:53:36 okay how do I bring this AI stuff in in
2:53:38 a way that amplifies those things that
2:53:40 are going to be good but that took me
2:53:42 four months and it's it's intense so I'm
2:53:46 not saying it's not intense but I am
2:53:47 saying if if you're not getting curious
2:53:50 about this stuff you're doing yourself a
2:53:52 real disservice because it's coming
2:53:53 faster than you think
2:53:57 um it was basically advanced
2:54:00 ITV
2:54:02 was basically Advanced networks I think
2:54:05 I know what you're talking about there
2:54:06 does chat CPT remember results no and
2:54:08 apply to better answers well
2:54:10 sort of open AI is using all of our use
2:54:14 of chat GPT to improve to fine-tune the
2:54:17 models so as a general rule it's getting
2:54:21 better for you personally it does not
2:54:23 you've got a chat history but but it
2:54:26 does not have a memory of what it's
2:54:28 created and you can't reference one chat
2:54:30 to the other they're distinct little
2:54:31 conversations that will change but just
2:54:35 we're early
2:54:37 we're six months into this and chat EBT
2:54:39 was written in two weeks so it's a it's
2:54:41 a very thin piece of software let's put
2:54:43 it that way
2:54:45 um
2:54:47 thank you I have scores and scores of
2:54:49 data that I'd love to have chat GPT do
2:54:51 some work on check out Lang chain
2:54:54 and I mean the thing about Lang chain
2:54:56 even though you need to get geeky with
2:54:57 it it's like it's like 80 lines of of
2:55:00 python
2:55:02 it's like some ridiculously small amount
2:55:04 of python code and you can have your own
2:55:08 sort of version of chat GPT that you can
2:55:10 dump your own data into and then start
2:55:12 querying against it
2:55:15 um
2:55:15 and do a whole bunch of other [ __ ]
2:55:17 go to the um the Lang chain subreddit
2:55:21 although subreddits start or reddit's
2:55:23 dark today so I don't know what's going
2:55:25 on there but go to the Lang chain
2:55:27 subreddit
2:55:29 um you know start learning about GitHub
2:55:31 and hugging face if you're not on those
2:55:33 greedy entrepreneurs don't care if they
2:55:35 burn the world down to make a dollar
2:55:38 um
2:55:40 yeah greedy entrepreneurs I I don't
2:55:42 think entrepreneurs in general I I mean
2:55:47 a couple of things start to happen in in
2:55:49 my opinion is
2:55:51 so I've heard some some stats that tech
2:55:54 companies that are embracing chat GPT
2:55:57 are seeing 30 to 40 percent productivity
2:55:59 increases
2:56:01 so
2:56:02 when you have a technology that that
2:56:04 generates a 30 to 40 percent
2:56:06 productivity increase you've got two
2:56:08 options essentially as an entrepreneur
2:56:10 if you're a private Equity Firm you're
2:56:13 like [ __ ] it cut 40 of the staff right
2:56:16 and I think that's the fear
2:56:19 if you're an entrepreneur
2:56:23 and you're about creating value
2:56:26 and you see a 30 to 40 percent
2:56:27 productivity increase you could say huh
2:56:30 we could either deliver 30 to 40 40
2:56:32 faster and be more competitive that way
2:56:35 Thirty to forty percent higher quality
2:56:39 um
2:56:40 you know or or
2:56:43 um
2:56:44 we could take on more clients right we
2:56:46 could scale
2:56:47 so
2:56:48 I don't I think as a blanket term it's
2:56:51 fine to [ __ ] on entrepreneurs and say
2:56:52 they're all going to [ __ ] rip us off
2:56:54 but a lot won't and and
2:56:58 a lot are going to take those
2:56:59 productivity increases and they're going
2:57:01 to start innovating they're going to
2:57:02 start using these tools to innovate new
2:57:04 ways of working so I think what you're
2:57:06 going to see is you're going to see
2:57:07 bigger old established companies that
2:57:09 are like we're fine we're the market
2:57:12 leader we've got three or four years to
2:57:14 deal with all these punks with their
2:57:16 artificial intelligence and
2:57:21 a 10 person company is going to come
2:57:24 knock them off their pedestal right I
2:57:26 think we're going to see a lot of that
2:57:27 [ __ ] so so the companies that jettison
2:57:31 30 of their Workforce a lot of those 30
2:57:34 percent jettisoned people also have
2:57:36 access to these tools
2:57:37 and they and they might know how that
2:57:39 company works better than the CEO does
2:57:41 and they're going to come back at them
2:57:43 two years from now and go knock knock
2:57:45 get the [ __ ] out of here
2:57:48 um so anyway so so I think it's going to
2:57:50 be good but it is going to be disruptive
2:57:52 and yeah there's going to be some shitty
2:57:53 entrepreneurs out there that do shitty
2:57:55 things that's human nature
2:57:58 what I did was send documentation to
2:58:00 gpt4 and had it format the data
2:58:02 interesting yeah I'm a programmer work
2:58:04 with chat EBT all the time and animator
2:58:07 with stable diffusion oh that's super
2:58:09 cool Harrow that's cool I was asking
2:58:11 about remembering wondering if it was
2:58:14 getting smarter thanks for answering
2:58:15 it's getting smarter it's it's just it's
2:58:17 not getting smarter for you as an
2:58:19 individual it's getting smarter as a
2:58:21 larger capability
2:58:22 braver Builder is no code web Builder
2:58:25 that's cool I haven't heard of that
2:58:29 it's funny I just got a notice from
2:58:30 Reddit about Lang chain it was listening
2:58:33 they're all listening to everything we
2:58:35 say
2:58:36 what are we gonna do
2:58:39 have you used zapier with chat CPT for
2:58:41 any automations yeah I've built a bunch
2:58:42 of
2:58:43 um content generation tools this is my
2:58:46 window I can't find my window
2:58:49 you know watching old people with
2:58:51 Technologies rather entertaining okay
2:58:56 um if I go to content Evolution so
2:59:00 I'm part of this group called content
2:59:02 Evolution which is basically a
2:59:04 Federation of uh marketing
2:59:06 marketing companies and and I I just got
2:59:09 promoted to the chief generative officer
2:59:12 of this organization but largely because
2:59:14 we built this tool so what you see here
2:59:17 is this thing called The Challenge
2:59:18 engine and what this is is a Google form
2:59:21 so I'm basically using a Google form as
2:59:23 input and it asks you you know what what
2:59:26 business category are you in and what's
2:59:28 your challenge and then you answer a few
2:59:30 questions about the challenge
2:59:32 and then what it does
2:59:38 e hold please what it does let me go
2:59:42 back to Gmail
2:59:51 is within about 30 seconds it gives you
2:59:55 kind of a Consulting response to your
2:59:57 problem
2:59:58 so so it basically says us the challenge
3:00:01 engine about you know I would like to
3:00:03 file for a patent for for an invention
3:00:05 of mine
3:00:06 so we kind of recap what the challenge
3:00:08 is and and some of the questions that
3:00:10 you answered and then in these different
3:00:13 categories strategic considerations
3:00:14 customers perspective in each of these
3:00:17 categories we give you three questions
3:00:19 that you could be asking yourself so
3:00:21 strategic what's the competitive
3:00:22 landscape like in the industry
3:00:25 um how will filling a patent for your
3:00:27 invention affect the competitive
3:00:29 Advantage what's the long-term vision
3:00:31 for your invention customer brand
3:00:33 organization and it it not only
3:00:35 generates this document but it generates
3:00:37 the HTML
3:00:39 so
3:00:41 so how we do that if you don't know what
3:00:42 zapier is it's you can kind of think of
3:00:45 it like Lego blocks for for connecting
3:00:49 different computer services together
3:00:52 so let me log in
3:01:04 all right so so for the challenge
3:01:05 engines here are the little Lego blocks
3:01:07 so the way these start is they start
3:01:09 with the trigger so something happens in
3:01:11 the world someone posts to LinkedIn or
3:01:13 they post a tweet or they post on your
3:01:15 slack Channel or in this case they fill
3:01:17 out that form
3:01:19 so when they fill out that form I then
3:01:21 take the the data the input from that
3:01:25 form and I roll that into a prompt that
3:01:28 I wrote for open AI so there's this
3:01:31 prompt that's pretty long that kind of
3:01:33 sets up you know this this bot is
3:01:36 supposed to act like a consultant and we
3:01:38 describe what the what content evolution
3:01:40 is and what they do and then these
3:01:43 little Fields here are the inputs from
3:01:46 that form so I say now you're going to
3:01:48 provide Sound Advice to a professional
3:01:50 device to a professional using these
3:01:52 variables so I Define the variables pop
3:01:54 the little form pieces in there and then
3:01:57 and then basically write a prompt that
3:01:59 that says you know go make a set of
3:02:02 three questions each in these different
3:02:04 categories create an introductory
3:02:06 paragraph and a conclusion paragraph
3:02:08 make sure that you you know put a call
3:02:10 to action for the organization there so
3:02:12 that that Lego block goes to open AI
3:02:16 generates the content
3:02:18 then we take the output of that send it
3:02:20 back to open Ai and say now go format
3:02:24 the HTML here's here's the rules for the
3:02:26 HTML and then you know here's that input
3:02:29 and then we take that HTML formatted
3:02:32 content dump that into a Gmail Lego
3:02:35 block
3:02:37 and that sends the email out and then
3:02:39 the final Lego block is we write all of
3:02:41 that as a row in a Google spreadsheet
3:02:45 so that is zapier and it's I mean that's
3:02:49 what's nice about building automations
3:02:51 like that is you really need to you
3:02:55 really need to get your prompt crafting
3:02:58 right or you get shitty content and you
3:03:00 don't want to automate shittiness
3:03:03 um so so it's a really good exercise to
3:03:06 learn those automations it's also super
3:03:08 powerful like I was showing a guy I was
3:03:10 on a phone with a guy
3:03:12 and he was like yeah I've used chat GPT
3:03:14 he was like yeah I'm good I'm I'm smart
3:03:18 I've used chat GPT
3:03:20 and I said you know are you aware could
3:03:23 do this that or the other he's like ah
3:03:24 yeah yeah yeah I said well let me show
3:03:26 you something I built and I showed him
3:03:27 this thing that I call it the social
3:03:29 media maker machine and you answer a few
3:03:31 questions about a topic and then it
3:03:33 generates like five pieces of social
3:03:35 media content they're pretty good
3:03:37 so I'm on the phone with him and I go to
3:03:39 this Google form and I ask him some
3:03:40 questions I said give me a topic and
3:03:42 answer a few questions about it and then
3:03:43 I hit you know submit and we're talking
3:03:46 and like a minute later he looks at his
3:03:49 email he goes oh this thing came in and
3:03:51 he looks at the email
3:03:53 and his jaw literally dropped open
3:03:57 and he said I he goes what is this I'm
3:04:01 like well that's the social media maker
3:04:02 machine Let's that's what chat GPT can
3:04:04 do and he goes
3:04:06 I had no idea and he goes I gotta go and
3:04:10 he hung up on me and it's like it's like
3:04:13 it's intense like like you think like
3:04:16 before you use this stuff you're like oh
3:04:18 it's kind of like Google but a little
3:04:20 better maybe I don't know people seem
3:04:21 excited about it but when you actually
3:04:23 use it and see what it's capable of like
3:04:26 it just I I don't know that moment for
3:04:29 me that's a moment of Joy when people
3:04:31 like wait it does that
3:04:33 holy [ __ ] what else can it do like
3:04:35 that's where you want to get get to that
3:04:37 moment for yourself
3:04:40 Wix is the is the no code web Builder
3:04:43 that I was thinking of also GoDaddy
3:04:45 Squarespace all of those yeah if you
3:04:47 want to build websites you don't need AI
3:04:49 for that right now although go to
3:04:51 futurepedia.io because I know that
3:04:53 there's a bunch of tools out there there
3:04:55 now that are like just type in the
3:04:58 iPhone app that you want or the website
3:05:00 that you want and just out of income
3:05:04 um I believe Tech Advance will just
3:05:07 become better and faster yeah we are
3:05:08 deaf it's so we're definitely already
3:05:11 hitting an acceleration where the the
3:05:13 Innovations are
3:05:17 there's there's been been this really
3:05:18 interesting thing because chat GPT was
3:05:21 adopted so quickly and so massively
3:05:25 the entire developer Community is like
3:05:28 all in on it on generative AI and so
3:05:31 what's happening is every time a white
3:05:32 paper is released or some new technology
3:05:34 is released especially in the open
3:05:36 source Community they're just [ __ ]
3:05:39 pouncing on it and like there's GitHub
3:05:41 repos just like popping up it's like
3:05:43 popcorn where GitHub and hugging face
3:05:46 there's just new models and new tools
3:05:48 and new interfaces
3:05:50 and they're starting to build on top of
3:05:52 one another so like you'll see a paper
3:05:54 come out and like three days later that
3:05:57 paper paper will be rolled together with
3:05:59 two other papers that came a week before
3:06:00 and that'll be a new paper and then that
3:06:03 paper gets dumped into some future like
3:06:06 it's just insane how how quickly it's
3:06:08 going
3:06:10 um
3:06:13 it's good you're you're Consulting with
3:06:16 one another about website builders in
3:06:18 here I appreciate that because my brain
3:06:20 shut down when I tried to remember some
3:06:22 of them so thank you for that this will
3:06:24 be similar to internet early days except
3:06:26 advancements will be faster totally
3:06:28 agree with that much faster and much
3:06:30 more profound again the core tool the
3:06:32 core output of what the World Wide Web
3:06:34 was and the core output of what this is
3:06:36 are like leagues apart you know boring
3:06:41 page with blue underlined words versus
3:06:45 world-class illustrations writing tables
3:06:48 and coding
3:06:50 you mean you can have gpt4 write queries
3:06:53 for your database yeah in fact it's it's
3:06:56 it's way more intense than right queries
3:06:58 for your database you can have it write
3:07:00 queries for your database but there's a
3:07:02 thing coming out called code interpreter
3:07:04 which is just another mode of chat GPT
3:07:08 where you can upload a data set you can
3:07:10 upload a spreadsheet and it will read
3:07:12 the spreadsheet and then you can just
3:07:14 start querying on it you can say what
3:07:16 are the interesting Trends in this data
3:07:18 and it will it'll create a whole Report
3:07:19 with graphs and [ __ ] like that you could
3:07:21 say are there any anomalies in the data
3:07:23 it will instantly identify the anomalies
3:07:25 in the data
3:07:27 um there was a guy one of the examples I
3:07:29 saw a guy uploaded the
3:07:33 um
3:07:33 Lighthouse data for the United States
3:07:36 like where all the lighthouses on the
3:07:38 coasts and then he said to code
3:07:41 interpreter
3:07:42 create a dark map of the United States
3:07:46 with each of the locations of the
3:07:48 lighthouses as a light blinking Dot and
3:07:51 it wrote The Python code to generate a
3:07:53 map of the United States and then
3:07:55 plotted all of the locations of the
3:07:57 lighthouses on the United States
3:07:59 blinking
3:08:01 like this [ __ ] is the [ __ ] that's coming
3:08:05 it's like
3:08:07 start playing if you're new here my name
3:08:09 is Kyle Shannon this is the AI learning
3:08:11 lab I'm a bit nuts and these things are
3:08:14 where you should go right now if you
3:08:15 haven't played with chat GPT
3:08:17 go there go there now
3:08:21 I don't get paid for any of this I don't
3:08:23 get any commissions though if you work
3:08:25 for open AI you should give me
3:08:26 commissions I'm like [ __ ] on here
3:08:28 every night talking about how awesome
3:08:30 you are and Sam Altman's a genius and
3:08:33 you know
3:08:35 you're out here doing God's work well
3:08:37 I'm out here doing work
3:08:40 um
3:08:41 dude you have great hair thank you I
3:08:43 appreciate that it is uh it sometimes
3:08:46 has a mind of its own it's behaving the
3:08:47 glue is holding tonight so I appreciate
3:08:50 that
3:08:51 why wouldn't AI destroy us
3:08:54 um same reason it you know the same
3:08:57 reason it will transform us for the good
3:09:02 um
3:09:03 here's the deal it is not going to
3:09:06 destroy so so there's a there's a
3:09:08 marketing guy named Paul rateser
3:09:10 r-o-e-t-z-e-r he's got an organization
3:09:13 called the marketing AI Institute you
3:09:16 should go listen to his podcast he's
3:09:17 been he's been at this for like five
3:09:19 years on the on the marketing side
3:09:22 really smart guy
3:09:24 he's got this saying
3:09:26 AI will not replace you but someone
3:09:28 using AI will
3:09:30 and I feel the same way about AI is not
3:09:32 going to destroy us someone using AI
3:09:34 will but the the
3:09:38 optimistic side of that coin is
3:09:41 um the good guys also have access to
3:09:43 these tools right so it's not just Bad
3:09:45 actors that have access to these tools
3:09:46 we do too
3:09:48 part of the reason this channel exists
3:09:50 is
3:09:51 I want people to get curious about this
3:09:53 and get educated so they're not
3:09:55 operating out of fear or ignorance just
3:09:57 like like learn what what it makes
3:10:00 possible and start to understand it
3:10:02 and then try to do good with it because
3:10:04 I feel that any small thing that we do
3:10:07 now that's positive is going to have a
3:10:09 disproportionately a disproportionate
3:10:11 impact you know
3:10:13 to the good and I think we can actually
3:10:16 make a difference but we actually have
3:10:17 to participate
3:10:20 um you're great I appreciate you doing
3:10:22 this thank you very much I appreciate
3:10:24 that
3:10:25 uh uh uh Show an example of Bing slash
3:10:29 create I will do that so if you want to
3:10:32 generate images you have
3:10:35 a bunch of options you've got
3:10:38 the best one in my opinion is
3:10:39 mid-journey
3:10:41 the second best one is stable diffusion
3:10:44 and stable diffusion has a lot more
3:10:46 developers working on it because it's
3:10:48 open source
3:10:50 and then you have Adobe Firefly which if
3:10:53 you're in business is good because it's
3:10:56 safe for commercial oops did I spell
3:10:59 that wrong Well I obviously spelled it
3:11:00 wrong
3:11:01 hold please old man can't type
3:11:06 warning we've got an old man trying to
3:11:09 type on the computer how do I install
3:11:11 the internet on this thing
3:11:14 so this is the the fourth one so I've
3:11:18 talked mid-journey is the best stable
3:11:19 diffusion is the most flexible and
3:11:21 probably powerful but I don't think the
3:11:23 quality is quite as good
3:11:26 um
3:11:26 Adobe Firefly and generative fill in
3:11:30 Photoshop are really good
3:11:33 um and they're safe for commercial but
3:11:35 they're but they're kind of third in
3:11:37 quality and then fourth in quality is
3:11:39 this which is Dali which is from open AI
3:11:42 the people that make Chach EBT but if
3:11:44 you go to bing.com create it's an image
3:11:47 generation tool so if we go let me try
3:11:49 to do something different we'll do
3:11:52 um Rocky
3:11:54 mountains at
3:11:57 sunset with a
3:12:00 dramatic
3:12:03 Thunder
3:12:05 head
3:12:07 cloud
3:12:10 over the mountains
3:12:15 all right let's see what it does
3:12:19 so the nice thing about
3:12:21 Dolly and and this is that it's free
3:12:25 um and with dolly in particular those
3:12:28 are not bad actually those are pretty
3:12:30 good
3:12:31 um so you know big nice Thunder Boomer
3:12:34 cloud and
3:12:37 I mean that's pretty solid
3:12:41 that does not look dissimilar to what I
3:12:44 see over the Rocky Mountains here
3:12:47 on a decently regular basis I'm a
3:12:49 photographer so I take a lot of
3:12:51 a lot of cloud pitches if you go to Kyle
3:12:53 shannon.com you can see some of my
3:12:55 photography
3:12:58 I actually have a lot of stuff that
3:12:59 looks like this
3:13:02 so yeah so that's bing.com create
3:13:07 does that welcome it looks like we got
3:13:09 some new people my name is Kyle Shannon
3:13:10 this is the AI learning lab I'm about
3:13:12 three hours into this live so I have no
3:13:15 moisture left in my mouth
3:13:17 um my I'm I'm moist
3:13:20 sorry I am I changed my shirt so you can
3:13:24 read the text
3:13:28 I'm an entrepreneur I'm the co-founder
3:13:30 and CEO of a company called storyvine
3:13:32 which is an automated video storytelling
3:13:34 platform
3:13:35 and I'm really here just to do anything
3:13:37 I can to demystify generative AI so if
3:13:40 you have questions about AI pop them
3:13:42 into the questions below I am painfully
3:13:45 behind on the questions so if it seems
3:13:46 like I'm ignoring you it's just because
3:13:48 I'm behind but I love you you're here oh
3:13:52 and go ahead tap the screen I know a
3:13:53 bunch of your tap on the screen now but
3:13:55 apparently if you tap the screen it'll
3:13:56 tell other people that we're here we can
3:13:58 invite them to our little party
3:14:00 this is also a Nuggets won the NBA
3:14:02 Finals celebration I'm in Denver
3:14:04 Colorado I don't really give
3:14:06 too much of a [ __ ] about basketball but
3:14:08 I'm a fair weather fan I like it when
3:14:11 they win championships that's pretty
3:14:12 [ __ ] cool
3:14:14 all right I'm 62 now and right with you
3:14:16 thank you yep us us gen xers got to
3:14:19 stick together
3:14:21 bitter you betcha
3:14:23 you want to be able to ask a question oh
3:14:25 darn it tick tock tick tock just jump me
3:14:28 to the bottom again because that's what
3:14:29 it does it does it doesn't like me it
3:14:32 doesn't like the way I do my my uh my
3:14:34 lives and how I answer questions if I'm
3:14:36 scrolling when someone puts in a new
3:14:38 comment it jumps me to the bottom
3:14:40 and I lose my place and then I get
3:14:42 cranky I like have to OCD my way back to
3:14:45 the exact question I was at whatever
3:14:49 talk to my psychologist
3:14:51 I don't have one
3:14:54 someone said they're freaking out so so
3:14:56 here's the thing about this once you
3:14:58 start playing with these things
3:15:00 I've been like this for six months
3:15:04 I am simultaneously as excited as I've
3:15:07 ever been about something and as
3:15:09 terrified as I've ever been about
3:15:11 something
3:15:12 and like both of those things are true
3:15:15 at the same time like that's that's
3:15:17 happening
3:15:18 pretty much on a daily basis
3:15:23 all right I'm just gonna pick some
3:15:25 random spot where I went back to Alec
3:15:27 Baldwin is overweight on occasion thank
3:15:29 you
3:15:32 you should use AI to turn your voice
3:15:34 into Rosie's voice I should do that that
3:15:36 would be good
3:15:37 all right Hunter Wampus good best AI
3:15:40 tool for placing my podcast transcript
3:15:42 and generating blog posts show notes
3:15:45 social captions
3:15:47 well what I would do so if like I do I I
3:15:50 do this thing called the AI salon and we
3:15:53 have two hour meetings and the the
3:15:55 transcripts are too big to to paste into
3:15:57 chat GPT so
3:15:59 that's why I have po.com up here
3:16:02 so I showed this earlier but I'll show
3:16:04 it to you as well if you go to poe.com
3:16:08 if you pay for their premium you get
3:16:10 access to three premium Services gpt4
3:16:13 Claude plus which is anthropics version
3:16:16 of gpt4 and then this Claude instant
3:16:19 100K that's a hundred thousand token
3:16:22 thing you can pay 75 000 words in that
3:16:25 so unless you're doing four hour
3:16:27 podcasts you can probably just dump your
3:16:29 whole transcript right into Claude 100k
3:16:33 I there might be other places that you
3:16:36 can use cloud 100K but I I don't I don't
3:16:39 know of them there there might be one on
3:16:41 the anthropic website but
3:16:43 that is how I would do that
3:16:47 um the other thing you can do if you
3:16:48 want to geek out and install Lang chain
3:16:51 you can just point Lang chain into
3:16:53 directory of a bunch of documents like
3:16:55 you could have a bunch of historical
3:16:56 transcripts in there and just point Lang
3:16:58 chain at it and embed all that [ __ ] and
3:17:01 and dump it into a vector database
3:17:04 um but you gotta geek out for that so
3:17:07 in in the absence of geeking out pay 20
3:17:09 bucks to Poe and just go use that thing
3:17:12 and then you know all of this stuff I I
3:17:15 would say
3:17:16 with any of these AI tools do not pay
3:17:19 the
3:17:20 um save
3:17:21 40 or 20 percent by by buying the year
3:17:25 subscription I wouldn't buy a Year's
3:17:27 subscription on [ __ ] right now because
3:17:29 everything's moving so fast that three
3:17:31 months from now most of what you
3:17:33 subscribe to will be obsolete or out of
3:17:35 business so I would just go month to
3:17:37 month
3:17:38 can chat gp2 T do SEO it can it can do
3:17:42 it really well and there's a bunch of
3:17:44 plugins for SEO so if you if you enable
3:17:47 plugins it'll go to like website
3:17:50 analysis and SEO analysis
3:17:52 I have a darker
3:17:56 um
3:17:57 I I have a
3:17:59 dark
3:18:01 um
3:18:03 I I don't think SEO in its current form
3:18:07 is long for this world
3:18:09 um
3:18:10 I think I think the the the the current
3:18:14 search method is broken really broken
3:18:17 and and it got stuck there because
3:18:20 Google effectively had a monopoly so
3:18:22 they just said this is the way it is and
3:18:23 [ __ ] deal with it and everyone dealt
3:18:25 with it and then Along Comes A Tool
3:18:27 where it's not it doesn't give you a
3:18:31 list of 40 things to click on it just
3:18:32 gives you an answer and and so
3:18:35 as the chat GPT tool gets more and more
3:18:39 sophisticated it's going to be
3:18:41 dynamically generating content for you
3:18:43 without you having to go to websites so
3:18:46 I think the whole idea of going to a
3:18:48 website and driving traffic to a website
3:18:51 is going to shift dramatically so I
3:18:53 would say short term yes it can do all
3:18:56 of the tactics that are SEO I just think
3:18:58 that SEO tactics are just going to see a
3:19:02 slow kind of slide decline in
3:19:05 ineffectiveness as more and more people
3:19:07 use these these tools that just give
3:19:09 them their answers
3:19:10 but we'll see this is why it's such an
3:19:13 existential threat for Google right now
3:19:15 best AI tool for placing podcast I did
3:19:17 that one what is there a way to create a
3:19:21 course using AI yeah absolutely things
3:19:23 on screen like you should be you should
3:19:25 go to chat.openai.com and just while I'm
3:19:28 rambling just just play so I'm just
3:19:30 going to leave it in chat GPT 3.5 and
3:19:33 I'm going to say Act
3:19:35 as an instructional designer
3:19:40 and
3:19:43 specializes in
3:19:46 um let's see team building and
3:19:51 Executive
3:19:54 Education period
3:19:57 um
3:19:58 I need an outline for a two-day workshop
3:20:07 with
3:20:09 20 executives
3:20:14 um who don't like each other
3:20:18 [Laughter]
3:20:23 it is very political and toxic
3:20:30 can you help me design a course
3:20:36 to mend the fences so I like purposely
3:20:40 put in like shitty arbitrary weird
3:20:42 language look how fast it's writing it
3:20:44 holy [ __ ]
3:20:46 so let's let's scroll up and see what
3:20:48 we've actually created here
3:20:51 um Workshop title building bridges
3:20:54 fostering collaboration and Trust in
3:20:56 high-stakes environment that's that's
3:20:58 good corporate speak for you're a bunch
3:21:01 of [ __ ] why don't we try not doing
3:21:03 that all right target audience 20
3:21:05 Executives who experience political and
3:21:07 toxic Dynamics within their team day one
3:21:09 setting the foundation Icebreaker
3:21:12 activities sharing individual
3:21:13 expectations establishing ground rules
3:21:15 for respectful Communications and active
3:21:18 participation let's go down and
3:21:20 challenge it on that one let's see I'm
3:21:22 going to say the third bullet in day one
3:21:25 wait the third bullet
3:21:27 in day one session one
3:21:31 can you tell me more about
3:21:36 day one session
3:21:39 one bullet three so one of the things
3:21:43 that's cool about chat EBT is it knows
3:21:46 what it typed before right so I can just
3:21:48 say go go to Bullet one you know or
3:21:51 bullet three Session One certainly day
3:21:54 one session one bullet three
3:21:55 participants collectively establish and
3:21:58 set ground rules that will govern
3:22:00 their interactions these ground rules
3:22:02 are designed to create a safe and
3:22:03 respectful environment so so it's
3:22:06 understanding that what I asked it was
3:22:08 you know a toxic environment so it's
3:22:10 it's reinforcing that we want to work
3:22:13 against that here are some steps to
3:22:15 follow explain the purpose of
3:22:16 establishing the ground Rules start by
3:22:19 emphasizing importance of setting clear
3:22:21 goals brainstorm ground rule ideas avoid
3:22:24 personal attacks
3:22:26 derogatory language discuss and refine
3:22:29 the suggested rules gain consensus
3:22:33 display blah blah blah blah blah
3:22:35 um let's see how would I
3:22:40 tactfully
3:22:44 um
3:22:48 call certain people out for their toxic
3:22:55 Behavior
3:22:59 preemptively
3:23:02 this is probably all really bad
3:23:05 um corporate
3:23:06 development practices but you can see
3:23:10 that it like
3:23:12 is there to support you and kind of
3:23:14 amplify what you do
3:23:17 come on Focus you dumb dumb
3:23:20 no ah
3:23:24 there you go
3:23:27 a toxic addressing toxic behavior
3:23:30 preemptively requires delicate approach
3:23:32 to ensure that it doesn't escalate
3:23:34 tensions that's right
3:23:36 create defensive response here are some
3:23:39 strategies set expectations use
3:23:41 hypothetical scenarios frame it as a
3:23:44 collective issue instead of individuals
3:23:46 emphasize the impact on team performance
3:23:48 like this is good good stuff right like
3:23:51 it understands empathetic engagement it
3:23:54 might not be empathetic but it
3:23:55 understands empathetic engagement so
3:23:58 um yeah so yes it can
3:24:01 is there a way to create a course yes
3:24:03 you go there and you say I want to
3:24:05 create a course and you know give if if
3:24:07 there's a particular
3:24:09 um course framework that you're a fan of
3:24:12 tell it to use that framework and and to
3:24:15 organize everything within that
3:24:16 framework
3:24:19 uh hello hello best AI tool for placing
3:24:22 okay hey Kyle hey Drew how's it going
3:24:25 is that true I think drew that used to
3:24:28 work at storyvine if that's you peace
3:24:30 man
3:24:31 what's happening
3:24:33 I don't know if you knew this but I'm an
3:24:34 influencer now on the tick tock
3:24:37 I Twerk and everything
3:24:39 no I don't
3:24:41 I spare everyone that
3:24:45 Has anyone used chat base I have not
3:24:49 better to Learn Python
3:24:52 better to Learn Python in chat GPT
3:24:56 well
3:24:59 I mean whatever's your way of learning
3:25:01 the nice thing about learning Python and
3:25:03 chat gbt is it can be a combination of
3:25:06 you learning the concepts but not having
3:25:08 having to learn the details it can do
3:25:10 the details but if you really want to
3:25:12 learn the details of the syntax and all
3:25:14 that sort of [ __ ] then
3:25:16 it might be better to just take like a
3:25:18 YouTube python code but
3:25:20 I am way too add and way too lazy for
3:25:23 that [ __ ] if I got a tool that's making
3:25:25 code for me well in fact like I did when
3:25:28 I made those content creators like it
3:25:30 was sending out emails and and they
3:25:32 looked like [ __ ] and I'm like oh I gotta
3:25:35 put HTML on them I'm like I haven't done
3:25:37 HTML in like 25 years I don't know how
3:25:39 to [ __ ] do HTML anymore like it's way
3:25:42 more sophisticated than the [ __ ] I was
3:25:44 doing
3:25:44 so I just had chat EBT do it I said you
3:25:49 know make this email
3:25:52 um
3:25:53 look pretty using HTML make sure that
3:25:56 the HTML is safe for eat you know email
3:25:58 programs because it's a really limited
3:26:01 subset of HTML and it just [ __ ]
3:26:03 knocked it out and it worked like within
3:26:06 10 minutes I had a version of HTML that
3:26:08 like I could you know go in there and
3:26:10 tweak [ __ ] if I needed to tweak it but
3:26:11 it's amazing it's just amazing it's
3:26:14 incredible I'll tell you it's incredible
3:26:16 I don't know if you if you haven't heard
3:26:18 of chat jpt it's gonna be the next big
3:26:20 thing
3:26:21 hey
3:26:25 I hope someone just came in right when I
3:26:27 did that that would be awesome
3:26:29 like what the [ __ ] is this
3:26:31 scrolling along
3:26:33 all right nobody hired me to make music
3:26:35 videos before AI so thanks to able to
3:26:38 Fusion well I mean you know that's an
3:26:40 interesting thing Harold is that
3:26:42 these tools you can sort of take your
3:26:44 creative intent and these tools can help
3:26:47 you
3:26:48 you know do stuff that you might not
3:26:50 have been able to otherwise do and if
3:26:52 you can go in and you can say yeah I can
3:26:54 create you an animation for 500 bucks
3:26:56 and execute it in a day knowing that you
3:26:58 can execute it in two hours
3:27:00 um it's probably going to be way cheaper
3:27:02 than what they've been quoted for that
3:27:03 right so
3:27:05 [ __ ] power on I think that's awesome
3:27:10 um I know I'm freaking out thanks y'all
3:27:12 yeah listen that's exactly why this is
3:27:14 here it's like go start playing with
3:27:16 this stuff and like take the thing
3:27:18 you're passionate about it the thing you
3:27:19 know some little something about and
3:27:22 figure out how these tools you know
3:27:24 either these word tools or the image
3:27:25 tools like stable diffusion can help you
3:27:28 and uh the heavy metal scene was so so
3:27:30 accepting of AI art yeah they should be
3:27:33 because those like heavy metal
3:27:35 illustrations are you know they're
3:27:37 [ __ ] intense and now you can just
3:27:39 knock them out like that
3:27:41 um that that's an interesting thing in
3:27:43 in the art World in particular there's
3:27:45 there's you know a fairly visceral and
3:27:48 violent
3:27:49 you know resistance against AI art
3:27:53 um and I understand it right like this
3:27:56 is people that have been you know
3:27:57 illustrators for 30 years all of a
3:27:59 sudden this [ __ ] tool can just knock
3:28:02 that [ __ ] out you know what takes them a
3:28:04 week or a month to illustrate is just
3:28:06 like that
3:28:08 um I think the Adobe I think Adobe
3:28:12 um
3:28:13 being cognizant of artists and building
3:28:17 their model in a way that doesn't
3:28:18 infringe on artists work is really
3:28:21 really smart I think that's going to
3:28:22 pave the way to people accepting it but
3:28:25 every time there's a new technology
3:28:26 there's there's resistance like this
3:28:28 every single time
3:28:31 um I like this take on being curious and
3:28:33 also disrupting yourself yeah like get
3:28:35 curious get adventurous
3:28:37 um
3:28:37 learn enough about this stuff to know
3:28:40 like whatever your job is go figure out
3:28:43 what this stuff can do better than you
3:28:46 can
3:28:47 and you'll find it and then when you
3:28:49 find it you'll go oh [ __ ] really
3:28:53 oh
3:28:55 and then just breathe like accept the
3:28:59 terror just sort of Park that on this
3:29:01 shoulder and then go okay
3:29:04 now that I know it can do that what else
3:29:07 could it do that that because I know
3:29:10 this space really well what could I make
3:29:12 it do that maybe someone who doesn't
3:29:14 have my experience won't figure that out
3:29:16 that's what will give you the advantage
3:29:19 and and then the like the terror and the
3:29:21 existential threat is absolutely going
3:29:23 to be there but just find the balance of
3:29:25 it about okay but what can it do
3:29:27 and what you might find is that like
3:29:30 oh well I really didn't like that part
3:29:33 of my job anyway so holy [ __ ] this thing
3:29:35 could automate the part of my job I
3:29:37 hated but you know what part I liked was
3:29:39 this piece and then you go start
3:29:40 exploring these tools with the part you
3:29:42 really love and you're like oh
3:29:45 this could be better right and like
3:29:47 that's that's I think how we navigate
3:29:49 through this
3:29:51 where's a good place to start for a
3:29:52 newbie right here that's why I keep
3:29:55 pointing like I've got you know some
3:29:57 sort of OCD spectrum disorder
3:30:01 go to
3:30:02 chat.openai.com go to chatgpd you're
3:30:06 talking about
3:30:08 uh
3:30:12 uh uh sure Harrow pop your pop your
3:30:15 channel in the chat
3:30:17 um and people should go check it out
3:30:18 that's that's pretty exciting
3:30:21 um what is as good as gpt4 that can run
3:30:24 locally on Unix no idea what to do I
3:30:28 don't know why you want to run it
3:30:29 locally on Linux like I would say just
3:30:32 to learn it just go play with this [ __ ]
3:30:35 and then if you're putting something
3:30:36 locally on on Linux
3:30:38 go check out a project on GitHub called
3:30:41 GPT for all GPT for all
3:30:46 um and that's something that you can
3:30:48 install on a local machine I'm pretty
3:30:50 sure there's a Linux Linux install but
3:30:52 if not you could probably hack one
3:30:53 together and then you can use lots of
3:30:56 the different open source large language
3:30:58 models to my
3:31:00 knowledge there's not an open source
3:31:02 large language model that's as good as
3:31:03 gpt4 there there are some that are now
3:31:06 really really close to GPT 3.5
3:31:09 and there are a bunch of projects that
3:31:11 you can install locally on your machine
3:31:12 but I would say if you're just getting
3:31:14 started [ __ ] it just go use this web
3:31:17 version you know like if you're afraid
3:31:19 of privacy and security and [ __ ] like
3:31:21 that then just put it on a burner laptop
3:31:23 like you don't you don't need a really
3:31:25 powerful machine to go use this they're
3:31:27 using all of Microsoft's iron right so
3:31:30 just don't put anything sensitive in
3:31:32 there and and just go play and then as
3:31:35 you get up to speed on that just start
3:31:37 investigating the open source llms
3:31:41 the reason I'm not talking about any
3:31:43 specific open source projects right now
3:31:46 like like the specific llms is like
3:31:48 every three days they top one another
3:31:50 like every three days there's a new open
3:31:52 source model coming out that some
3:31:53 [ __ ] remarkable you know advancement
3:31:56 it was like for for 37 cents we trained
3:32:00 this thing that can you know fly you
3:32:03 know fighter jets
3:32:04 whatever like it's just like every three
3:32:07 days there's just new [ __ ] coming out so
3:32:10 um so yeah go check out GPT for all
3:32:14 why is the system so scared of the woke
3:32:17 mind agenda I can't really talk about
3:32:18 any of that the two times I started
3:32:21 talking about bias
3:32:23 in the models I got booted off The live
3:32:27 and and booted out for seven days so I
3:32:29 don't want to talk about it but it's
3:32:31 basically
3:32:33 It's a combination of the data it's
3:32:35 trained on and then the human
3:32:36 reinforcement learning you've got humans
3:32:38 that are saying that's a good answer
3:32:39 that's a bad answer
3:32:41 you know every human that partic
3:32:43 participates in this they have their own
3:32:45 biases and and it can inject bias into
3:32:47 the system there's a there's a system
3:32:49 called anthropic uh or there's a company
3:32:52 called anthropic it's a bunch of EX open
3:32:55 AI people and they've got a model called
3:32:57 the Constitutional learning model that
3:32:59 rather than having humans doing the
3:33:01 thumbs up thumbs down they write a
3:33:02 constitution they have a set of rules
3:33:04 like
3:33:05 you know don't Destroy Humans increase
3:33:08 Prosperity
3:33:10 um don't be an [ __ ] whatever the
3:33:12 rules are right and then when it
3:33:14 generates an answer it it weighs that
3:33:16 answer against the Constitution and it
3:33:18 will iterate it until it passes the
3:33:20 Constitution so there's there's models
3:33:22 out there that are starting to
3:33:25 um deal with the bias thing
3:33:28 but the deal with bias is you're never
3:33:31 going to have a single large language
3:33:34 model that pleases everyone so I think
3:33:36 what we'll see in the future is you'll
3:33:38 have different kinds of models this one
3:33:40 might lean more that direction this one
3:33:42 might lean more that direction this one
3:33:44 might be neutral this one might be good
3:33:46 for science this one might be good for
3:33:47 writing you know so I think we'll start
3:33:50 to see choices when it comes to
3:33:53 the flavor of your large language model
3:33:56 but I can't really talk about it because
3:33:58 I seem to keep getting
3:34:00 kicked off this thing when I talk about
3:34:02 anything even remotely political so [ __ ]
3:34:05 it
3:34:11 my work is already talking about adding
3:34:14 AI to help do the easy stuff so we can
3:34:17 get more clients yes
3:34:19 let chat GPT do the account management
3:34:22 [ __ ] and some of the project management
3:34:24 drudgery and have it write follow-up
3:34:26 letters and have it
3:34:28 like help do some of the back office
3:34:30 stuff and and so it can free you to do
3:34:33 the higher level engagements that's
3:34:34 absolutely what this is about and I
3:34:37 think the Smart Companies
3:34:39 are not going to lean into oh we can lay
3:34:41 people off the Smart Companies are gonna
3:34:43 go oh holy [ __ ] instead of spending all
3:34:46 this money on all this shitty low value
3:34:47 activity let's automate that and let's
3:34:50 all focus on the better [ __ ] that's more
3:34:53 fulfilling and is better for our clients
3:34:55 I work for a large technology company
3:34:57 based on internal IP value how can chat
3:35:02 GPT help with internal IP
3:35:05 I think internal data sets are are the
3:35:09 the actual superpower of
3:35:13 all companies especially big companies
3:35:17 so there's a thing that was announced
3:35:20 six weeks ago eight weeks ago called
3:35:23 Bloomberg GPT
3:35:25 and I don't know if they've created it
3:35:27 yet but they announced it and I assume
3:35:29 you have to pay a lot for it and it's
3:35:31 basically chat GPT but aimed at
3:35:33 Bloomberg financial data like all the
3:35:36 you know proprietary Bloomberg [ __ ]
3:35:39 um they're gonna make a [ __ ] fortune
3:35:41 on that right and I think any company
3:35:43 that's got internal data that is of any
3:35:47 kind of value you basically just plop
3:35:50 this large language model on top of it
3:35:52 and it makes it all discoverable in you
3:35:56 know massively unique and Powerful ways
3:35:57 now there's all sorts of
3:36:01 security and privacy issues like you're
3:36:03 going to need to run your own machines
3:36:05 and you know firewalls and all that sort
3:36:07 of [ __ ] and figure out what you put out
3:36:08 to the world and
3:36:10 but I think I think there's deep deep
3:36:13 tremendous value is I think that's where
3:36:16 kind of long-term value is going to come
3:36:18 from big companies
3:36:20 with data that was historically kind of
3:36:23 locked into these obscure databases that
3:36:25 you couldn't really really get value out
3:36:27 of without doing lots of programming you
3:36:29 can now just
3:36:30 you know get into a vector database and
3:36:33 just chat GPT the [ __ ] out of it so
3:36:36 um
3:36:37 so so what I would I would look into you
3:36:40 know Azure cloud services
3:36:43 um uh open AI has Standalone like chat
3:36:47 GPT servers that you can you can spin up
3:36:49 they like start at like 25 Grand a month
3:36:53 or a year
3:36:55 I think it's 25 Grand a year but they go
3:36:57 up to like a million or a million five
3:37:00 right for for these are Big Azure
3:37:02 instances I'm sure
3:37:04 um but I think that's huge I would get
3:37:06 your I.T department on that ship pronto
3:37:09 have them start experimenting with Lang
3:37:11 chain and and look at Azure and what
3:37:14 they're doing and Amazon AWS has
3:37:17 announced they've got a bunch of tools
3:37:18 so I I'm sure there's tools out there
3:37:21 right now that you can start building
3:37:22 that [ __ ]
3:37:24 where can I download gpt4 without all
3:37:27 these scam sites that's why these are
3:37:29 here
3:37:30 you get my money I'll catch everything
3:37:31 don't go to the scam sites
3:37:34 um
3:37:35 chat.openai.com you don't download
3:37:39 um
3:37:40 T you you can download an IOS app
3:37:44 that is the official one but I wouldn't
3:37:46 even do that just point your browser
3:37:48 whether you're on a mobile device or a
3:37:50 desktop to chat.openai.com or bing.com
3:37:54 those top two bing.com is free
3:37:58 chat.openai.com they're the ones who
3:38:01 made it so that's the official site
3:38:03 anything else is [ __ ]
3:38:06 so avoid the scammers there's so many of
3:38:09 them now
3:38:11 oh great Harrow just told you he told he
3:38:13 probably told you that 15 minutes ago so
3:38:15 so thank you Harold
3:38:17 um I am there do I ask the question to
3:38:21 download
3:38:23 you don't download
3:38:25 gpt4
3:38:27 so
3:38:28 when you go to chat GPT
3:38:31 you'll see this site you you'll have a
3:38:33 chat history over here and you'll have
3:38:37 I got this big chat window there's
3:38:39 nothing to download so you got GPT 3.5
3:38:41 gpt4 if you're not paying for plus you
3:38:44 won't see four but GPT 3.5 you just have
3:38:48 this little window on the bottom and
3:38:49 you're like
3:38:50 right
3:38:52 uh me a pop ballad about
3:39:00 um
3:39:01 little yellow birds
3:39:08 and it will write you a ballot about
3:39:10 little yellow birds
3:39:13 right
3:39:14 in a world please guys there's a tale to
3:39:17 be heard
3:39:19 this is gonna be good
3:39:22 a little yellow bird singing their songs
3:39:24 word by word they dance on the branches
3:39:27 like sunshine on wings their Melodies
3:39:29 are pure the happiness it brings
3:39:32 um so yeah you shouldn't be told to
3:39:35 download anything
3:39:39 um if you really want the official chat
3:39:42 GPT app go to openai.com
3:39:48 and the first thing that you see is
3:39:51 there's an IOS app for chat GPT and you
3:39:54 click on that link don't search for it
3:39:56 in the App Store there are too many scam
3:39:58 sites to figure it out
3:40:01 I can't figure out why Apple hasn't shut
3:40:03 that [ __ ] down but they haven't
3:40:07 so that's why this stuff's up here go
3:40:10 there
3:40:11 Public Service Announcement Do You Know
3:40:14 Where Your Children Are
3:40:15 hey gen xers remember all those
3:40:17 commercials
3:40:18 all right Chief generative officer I
3:40:21 know isn't that a cool title I didn't
3:40:22 come up with that the guy that runs
3:40:24 content Evolution came up with it and he
3:40:25 goes he goes yeah you built this thing
3:40:27 that was pretty good I think we should
3:40:28 memorialize this how would you like to
3:40:30 title Chief generative officer I'm like
3:40:32 I'm [ __ ] in
3:40:34 it's like yes it was funny when I put
3:40:37 that on LinkedIn people were like oh my
3:40:40 congratulations sir well
3:40:44 I don't mean to brag
3:40:49 uh I'm gonna try having AI compose my
3:40:53 emails tomorrow okay absolutely do that
3:40:55 gossamer prints like start that it's a
3:40:58 perfect way to do it like start with
3:41:00 what I call parlor tricks have it write
3:41:02 songs have it write poems make a table
3:41:04 like just dick around with it and then
3:41:07 and then yeah like start to think about
3:41:09 what you do at work and start to think
3:41:11 about the things that you do at work
3:41:12 that take you way [ __ ] longer than
3:41:14 they should
3:41:15 and just kill your brain
3:41:19 and it could probably do it better and
3:41:21 here's another thing to do if you're
3:41:22 gonna have it write emails this is a
3:41:24 cool thing go
3:41:26 based on this email and then paste in
3:41:29 one of your existing emails that you've
3:41:31 written that you're really happy with
3:41:32 that it's like you Wordsmith it and it's
3:41:34 good
3:41:35 so say based on this email paste in your
3:41:38 email write me a new email to this
3:41:40 person about this topic using that one
3:41:43 as an example
3:41:44 and just watch what it does you'll be
3:41:46 like holy [ __ ] I'm never not doing this
3:41:48 again
3:41:50 it's like
3:41:55 um content Evolution yeah content
3:41:57 evolution
3:41:59 so in depth I love it I don't know what
3:42:01 I was doing but cool seriously good
3:42:02 stuff yes double tap the screen
3:42:04 Professor Sonia yes tell people to
3:42:07 double tap the screen I think oh it
3:42:09 looks like some new people are in here
3:42:11 um they're leaving because I'm just
3:42:13 rambling uh if you're new here my name
3:42:14 is Kyle Shannon this is the AI learning
3:42:16 lab I've been I think I'm on three and a
3:42:19 half hours now I've been doing this live
3:42:21 for three and a half hours because I'm a
3:42:23 dum-dum
3:42:24 um
3:42:25 I'm an entrepreneur I'm a CEO of a
3:42:28 company called storyvine these sites
3:42:31 here are just here if you haven't played
3:42:32 with chat GPT please go try it
3:42:36 um it's the the price of admission for
3:42:39 this free live stream is go to
3:42:41 chat.openai.com and start [ __ ] around
3:42:44 with chat gbt if you haven't done it yet
3:42:45 also bing.com the chat button there is
3:42:48 also chat GPT it's gpt4 for free
3:42:53 um
3:42:54 and the other things are other resources
3:42:56 so is this live yes it's live but I'm
3:42:58 very behind on my questions what is a
3:43:00 self-hosted AI that can run on my own
3:43:02 Hardware
3:43:04 um
3:43:05 go check out GPT for all
3:43:08 and then you can use different models
3:43:11 within it if if you want to start
3:43:13 installing things locally on your
3:43:15 machine you're going to need to geek out
3:43:17 a bit on all the different open source
3:43:18 large language models because there's a
3:43:20 lot of them
3:43:21 and they keep getting updated and um
3:43:25 you're just gonna have to play like
3:43:27 where I am right now is things are
3:43:29 changing so fast I know that gpt4 that
3:43:33 that thing right there
3:43:35 is better than any tool I've ever used
3:43:37 in my life
3:43:38 by orders of magnitude
3:43:41 and so I'm just kind of sticking with
3:43:43 that right now all this all this other
3:43:45 stuff
3:43:47 until the noise kind of settles down
3:43:50 like I could spend the next decade
3:43:51 seriously with just GPT as it stands
3:43:54 today without plug-ins
3:43:57 I could spend the next decade figuring
3:43:58 out what the [ __ ] to do with that thing
3:44:00 so there is so much depth there it's
3:44:02 it's it's it's actually hard to
3:44:04 comprehend
3:44:06 there's no precursor to this
3:44:08 technological Advance intelligent
3:44:09 technology has never existed yeah I like
3:44:12 that's the thing I like
3:44:15 everything that I've experienced in my
3:44:17 life has been incremental advances like
3:44:20 when I was in seventh grade and I
3:44:22 experienced the personal computer you
3:44:24 could kind of look at the computers
3:44:26 before and go oh this is just that but a
3:44:28 little smaller
3:44:30 um the World Wide Web was you know kind
3:44:33 of like the internet but you could click
3:44:35 on it but like it was
3:44:37 this stuff is like it looks like a
3:44:40 Google Input box
3:44:42 but then what it does ain't Google it's
3:44:45 like we're not in Kansas anymore
3:44:47 the difference will be Stark yep totally
3:44:50 agree how do you learn how to use AI to
3:44:53 catch up to the curve for one thing
3:44:55 Austin
3:44:56 they ain't no catching up this curve is
3:44:59 is accelerating so far behind but beyond
3:45:02 any of our capacity to keep up with it
3:45:07 you can only choose to step onto the
3:45:11 train knowing that you're never going to
3:45:13 get to the front of it because it's just
3:45:15 it's it's it's Way Beyond any of our
3:45:18 capacity to keep up with it it's
3:45:19 advancing that fast
3:45:21 the way to do it in my opinion and it's
3:45:24 why I have these things on screen here
3:45:26 go to chat.openai.com start playing with
3:45:28 chat GPT just start learning what it
3:45:31 does and then take what you know what
3:45:33 you're passionate about
3:45:35 and start asking it to do work for you
3:45:37 or do projects for you or like the the
3:45:41 one guy in here just said I think I'm
3:45:42 going to use it to write emails for me
3:45:44 at work tomorrow go have it write emails
3:45:46 for you at work just start learning what
3:45:49 this stuff does and get curious about it
3:45:51 it's really the only way in
3:45:54 um
3:45:55 because it's I it's it's it's it's
3:45:58 impossible to understand it from the
3:46:01 outside is the only thing I can say
3:46:04 um and you're not behind that's the
3:46:06 other thing chat GPT that that site
3:46:11 came out on November 30th 2022. I I
3:46:14 think this it marks a moment in history
3:46:16 for sure but it's only been out for six
3:46:18 months and it's we're early right Ai and
3:46:23 machine learning has been around for
3:46:24 decades but this let's call this popular
3:46:26 AI right
3:46:28 popular AI
3:46:30 AI for the rest of us has only been
3:46:33 around for six months
3:46:34 and so you're it's you're very very
3:46:37 early anyone that steps onto this train
3:46:40 right now for probably the next year is
3:46:44 setting themselves up to have massive
3:46:46 opportunity because as as businesses
3:46:49 start to realize holy [ __ ] we need to
3:46:52 get into this now
3:46:53 who can spell AI we're gonna make them
3:46:56 the VP of machine learning for the
3:46:58 company like like there's going to be a
3:47:01 panic that sets in so anyone that can
3:47:03 kind of raise their hand and go I know
3:47:05 how to do this [ __ ] you're going to
3:47:07 Serve Yourself well so don't worry about
3:47:08 that you're behind I like the the last
3:47:14 four days I've not been paying attention
3:47:17 to the new [ __ ] and I I haven't been
3:47:19 posting Tick Tock videos either I've
3:47:21 been doing a lot of lives
3:47:24 um I feel so out of touch like four days
3:47:26 of not paying attention to it like
3:47:28 acutely I feel like completely lost
3:47:30 right now but whatever doesn't matter
3:47:32 just get on the train start learning it
3:47:35 start having fun with it
3:47:37 oh thank you very much someone said they
3:47:39 love me thank you I appreciate that is
3:47:42 there anything better than GPT Ford not
3:47:44 right now gpt4 is is has set the
3:47:47 standard
3:47:49 um Bard is a train wreck I think Bard's
3:47:51 not worth using I think the the
3:47:54 um anthropic models Claude Claude plus
3:47:57 and and Claude 100K and Claude instant
3:48:01 are all good
3:48:03 um
3:48:04 Bing is gpt4 but it's a little bit
3:48:07 hobbled because of some [ __ ] that
3:48:10 Microsoft went through it'll get better
3:48:11 but Bing's not quite as good as the
3:48:14 original chat EBT so right now chat GPT
3:48:18 4 is King of the Hill in my opinion
3:48:21 are you working on any AI projects well
3:48:24 I'm I'm building some of these content
3:48:26 generation tools I rolled whisper
3:48:29 transcriptions into my video platform
3:48:31 about five months ago
3:48:35 and then what I'm doing right now is
3:48:38 um creating a whole new vision for the
3:48:41 company really where
3:48:43 so so the way our our technology works
3:48:46 we have an app that someone answers a
3:48:48 series of questions in the app like to
3:48:50 tell a story like when were you
3:48:52 diagnosed and how did that feel and how
3:48:54 are you treating it now and what are
3:48:55 your hopes and dreams for the future so
3:48:56 they answer those questions that goes up
3:48:58 to the cloud and it creates a fully
3:49:00 edited video within about five minutes
3:49:02 so that's been the product for 11 years
3:49:05 so now what I'm doing is I'm taking the
3:49:07 transcript of those videos and I'm using
3:49:09 generative AI to create a whole Omni
3:49:11 Channel content world around that story
3:49:14 so I take the core story and then I
3:49:16 Remix it into all these different
3:49:18 outputs so that's the thing I'm working
3:49:20 on right now and it's it's insane
3:49:23 because it's like I'm I'm taking
3:49:25 I'm taking a company that's essentially
3:49:27 had the same product for 11 years and
3:49:29 I'm having to completely rethink it and
3:49:31 completely rethink how I position it how
3:49:34 I talk about it
3:49:36 what I retain from the value proposition
3:49:38 what I jettison
3:49:41 um it's intense it's it's existential
3:49:43 [ __ ]
3:49:50 didn't get dumber when the calculator
3:49:52 was invented AI will allow humans to be
3:49:54 more creative one thousand percent agree
3:49:57 absolutely I have never been more
3:50:00 creatively engaged
3:50:03 and alive in my life
3:50:07 like these tools are making me smarter
3:50:10 more creative and I'm executing that
3:50:13 creative work better
3:50:16 it's insane
3:50:18 it's just stupid it's just stupid good
3:50:21 is there an AI tool to automate formulas
3:50:24 for Excel oh yeah so for one thing
3:50:27 Microsoft is rolling they're calling it
3:50:29 co-pilot but they're rolling gpt4 into
3:50:32 all of their products all of Office 365
3:50:35 so it's going to be built in there so
3:50:36 you can just wait for that or if you use
3:50:40 Google Sheets and I'm sure there's one
3:50:41 for Excel now there I know for Google
3:50:43 Sheets there's a an extension called
3:50:47 um GPT for Sheets and basically you
3:50:50 install it and then every cell of your
3:50:52 spreadsheet is a call to open AI if you
3:50:54 want it to be you basically just call it
3:50:56 like a formula like like uh whatever you
3:50:59 know slash
3:51:00 GPT and then you get a whole list of
3:51:04 things it can do and you can have one
3:51:06 cell generate a list of 10 names and
3:51:08 then you can take each of those names
3:51:09 and have it generated description and
3:51:11 you can take each of those descriptions
3:51:12 and have it generate a blog post and
3:51:14 it's it's absolutely insane so
3:51:17 um GPT for sheets for sheets and then
3:51:21 there's some equivalent for Excel I'm
3:51:22 sure or just wait for both Google and
3:51:25 Microsoft to roll them into the tools
3:51:28 I would guess within six months there's
3:51:31 there's a lot of competition right now
3:51:32 like Microsoft is giddy over how much
3:51:36 they're [ __ ] with Google so Google's
3:51:38 trying to catch up and Microsoft is like
3:51:41 trying you know
3:51:43 they keep doing that so there's there's
3:51:45 a little competition there it's good
3:51:49 um doesn't Excel already have ai built
3:51:51 in uh it's going to it's it's going to
3:51:53 be fully integrated
3:51:56 um there might be little Snippets here
3:51:57 and there I mean listen
3:51:59 we've had AI in our software for like
3:52:01 decades right like when it does spell
3:52:04 check or when it finishes the sentence
3:52:06 for you in Google that's all AI right so
3:52:09 all the these large language models are
3:52:12 is like way more intense versions of
3:52:14 that right like it's creating whole
3:52:15 pages instead of you know little
3:52:18 sentence Snippets
3:52:20 do any do you know any good prompts yeah
3:52:22 that's why I put this here prompts.chat
3:52:25 oh there's two things one is go to
3:52:27 props.chat go to that document it'll
3:52:29 teach you about prompting at the
3:52:31 beginning there's some eBooks and [ __ ]
3:52:33 like that that are free and then it's
3:52:35 just a whole list of prompts where you
3:52:36 can have chat EPT act like a certain
3:52:39 Persona so go there that'll teach you a
3:52:41 lot the other thing I would say about
3:52:43 prompts is don't don't worry as much
3:52:47 about learning
3:52:50 the specific prompt mechanics as the
3:52:54 concept that
3:52:56 it's you don't need to write a perfect
3:52:58 prompt and then get a perfect response
3:53:00 it's much more like a conversation with
3:53:03 a colleague
3:53:04 hey I'm thinking about doing this you
3:53:06 know can you help me with that sure
3:53:07 here's a list oh that's not what I had
3:53:09 in mind can you make the I didn't want a
3:53:12 list I wanted I wanted a paragraph where
3:53:15 you explain this this concept oh here's
3:53:17 that concept oh yeah that's more like it
3:53:18 now can you give me a list that talks
3:53:20 about like that back and forth is very
3:53:23 much the experience of working with
3:53:25 these things it's not like don't worry
3:53:27 too much about that break your thought
3:53:29 up into smaller sections like
3:53:32 rather than saying you know please write
3:53:35 me the Great American novel
3:53:37 you know start out having a conversation
3:53:39 about hey I'm thinking about writing a
3:53:41 book and I have some ideas can you help
3:53:42 me organize my thoughts sure I hope you
3:53:44 organize your thoughts right and then
3:53:46 and then start the conversation and then
3:53:49 in doing that you can get to writing the
3:53:51 Great American novel
3:53:54 all right
3:53:55 I am starting to burn out I think I've
3:53:58 been four hours at this or close to it
3:54:00 three and a half anyway all right I want
3:54:02 to get into AI art learn how to install
3:54:05 automatic 1111 for stable diffusion oh
3:54:08 if you want yeah Harold yeah if you want
3:54:10 to get into air learn how to install
3:54:12 automatic 1111 for stable diffusion for
3:54:14 free agree with that it took me nine
3:54:17 attempts nine failed attempts before I
3:54:19 got that up and running
3:54:21 um so you have to you have to be pretty
3:54:23 geeky if you don't want to be geeky
3:54:25 um I would look at a a project called
3:54:28 leonardo.ai which is a web front end uh
3:54:32 for stable diffusion that's really
3:54:33 [ __ ] good and you don't have to
3:54:35 install it you do have to pay for it
3:54:37 well no it's not true
3:54:38 leonardo.ai I think you have to get on
3:54:41 the wait list but their free version
3:54:43 they give you like 150 credits a day and
3:54:45 I think it's like five credits an image
3:54:48 so you get a fair amount of images per
3:54:50 day for free
3:54:51 um so so leonardo.ai you don't have to
3:54:54 geek out but if you want to geek out the
3:54:56 developer Community for stable diffusion
3:54:58 is insane there's all sorts of tools but
3:55:01 you gotta
3:55:02 you gotta have pretty you gotta have the
3:55:05 uh the code Noggin wired pretty good for
3:55:08 that [ __ ] you've been replaced by AI is
3:55:11 there a tool for automating formulas in
3:55:13 Excel we talked about that
3:55:15 all right I'm going to do some amount
3:55:18 more questions I always do this I'm like
3:55:21 I'm gonna go soon and then like three
3:55:22 hours later I'm like I'm gonna go soon
3:55:26 excuse me
3:55:31 all right let me see where it was in the
3:55:33 old questions yeah tap that like screen
3:55:35 I don't know what that does but the
3:55:37 hearts go up and it makes me smile
3:55:41 hell yeah my mom had eight kids to keep
3:55:44 track of she never knew where any of us
3:55:46 were that's a gen xer for sure
3:55:49 why do you drink from the house because
3:55:51 we weren't allowed in the house
3:55:57 no one had sunscreen look who are you
3:56:00 people
3:56:01 all right I remember when Photoshop came
3:56:03 out and everyone said that what you were
3:56:05 creating wasn't real art yeah we've seen
3:56:07 this there's one of the uh the the
3:56:09 presentation that I did
3:56:11 I did was this one
3:56:14 which is when photography first came out
3:56:17 there's this painter named Paul De La
3:56:19 Roche that saw the first photograph at
3:56:22 daguerreotype and he said from today
3:56:23 painting is dead
3:56:25 so it's like every time there's a new
3:56:27 technology the same [ __ ] happens that's
3:56:31 not real art [ __ ] you machine people
3:56:37 same [ __ ] every time every single time
3:56:40 but whatever
3:56:43 again
3:56:45 I don't fault people for feeling pissed
3:56:47 off and having those arguments I'm just
3:56:50 not interested in having those arguments
3:56:52 because they're irrelevant
3:56:54 to what's happening like it's com the
3:56:58 image stuff is here whether you want it
3:57:00 or not like you can't uninvent it right
3:57:03 and this what what these things are
3:57:06 going to turn into is here whether you
3:57:08 want it or not
3:57:09 so if you're a copywriter and you've
3:57:11 been a copywriter for 30 years
3:57:13 it sucks I get why you're pissed but the
3:57:17 game has changed these tools completely
3:57:20 transform that industry overnight if
3:57:23 you're a voice over actor The Voice
3:57:25 synthesis that's coming
3:57:27 that like everything's different
3:57:30 so yeah
3:57:33 GPT for all you two can make money
3:57:44 hell yeah oh yeah this is the eight kids
3:57:47 okay
3:57:49 I got lost in my little questions is the
3:57:52 premium post subscription for access to
3:57:54 gpt4 Claude Plus
3:57:56 and Claude instant a good deal I think
3:57:58 it is well
3:58:01 pose weird but Poe gives you access to
3:58:05 the Claude instant 100k
3:58:08 thing so so you can do a chat with a 75
3:58:12 000 Word document and just paste it into
3:58:14 the chat window and interact with it so
3:58:17 for me it's worth it for that alone
3:58:20 um
3:58:21 I don't think the PO interfaces as
3:58:23 elegant as chat GPT so I prefer chat
3:58:26 gpt's interface but if I want to do long
3:58:28 documents I go to Poe but just for that
3:58:31 so I don't know
3:58:33 I also I really do like having access to
3:58:36 the anthropic the Claude models because
3:58:38 I I'm a fan of the anthropic approach
3:58:43 I have a question first time here
3:58:46 come for the hair stay for the smart ass
3:58:49 comments yes first time here are the AI
3:58:52 results generated
3:58:54 are the AI results generated plagiarism
3:58:57 free
3:58:58 yes ish
3:59:02 so
3:59:03 the way these things work the way the
3:59:05 large language model works is it's
3:59:07 actually writing word by word
3:59:10 the response
3:59:12 so it's writing original content it is
3:59:15 not copying and pasting blocks of text
3:59:18 from this copywritten document and that
3:59:20 copywritten document and smooshing them
3:59:22 together that's not what it's doing
3:59:24 that said the base model is trained on
3:59:28 essentially all of the internet so in
3:59:31 that training data there is likely no
3:59:34 there is definitely copywritten material
3:59:36 that it's been trained on
3:59:39 where's the boundary between if it was
3:59:42 trained on that and something in this
3:59:45 original generated piece of content has
3:59:47 similarities where is it too close where
3:59:50 someone could sue you I don't know the
3:59:53 IP attorneys are going to make a [ __ ]
3:59:55 fortune over the next 20 years
3:59:57 um I would say this is way more
4:00:00 clear-cut in the image space than in the
4:00:03 word space but
4:00:05 lawyers of lawyers there's going to be
4:00:07 lawsuits in the word stuff
4:00:09 um
4:00:10 on the image side if you're with a big
4:00:13 company use Adobe Firefly and Photoshop
4:00:16 the generative AI stuff in there because
4:00:18 that model has been trained on no
4:00:20 copywritten content
4:00:23 um I think all of the large language
4:00:25 models have so but it's not plagiarism
4:00:28 but there's but it's a really weird gray
4:00:31 area where
4:00:32 you can't necessarily figure out the
4:00:35 percentage of what pieces were
4:00:37 contributed to your thing because it's
4:00:40 going word by word and it's coming out
4:00:42 of these semantic tangled webs of
4:00:45 words I don't know uh good question no
4:00:50 good answer
4:00:51 is this usable just on your phone
4:00:54 um
4:00:56 there some of the open source projects
4:00:59 are now installable on a phone
4:01:02 um they're probably more like gpt3 than
4:01:07 3.5 and they're definitely not gpt4 but
4:01:10 I would just just go start hanging out
4:01:13 on the subreddits
4:01:15 um where they're talking about open
4:01:17 source llms large language models and
4:01:20 and just start following those projects
4:01:22 because it's changing daily what's
4:01:25 possible I wrote a 14-day itinerary to
4:01:28 Italy with 13 excursions today but it
4:01:31 was hard to train it but I got it yeah
4:01:32 that's cool I
4:01:34 what's this live will end in one minute
4:01:36 verify to continue
4:01:41 ah
4:01:43 I think I'm still good I think I stayed
4:01:45 on too long it made me do a caption to
4:01:47 keep my live they're like this dude's a
4:01:49 [ __ ] robot
4:01:50 nobody goes live for this long I don't
4:01:53 know why I'm doing this this long I got
4:01:54 work tomorrow
4:01:55 but whatever it's fun
4:01:59 um yeah it's cool so so doing itinerary
4:02:01 stuff like that like
4:02:03 I think one of the best ways to learn
4:02:05 this thing is like give yourself a
4:02:07 specific project like like a 14-day
4:02:09 itinerary because you know what good is
4:02:11 right and so the first time you do it
4:02:14 it's shitty and then you're like no
4:02:15 that's not right and then but all that
4:02:16 navigating to get to the thing that you
4:02:19 knew you wanted like that's how you
4:02:20 learn this [ __ ]
4:02:22 because it because it's like I don't you
4:02:24 just can't learn it you can look at
4:02:27 technically how to do prompting but it's
4:02:29 a until you do problem solving like that
4:02:31 it's really hard to understand
4:02:33 all right super informative I appreciate
4:02:35 that I think days of General AI are
4:02:38 drawing to a close I don't quite know
4:02:40 what you mean there
4:02:42 General AI I don't know what that means
4:02:45 can you share what types of data sets
4:02:48 most llms are trained on
4:02:51 um they're they're all over the place
4:02:53 there's interviews with Sam Altman where
4:02:55 he talks about the kinds of data he and
4:02:57 he said
4:02:59 one of the reasons
4:03:01 um Chachi BT's better that gpt4 is
4:03:04 better than others and GPT 3.5 is better
4:03:06 to others they spent a lot of years
4:03:11 they recognized early on that if you
4:03:13 threw high quality data and lots and
4:03:16 lots of processing at these things that
4:03:18 they could get really good
4:03:20 and so
4:03:21 they spend a lot of their early years
4:03:24 categorizing and you know cleaning data
4:03:27 like basically finding all the data
4:03:30 so I think I think open AI seems to have
4:03:34 a head start on just you know get as
4:03:36 much data as [ __ ] possible and get it
4:03:39 in a trainable format basically and just
4:03:43 jam it into the machines
4:03:46 um but I don't know I don't know that I
4:03:48 I'm
4:03:49 I'm sure in the developer Community
4:03:51 there's there's all sorts of places to
4:03:54 learn that there's a guy if you're
4:03:55 really curious about that who's this
4:03:57 Prometheus Prometheus bound 25 there's a
4:04:00 guy on YouTube named David Shapiro who's
4:04:03 who's written five books on AI and he's
4:04:06 really smart and really good he probably
4:04:08 talks about this I'm sure he's got a
4:04:10 video that's that's on that
4:04:12 um but you could join some of his
4:04:14 projects
4:04:16 um his open source projects and and
4:04:18 start hanging out with him and learn
4:04:19 some stuff
4:04:23 oh smaller models
4:04:25 are going to replace the big General
4:04:26 large language models yeah I think
4:04:28 that's probably true
4:04:30 um we'll see we'll see if the that
4:04:32 Google engineer is right that said we
4:04:33 have no moat what is Bing and Poe so
4:04:36 Bing
4:04:38 is Microsoft's Bing it's their search
4:04:40 engine but there's a chat button there
4:04:41 if you hit that chat button that's gpt4
4:04:45 it's that's chat GPT and then Poe is
4:04:48 just like another GPT but you have a
4:04:51 bunch of different models you can choose
4:04:53 from there's Sage gpd4 Cloud plus Cloud
4:04:56 instant 100K Claude instant chat EBT so
4:05:00 um so that's what those are they're just
4:05:02 all just different variations on the
4:05:04 theme
4:05:06 rather more specialized models are going
4:05:08 to be oh are going to replace
4:05:10 general ones well I think we'll have
4:05:12 like the core models will be the basis
4:05:15 for all this and then yeah I think
4:05:16 you'll have all sorts of spec
4:05:18 specialized things and and like the
4:05:20 plugins are already starting to point to
4:05:23 that future where the plugins right now
4:05:25 are interfacing with existing websites
4:05:27 and existing logic but they're probably
4:05:29 just going to be interfaces to
4:05:31 specialized llm so I think you're
4:05:33 probably right I think a lot of that
4:05:35 detail though will drift into the
4:05:37 background like we as end consumers
4:05:39 won't need to know any of that [ __ ]
4:05:41 how do you keep up with a website
4:05:43 Twitter
4:05:44 and I've got I started a newsletter to
4:05:46 keep up with it so it's if you go to
4:05:48 Twitter and go to at everyday AI news
4:05:51 that's my
4:05:53 the link to subscribe to the newsletter
4:05:55 is there
4:05:57 um I just pay attention to it all the
4:05:59 time because I think it's really
4:06:00 important
4:06:01 I exported my Tick Tock stats as a CSV
4:06:04 interesting pasted in the chat GPT for
4:06:07 analysis for recommendations was amazing
4:06:09 how insightful the response was I think
4:06:11 that's going to be huge I think that
4:06:13 idea of
4:06:15 um
4:06:16 turning an agent loose on your entire
4:06:18 digital history
4:06:20 and then being able to interact with it
4:06:22 tell me about the major phases of my
4:06:25 life as evidenced by my my social media
4:06:29 footprint like holy [ __ ] that's going to
4:06:31 be intense
4:06:35 oh yeah I remember when I posted that
4:06:39 Microsoft has never been great in its
4:06:41 whole life Microsoft in the last six
4:06:43 months is
4:06:45 acting like a startup for the first time
4:06:47 since
4:06:49 they so so in the early days they were
4:06:52 sort of like a startup but but not
4:06:54 really like apple always sort of kicked
4:06:57 their ass on the start FB feel
4:07:00 um when they decided to go from shrink
4:07:03 wrap to internet
4:07:05 they when they decided to go they went
4:07:07 and they went strong
4:07:10 I have been a Microsoft Basher my whole
4:07:13 life I'm an apple Fanboy like deep deep
4:07:16 to the core and
4:07:19 the last six months what Microsoft has
4:07:21 done with chat GPT and [ __ ] with
4:07:24 Google has been really impressive and
4:07:27 and how well thought out
4:07:29 the Integrations are with chat GPT
4:07:32 I'm
4:07:34 I like I hate to admit it but they're
4:07:36 they are
4:07:38 um
4:07:39 they're kicking ass right now they might
4:07:42 [ __ ] it up but they're kicking ass right
4:07:43 now is there money in studying AI
4:07:46 I don't know I mean if if you want to go
4:07:48 work for
4:07:51 um you know open AI if you want to go
4:07:53 work for one of the big companies or one
4:07:55 of the big research firms
4:07:57 um
4:07:59 learning the technical side of it's fine
4:08:01 I mean you're entering a world that's
4:08:03 going to have a lot of competition
4:08:07 um and a lot of um
4:08:11 a lot of saturation
4:08:14 um if you're not super technical then
4:08:16 there's no really there's no not really
4:08:18 a place to study it there's no experts
4:08:21 right now right five years from now
4:08:23 there will be but right now it's like
4:08:26 all these tools are kind of disparate
4:08:28 they're all kind of janky some of them
4:08:30 are going to survive some aren't like
4:08:31 the the biggest skill right now is
4:08:33 curiosity and exploration
4:08:37 um unless you really want to get into
4:08:38 the machine learning side of it
4:08:40 um Google just just
4:08:42 um released a course on it
4:08:45 um I think it's on generative AI but
4:08:47 I've heard it's pretty good
4:08:49 um I don't know I listen I think I think
4:08:52 right now studying any aspect of AI is
4:08:56 not a bad thing
4:08:58 do I subscribe to any AI apps I pay for
4:09:01 chat GPT plus I pay for Poe
4:09:04 I pay for mid-journey 10 bucks
4:09:08 I pay for
4:09:10 Google collab for when I want to run
4:09:12 stable diffusion
4:09:16 um
4:09:21 I pay for zappier for the automations
4:09:23 that's about it
4:09:25 I I think most of these apps are going
4:09:27 to just be gone
4:09:29 most of the apps right now are like
4:09:31 money grabs to to try to take advantage
4:09:33 of [ __ ] but a lot of these are just
4:09:35 going to turn into features Within other
4:09:37 applications
4:09:40 is GPT for sheets through open AI though
4:09:42 worried about data privacy
4:09:46 um
4:09:47 you're using your own API key so
4:09:50 basically it's just a function that
4:09:52 calls their API and uses your own API
4:09:54 key so the the
4:09:57 the data privacy and security of that
4:10:00 connection is the thing that you're
4:10:02 trusting or not but you like it's an
4:10:04 engagement with open AI I
4:10:06 I suppose GPT for Sheets could be
4:10:09 skimming data out of there but I doubt
4:10:11 it I assume the API is some version of
4:10:14 secure I don't know enough about that
4:10:15 [ __ ] to know how how they would exploit
4:10:18 that I mean I suppose it's possible if
4:10:20 you're concerned about that just wait
4:10:21 for Microsoft like they're gonna like
4:10:23 copilot's gonna come out soon
4:10:26 six months maybe three
4:10:29 where's the best site for AI generated
4:10:31 video and artwork
4:10:33 um
4:10:34 Lexica is good for images
4:10:38 um
4:10:39 I don't know that there's any good sites
4:10:41 for video right now like Runway ml is
4:10:44 the one that's doing text to video right
4:10:46 now they're kind of the leaders in that
4:10:48 um but text to video and AI generated
4:10:51 videos really early right now so there's
4:10:53 not like all the examples I've seen are
4:10:55 sort of they're kind of e oh my God four
4:10:58 hours how do you do it I don't know I'm
4:10:59 I'm stupid at this point my ass is numb
4:11:01 I'm thirsty as [ __ ] I can't talk anymore
4:11:04 but I can't stop I just love you people
4:11:06 you're awesome
4:11:12 oh
4:11:15 should we ask how its day is first you
4:11:18 can
4:11:20 I I ask I tell it please I thank it
4:11:26 be friends with your little co-pilot why
4:11:28 not
4:11:29 I'm learning code now how will AI affect
4:11:31 this
4:11:32 a lot
4:11:33 field yeah so if you're learning to code
4:11:37 now
4:11:39 do a couple of things go to chat GPT and
4:11:42 just describe what you just coded and
4:11:45 see how it does compare its code to your
4:11:47 code I would say that would be a thing
4:11:50 then what I would do is copy your code
4:11:52 paste it in there and have it analyze
4:11:55 your code and tell you how it could be
4:11:56 improved made more efficient or if it
4:11:58 finds any bugs
4:12:00 um
4:12:02 learn to code
4:12:05 inside the knowledge that
4:12:07 the Tactical Act of coding
4:12:12 the majority of it's going to go away
4:12:15 but the act of
4:12:19 figuring out what you want to code and
4:12:21 doing the critical thinking for the code
4:12:23 gets more important
4:12:26 so if you're not using GitHub copilot go
4:12:28 start using GitHub copilot just start
4:12:31 learning what the tools are and how they
4:12:32 do what they do and how good they are
4:12:34 and how fast they are
4:12:36 and
4:12:39 you'll get depressed but then on the
4:12:41 other side of that is some sort of
4:12:43 freedom of like okay let me learn how to
4:12:44 use these [ __ ] tools because then you
4:12:46 can be a coder that also uses these
4:12:48 tools and you'll be in demand
4:12:51 because people are going to need coders
4:12:52 but they're going to need AI literate
4:12:54 coders
4:12:56 all right I appreciate you sharing
4:12:58 generative AI race means
4:13:01 um well so the generative AI race means
4:13:03 this
4:13:05 open AI launched chat GPT in November
4:13:08 30th 2022.
4:13:11 it got to 100 million users in six weeks
4:13:14 it was the fastest adoption of
4:13:15 technology in history it made every
4:13:17 other company freak the [ __ ] out like
4:13:20 Microsoft and openai were like poor
4:13:23 Google was like we're [ __ ] Meadow was
4:13:26 like we've got AI too
4:13:28 AWS was like uh we are strongly
4:13:31 considering Ai and you can use all these
4:13:33 other models and we definitely have one
4:13:35 we will announce at some point in the
4:13:37 near so the generative AI race is their
4:13:42 like the predictions of how big this
4:13:44 industry is going to be are just mind
4:13:47 watering
4:13:49 eye watering
4:13:52 mind shattering I don't know whatever
4:13:54 it's late
4:13:56 um
4:13:58 you know so the race is all these
4:14:02 companies trying to
4:14:04 you know
4:14:07 be the big players in the generative AI
4:14:10 game
4:14:12 and and what Sam Altman once is he wants
4:14:14 to get to AGI artificial general
4:14:16 intelligence first I don't know how they
4:14:19 Define that but that's what he wants
4:14:20 we're not there yet Quantum Computing
4:14:22 entering the AI race soon yeah I heard a
4:14:25 thing today that they're doubling
4:14:29 Quantum Computing I think they're up to
4:14:31 500 qubits in in some computer some chip
4:14:35 I saw today
4:14:36 and they're they're doubling the number
4:14:39 of qubits
4:14:41 um every year so they're up to like 500
4:14:44 now so if it's going up if it's if it's
4:14:46 doubling every year
4:14:49 um because because every every qubit is
4:14:51 an exponential
4:14:53 power increase so like 500 cubits is
4:14:56 like whatever power to the 500th so it's
4:15:00 insane I don't know [ __ ] about the
4:15:03 quantum Computing world I just know that
4:15:04 they're going to be able to do some
4:15:06 pretty serious uh processing with AI
4:15:09 stuff with it
4:15:13 leave it's been four hours
4:15:17 oh hi again people are coming back
4:15:19 they're like you're still here dude go
4:15:21 to bed
4:15:22 all right I'm gonna go to bed everyone
4:15:24 peace out I hope this was helpful look
4:15:26 even my laptop went to sleep
4:15:29 um I appreciate you all hanging out and
4:15:31 uh I'll see you whenever I'll probably
4:15:33 see you tomorrow
4:15:35 peace everybody