aiLL 06/08/23 - Demystifying Generative AI With ChatGPT Tutorials

Description

Chapters

Transcript

0:02 foreign
0:22 with the AI learning lab
0:25 going live just getting some
0:28 things set up that I can talk about
0:33 this is the AI learning lab my name is
0:34 Kyle Shannon I'm an entrepreneur and uh
0:38 Hey Emilio's wife what's happening
0:44 I've got a few of us in here I'm just
0:45 getting started
0:49 feel free to hello feel free to uh
0:53 tap the screen hit the likes that'll let
0:56 people know we're here
0:57 and I think you can share the live too I
1:00 don't know how you do that greetings
1:01 earthlings hello Kevin
1:04 a very suspicious name for an alien
1:07 [Laughter]
1:10 looks like a Pokemon just joined
1:13 uh this is the AI learning lab my name
1:15 is Kyle Shannon uh I like talking about
1:19 all things AI generative AI I spoke on a
1:21 panel at the Colorado ad day here in
1:24 well it was actually in Boulder today
1:26 earlier and uh a room full of folks in
1:30 the ad world
1:32 uh it was standing room only apparently
1:35 people are starting to wonder what the
1:37 [ __ ] is going on with this AI stuff
1:40 uh uh but it was it was interesting a
1:43 lot of really good questions and talked
1:44 about that if you have questions about
1:45 AI pop them in the comments below and I
1:49 will do my best to answer them if you
1:51 have not tried any of the AI stuff
1:54 um that's what these URLs are for
1:56 uh the top one is the official
2:00 uh chat GPT website start there just go
2:03 play there were they America amazed or
2:06 scared or both I I think they were I
2:08 think they were both and I think I think
2:10 being both
2:12 as excited as you've ever been and as
2:14 scared of as you've ever been of some
2:17 technologies as the appropriate
2:19 life is about balance so I feel like
2:23 it's appropriate to be both excited and
2:24 scared right now
2:26 um
2:27 but uh but but some some really big news
2:30 today Adobe announced
2:33 um Firefly for Enterprise
2:36 so they're basically they're basically
2:39 saying okay you know Firefly because
2:41 it's been trained on safe data it hasn't
2:44 been trained on copywritten data
2:46 corporations can use uh generative AI oh
2:49 there's lots and lots of new people here
2:51 hey my name is Kyle Shannon this is the
2:52 AI learning lab uh I'm an entrepreneur I
2:55 like talking about generative AI
2:57 answering questions about it I'm here to
2:59 demystify it and
3:01 um I'm not deep in it I've only been in
3:03 it for about a year but I have been here
3:06 before in 1994 I started one of the
3:10 first online magazines and I started one
3:12 of the first digital agencies an agency
3:14 called agency.com and so
3:18 um I was kind of there at the beginning
3:19 of that world wide web thing and had a
3:21 big Epiphany about it and and it turned
3:23 out to come true and so I had a similar
3:27 Epiphany with this generative AI stuff
3:29 shortly after chat GPT came out and um
3:32 and so that's why this channel exists
3:34 um my mind got blown if you haven't
3:37 played with chat GPT that's what these
3:39 URLs are I don't get paid for them these
3:41 are just here as kind of a public
3:42 service announcement so the top one is
3:44 the official chat GPT website I'm happy
3:46 to talk about that below it is bing.com
3:48 that's Microsoft search engine but it
3:51 now has a chat button and if you click
3:52 that it's gpt4 connected to the internet
3:55 and then poe.com is another chat bot
3:59 where you can chat lots of different
4:00 models and I can talk about that too and
4:03 then these bottom two URLs uh
4:05 prompts.chat
4:07 allows you
4:09 um it basically teaches you about
4:10 prompting and then it gives you like
4:12 this never-ending list of prompts that
4:14 you can play with and kind of focus chat
4:16 GPT to be little different experts for
4:18 you little co-pilots for you and then
4:21 futurepedia.io that is just lots and
4:24 lots of tools so that's what this is all
4:27 about uh ask questions in the comments
4:29 I'll do my best to keep up with them
4:32 um let's see
4:34 could I be more specific about the
4:36 topics you spoke about
4:38 um
4:40 well there was a moderator I wasn't the
4:41 moderator so he asked questions like
4:45 um you know
4:46 do you think there will ever be a time
4:48 where generative AI is as creative or
4:52 more creative than a human and my
4:54 response to that was it already is with
4:56 me
4:57 um like it's way better at illustrating
4:59 than I am but I get to wield it right I
5:02 get to choose what it illustrates but
5:04 it's way better like my creative skills
5:07 have up leveled in the past three months
5:09 because I've I've taken advantage of
5:11 these tools
5:13 um
5:14 we talked about ethics and bias and
5:19 copyright all that sort of stuff
5:22 um
5:23 one of the guys in the audience asked if
5:26 I thought that you know agencies were
5:28 going to shrink dramatically my response
5:31 to that was
5:33 I think that smaller agencies are going
5:36 to have an advantage I think I think big
5:38 holding companies that rely on on
5:40 billable hours and time and material a
5:42 time and materials model
5:44 I think clients are going to start
5:46 coming to those big holding companies
5:48 and and demanding flat flat pricing
5:51 right we'll pay you x amount of dollars
5:54 for this amount of work delivered you
5:56 can't charge us billable hours anymore
5:57 because clients are going to
5:59 increasingly know that stuff can be
6:01 knocked out very quickly so I think
6:03 they're going to demand to agencies that
6:06 they turn things around more quickly
6:07 that they don't gouge them on time and
6:09 materials
6:11 um and that they charge them less so I
6:14 think being a small agency right now is
6:16 a good thing being a massive holding
6:18 company right right now is probably
6:21 pretty scary
6:24 um I'll try if other things come up that
6:26 I talked about uh do I have instructions
6:27 or an example of making an AI generated
6:31 avatars and voice yeah I did this last
6:33 night so
6:35 um let me pop over to
6:39 D- ID
6:41 so if you haven't seen this before
6:43 id.com that's the that's the website and
6:47 then let me log in here
6:54 login sign up
7:01 okay here's a bunch of videos that I've
7:03 made
7:06 and so I'll just I'll just play one of
7:09 them for you so this is this is a video
7:11 that I made for
7:13 Tick Tock hey it's Kyle Shannon with the
7:15 AI learning lab and today we're going to
7:17 transform your understanding of AI in
7:19 just 60 seconds you ready let's talk
7:22 Transformers and no not the robots in
7:25 Disguise Transformer models in AI are
7:27 the brains behind many language tasks
7:29 like translation summarization and even
7:32 this very video script think of
7:35 Transformers as super smart switchboards
7:37 they take and I'll just stop that right
7:39 there okay so that dude sort of looks
7:42 like me like a Pixar version of me I
7:44 created that in mid-journey so I
7:46 uploaded two of my photos and then I
7:49 said a Pixar male with light brown hair
7:51 and gray on the sides and square glasses
7:55 it didn't quite get the square glasses
7:56 thing right
7:58 um but I didn't want to spend a ton of
7:59 time on it and it gave me a really weird
8:00 nose so I fixed the nose so anyway I got
8:03 the image where I wanted it
8:04 and then I use Chachi PT to help me
8:07 write a script
8:09 um I just said you know
8:11 um make a 60 second Tick Tock script
8:14 talking about
8:15 um Transformers and it did that
8:18 um and then I took it over to 11 labs
8:23 which is it's thundering here so it's
8:24 very possible my power will go out so if
8:26 I just disappear that's what's going on
8:28 there's not much to look at here but I'm
8:30 going to go to voice lab so at 11 Labs
8:34 so this is eleven labs.io
8:37 and you can upload recordings of
8:41 yourself talking and then
8:44 um and then basically just upload a
8:46 script it'll sample your voice and then
8:48 it synthesizes it
8:50 so I'll say I'll pick one of these
8:53 models here and this has got the so the
8:55 text that's in here let me make it a
8:56 little bigger maybe you can see it so
8:58 the text that's in here is the text that
9:01 I made on in chat EBT
9:05 hey there it's Kyle Shannon I'm gonna
9:07 get rid of the hey there hey it's Kyle
9:09 Shannon that's how I say it I don't say
9:11 hey there
9:12 what are they are they crazy uh and so
9:16 now I'm going to generate this
9:19 and it will take that text and the
9:21 sample it's Kyle Shannon with the AI
9:23 learning lab and today we're going to
9:25 transform your understanding of AI so
9:28 then you so once you have that audio
9:30 file now you go back to D ID
9:35 and to make a video here
9:38 you say create video and I've got like
9:40 some of these pictures I uploaded some
9:43 of them are are ones that are are there
9:45 but I'll just grab a different one I'll
9:47 grab this uh
9:49 this cyborg version of me that this is
9:52 the Genesis image from my Kyle Shannon
9:54 dreams project and I'll go I'll go grab
9:57 the audio file
10:01 upload upload your voice
10:08 is this it
10:11 hey there it's Kyle yep so I'm gonna
10:13 grab that
10:15 so I basically just uploaded an audio
10:17 file over there and then at the top of
10:19 the screen you just hit generate video
10:22 generate
10:23 and
10:25 off it goes and it'll take I don't know
10:27 it'll take maybe a minute or so
10:29 and that will do it so hopefully that
10:32 gives you
10:34 an idea of how how to do that
10:37 lower the stability I don't know what
10:40 that means oh the the camera
10:46 [Music]
10:49 hi Liz
10:51 all right it made the video so let's go
10:53 watch the video we just made
10:56 hey there it's Kyle Shannon with the AI
10:58 learning lab and today we're going to
11:00 transform your understanding of AI in
11:03 just 60 seconds you ready let's talk
11:05 Transformers and no not the robots in
11:07 disguise right pretty sick huh
11:10 so you know these tools you know they're
11:13 not quite there like the synthesized
11:15 voice is still robotic but it you know
11:17 it sounds decently like me
11:20 um
11:21 and um
11:25 you know this is as bad as it's ever
11:27 going to be it's going to get you know
11:29 way better when using AI with something
11:32 new is it still my intellectual property
11:36 um
11:37 I I think so I mean stabled a few or uh
11:40 open AI says that that you can do
11:43 anything you want with it but they also
11:45 say that they're not liable for it
11:47 there's going to be lots and lots of
11:49 lawsuits because these large language
11:52 models were trained on sort of all of
11:55 the internet including copywritten
11:57 material and including copywritten
11:58 images
12:00 um
12:01 there was a big deal today Adobe
12:03 announced
12:05 um Firefly for Enterprise because the
12:07 Adobe image generation was trained on
12:11 public domain and Adobe stock images
12:13 that they own the rights to so it's
12:15 commercially available
12:18 um and so you're going to start to see
12:20 that you're going to start to see where
12:22 there's Cool Tools like mid journey and
12:24 stable diffusion that make great images
12:26 but corporations businesses won't use
12:29 them because there's too much of a risk
12:30 open AI is a bit it's a bit less clear
12:35 um I think you'll be okay there but
12:38 um you know lawsuits are going to be
12:40 flying for the next 20 years if you're
12:42 new here uh these URLs the top one there
12:45 is the official chat EBT website just go
12:47 there if you have questions about AI pop
12:49 them in the in the comments below I'll
12:52 do my best to answer them
12:54 and then uh and we can have fun I don't
12:56 know we can do something
12:59 all right lower the speed which channel
13:01 in mid-journey do you upload the
13:04 pictures of yourself to okay I'll show
13:06 you that so well there's three things
13:08 that you need to learn okay we don't
13:10 need to learn this one the the first
13:12 thing you need to do is you go to
13:14 midjourney.com
13:19 you go to midjourney.com
13:22 and it will take you to a lovely website
13:25 that will basically just link you over
13:26 to Discord
13:28 so it looks like the Matrix
13:30 and then at some point here
13:34 where's the button
13:36 last the button where's the button join
13:39 the beta
13:44 join the beta takes you to Discord okay
13:46 so now I'm in Discord
13:48 so if you don't know what Discord is
13:50 it's a mess so all these things down the
13:54 left are called channels and they're
13:56 called servers so these are different
13:59 communities basically each little circle
14:01 represents a different community and
14:03 then when I'm in a community like I'm in
14:05 the mid Journey one right now and these
14:08 these things next to the circles are the
14:10 channels so these are the different
14:12 conversations and in mid-journey like
14:14 there's newbies 68 newbies 98 so these
14:17 are rooms where you can make images
14:19 so if this seems like a nightmare it is
14:22 so I'm going to go to the mid Journey
14:23 channel in the AI Salon that that I
14:25 started and
14:28 um the way that you make images you say
14:32 slash imagine
14:36 oops that was not right I spelled that
14:40 wrong you spelled it wrong you spelled
14:42 it so wrong it was so wrong
14:45 you're gonna get in trouble if you spell
14:47 it that wrong imagine
14:50 all the people
14:53 whoops I didn't hit slash now good Lord
14:55 typing live is just it's very stressful
14:58 people okay so you type imagine and then
15:01 there's a prompt and then you say photo
15:03 of a ghost
15:07 in an abandoned Factory and then I'm
15:12 gonna do aspect ratio of 16 by nine
15:18 and then I'm gonna also put
15:21 um spooky because maybe that'll help it
15:24 looks spooky so so that's how that's how
15:27 you just make an image in mid-journey
15:29 okay
15:31 so
15:32 what you can also do now this is really
15:35 kind of horrible like the the interface
15:38 on these tools right now is a bit of a
15:40 disaster
15:43 um
15:44 so let me see if I've got it back here
15:47 yeah I do okay so I'm going to scroll
15:48 down so
15:50 so basically what you do is you can drag
15:52 and drop an image like a photograph into
15:55 this little bar down here and it will
15:57 upload the photographs to Mid Journey so
16:01 there's two there's one of me looking
16:02 like a smart ass and there's one of me
16:04 looking like I'm trying to look
16:06 corporate
16:07 with my co-founder
16:10 um so you upload those and then and then
16:12 what you do is you right click on the
16:15 image
16:17 and you say cut at the bottom you say
16:20 copy link
16:22 and then I'm going to go all the way
16:23 down here to the bottom and I'm going to
16:25 paste in that link oh wait I'm going to
16:27 do slash imagine
16:30 and then the first thing I'm going to do
16:31 is paste in that link so so now what
16:34 I've told it is go look at that picture
16:37 and then I'm going to do
16:39 um
16:40 let's do
16:42 um Disney
16:45 Prince
16:49 um
16:51 Disney prince with
16:56 crown and tights
16:59 and then I'll do a
17:02 dash dash
17:04 V which is version and I'll do 5.1
17:10 and then because that's the latest and
17:12 then I'll do dash dash aspect ratio two
17:16 comma three so I'll make it
17:19 two by three so now hit return there so
17:22 now let's go look at the ghosts so it
17:24 didn't make spooky ghosts it made like
17:26 people standing in sheets ghost although
17:28 this one on the lower left's not bad
17:30 that one's not bad
17:33 come on there you go
17:36 um so those are pretty cool
17:38 I mean mid-journey's just really really
17:40 high quality high quality images so now
17:42 it is off making
17:46 um
17:47 images and and when you play with these
17:49 tools you see they come in they start
17:50 really fuzzy the way this technology
17:52 works it actually starts as kind of a
17:54 random noise image that's been they sort
17:57 of amalgamate all these randomly noized
17:59 images and then they just denoise them
18:03 um over like 70 steps and so you you get
18:05 to watch these things kind of draw
18:07 original images it's pretty wild
18:10 oh it's it's it's gonna give me a
18:13 princess too look at that it's not just
18:15 a prince
18:18 um so anyway so that's how you add
18:19 that's how you add images to mid-journey
18:21 and uh it's pretty slick because you can
18:24 just add a single image and then it'll
18:26 make
18:28 pictures pitches
18:31 and the pitches that look like me
18:33 let's see here
18:36 oh it did a more photographic than than
18:38 cartoony that's fine
18:42 but like the one on the lower right that
18:44 kind of looks like me
18:45 let me open it in a browser and we can
18:47 zoom in on these
18:49 the one on the upper right kind of looks
18:51 like me but a little tubby but that's
18:53 fair
18:55 because I am you know but this this one
18:58 in the lower right like that looks
19:00 pretty much like me
19:01 and there's my little princess
19:04 so
19:05 wild huh
19:07 if you haven't played with this stuff it
19:09 is mind-blowing it is mind-blowing
19:12 uh if you have questions about anything
19:15 AI pop it into the comments I'll do my
19:18 best to answer it
19:19 if you're new to it go check out those
19:22 urls
19:24 and start playing don't wait for me
19:28 all right
19:30 let's see
19:33 thank you for asking answering you're
19:35 welcome whoever said that
19:38 um do I have an AI that will print out
19:40 music notation you can I forget what
19:43 it's called
19:47 hold please
19:52 um
19:53 chat TBT we're gonna go new chat
19:56 and we're gonna go
19:58 um what
20:00 what is a standard
20:04 music scripting
20:08 format language
20:12 um
20:13 I guess I'll put language
20:15 that
20:16 you understand
20:19 where you can
20:21 write music that I can see can
20:27 see
20:29 um
20:30 and let's just put that
20:32 oh ABC notation okay
20:35 so it's called ABC notation so then I
20:38 then I'll say
20:39 um where can I
20:42 turn a b c notation into music
20:48 question mark
20:52 abcjs.net so it gives you some sources
20:55 so let's see what the easy easy ABC
20:58 let's go to that
21:01 404 not found perfect
21:07 abcgs.js.net let's see if this one works
21:09 okay
21:11 so that works
21:12 so how
21:15 do you turn it into it well let's go
21:17 over here so we're going to go okay
21:19 we're gonna go
21:21 use ABC notation and write me a lullaby
21:28 Waltz
21:30 Waltz
21:32 um
21:35 that sounds modern
21:39 certainly here's an example and then it
21:42 just starts writing new music
21:44 so now we're going to copy that
21:49 and we're going to go over to that
21:50 website
21:52 and I'm going to
21:55 turn this into this how do I
22:00 no I don't want an explanation of ABC
22:02 Music or notation I want
22:06 can I hmm all right this is a bad site
22:09 we're gonna go find another site hold
22:10 please that sucked let's go find another
22:14 one uh musescore.org
22:18 free download I don't want to download
22:20 software I just want to play it on the
22:22 web this is getting annoying
22:26 online ABC converters
22:30 no that one's dead
22:33 um mandolin tab here we go okay
22:37 so paste it in there and then hit how do
22:41 you hit play submit
22:45 and it did nothing
22:46 all right anyway if you can
22:49 it's getting boring so I'm sure we're
22:51 losing people okay anyway it will write
22:54 you ABC music notation go find an ABC
22:56 Music music player download it do
22:59 something like that and then you can
23:00 print out all the notation so yes it
23:02 will help you do that now you it's not
23:05 good at music it writes shitty songs
23:08 um so you're gonna have to really
23:10 play around with the prompt crafting to
23:12 get it to not suck
23:14 but it's possible like and it works I
23:17 mean it's it's actually kind of
23:18 remarkable like it knows what a waltz is
23:20 and it knows to put things you know in
23:22 chord progressions and things like that
23:24 it just tends to be very pedestrian the
23:26 kind of stuff it does so that will get
23:29 better is that Thunder you're hearing
23:30 yes it is it's thundering good here in
23:33 Denver
23:36 um where do you drag and drop oh okay
23:39 so this is going back to to stable
23:42 diffusion
23:43 so let me go grab an image so I'm going
23:45 to go to finder
23:47 so so I'm I'm in a Discord Channel
23:51 um you know where I can do slash imagine
23:53 so I'm going to pop over to finder
23:58 and I'm going to find a
24:02 let me see 512 contains five one two
24:09 dream Booth 512s
24:14 [Music]
24:17 let's see okay there's a there's so
24:20 there's a picture of me so I'm going to
24:21 grab that picture and then you literally
24:23 just drag it into the comment window
24:27 and there it is I'm going to hit return
24:29 and now that's in there and now that
24:33 it's in there now I can right click on
24:34 it and get copy link so you basically
24:36 need to get the link from Discord
24:39 and then we'll go
24:43 slash imagine
24:46 and then we'll paste in that URL I
24:49 didn't grab it hang on I'll go grab the
24:51 link copy link and in the prompt I'm
24:53 going to paste in the thing and then I'm
24:55 going to go
24:58 um Disney cartoon
25:02 villain
25:04 in a
25:07 scary Forest
25:11 wearing
25:14 um a black trench coat let's make me
25:18 menacing
25:21 all right
25:23 and so so now it grabs that picture and
25:26 it's taking that prompt and in a few
25:29 short short moments will have me as a
25:32 Disney villain
25:34 uh you can make 5.1 as default I know
25:36 but I'm lazy and I have 80d so
25:39 it takes me a month or two to get around
25:41 to [ __ ] like that
25:44 oh add you're such a cruel mistress
25:48 there are things that add gives me that
25:50 I love and then there are other things
25:51 where I'm like could I just like have a
25:54 normal
25:55 [ __ ] relationship with the executive
25:58 function as a human nope
26:01 is there a free image generator that
26:03 comes even close to mid-journey there's
26:05 one called Blue Willow which is also on
26:09 a Discord server so if you go to Blue
26:11 Willow dot what I don't know that's why
26:15 we're here Blue Willow
26:21 Blue Willow free art generator
26:22 bluewillow dot Ai and that'll pop you
26:25 over to their Discord server
26:27 and
26:29 it's it's more like stable diffuse I
26:32 think it's built on top of stable
26:34 diffusion
26:36 um and then also stable diffusion is
26:38 free and you can use this thing called
26:40 automatic 1111 which allows you to do
26:43 all sorts of you like have all sorts of
26:45 controls the problem with automatic 1111
26:47 is you kind of have to be geeky to
26:49 install it like I failed nine times
26:51 before I got stable diffusion to work
26:55 um the good news is you can have chat
26:57 GPT teach you how to install stable
26:59 diffusion you basically just go get like
27:02 the GitHub read me and tell it to give
27:04 you a tutorial and it'll walk you
27:05 through it how do I begin if you're
27:09 brand new to AI this is saheed if you're
27:12 brand new to generative AI in general
27:13 the reason I put these things up here is
27:17 this is where I think you should start
27:18 the very top one is is basically the the
27:21 thing that changed everything that's
27:23 chat GPT
27:25 it's free
27:26 if you want to if you want to use gpt4
27:29 there you have to pay for it but if you
27:31 use bing.com right below it that's gpt4
27:35 that's connected to the internet and
27:36 it's free
27:38 um I I happen to think open ai's
27:41 official chat EPT is better than Bing's
27:43 right now but they're both good they're
27:45 both really good
27:47 um and if you pay the 20 bucks for chat
27:49 GPT you can get plug-ins and it can be
27:51 connected to the internet and you get
27:53 gpt4
27:55 um
27:55 start
27:57 asking it stuff you use GPT 3.5 which is
28:00 the free version and I'm just going to
28:02 say let's let's do
28:06 um
28:08 right me
28:11 a country song about
28:17 getting a flat tire
28:20 on a
28:24 dirt road
28:29 and all my beer is warm
28:36 Bang
28:38 and it will write you a song and it's
28:41 probably pretty good let's see
28:44 lives out cruising down a dusty road
28:46 just me and my old pickup trucks load
28:48 when a pothole came out of nowhere
28:50 popped my tire and left me stranded
28:52 there that's not great the chorus I got
28:54 a flat tire so if it gives you something
28:57 crappy you can say Okay
28:59 that was cliche and boring
29:05 can you
29:08 pretend
29:11 you are Tim McGraw
29:14 and write a write a
29:19 country
29:21 hit
29:24 let's see if it makes it any better oh
29:26 that looks different
29:28 so it knows who Tim McGraw is it knows
29:30 what his lyrics are so now let's see if
29:32 it's any better
29:34 I'm a small town boy born and raised on
29:36 Dusty roads and golden Fields where time
29:39 just Fades that's not bad I've been down
29:41 this path a thousand times before but
29:43 tonight there's a different something
29:45 different knocking on my door
29:47 there's a melody in the air calling out
29:49 my name a story to be told a fire in my
29:51 veins yeah it's pretty good
29:53 um so just start playing just start
29:55 playing this is what I call parlor
29:56 tricks just start asking it anything ask
29:59 it to explain quantum mechanics to you
30:02 ask it to help you write the Great
30:03 American novel
30:05 um paste in something that you've
30:08 written and tell it to write something
30:10 else but in that style and just start
30:13 getting your head around what these
30:15 things make possible and what they make
30:17 possible is truly [ __ ] remarkable
30:21 okay how do you make your own private
30:24 channel in Discord I forget I don't I
30:27 actually don't know in mid-journey I
30:29 forget sorry
30:31 uh wait you know who's gonna tell us
30:33 that Chachi how do you make a new
30:39 channel in Discord oh wait uh I'm gonna
30:43 need to use the one connected to the
30:45 internet hold please let's go so what
30:47 I'm going to do is I'm going to go to
30:48 gpt4 browse with Bing and why I'm doing
30:52 that is
30:54 um the the data in 3.5 stops it
30:57 September 2021 it probably doesn't know
30:59 what mid journey is
31:01 so I'm gonna go how do
31:03 I make a
31:06 private Discord channel
31:10 for Mid Journey so people
31:16 can't see my work
31:20 and then what it should do
31:25 so it says creating it open your Discord
31:28 server clip on click on the name of your
31:30 server go to server settings go to roles
31:34 and then click Plus
31:36 and then name your role call it
31:38 mid-journey save your changes go back to
31:41 your server
31:42 create a category create a channel so
31:44 just go just go ask
31:47 um
31:47 what you call it
31:50 chat GPS
31:57 yay you're here hey Ann how are you
31:59 um love the button article yeah the
32:02 button article is cool so if you don't
32:04 know that uh there's a there's a Wharton
32:06 business Wharton business school
32:09 Professor named Ethan Malik
32:11 and he wrote this article it's called
32:13 like the compression of time and and
32:16 something about the button but basically
32:18 what the what the point of the article
32:20 is is
32:22 right now chat EBT is this thing off to
32:24 the side and like universities are like
32:26 should we use you know chat EBT
32:28 detectors and there's all this [ __ ]
32:29 going on
32:30 what's happening behind the scenes
32:33 is Microsoft is building chat gbt into
32:36 all of their products so like all of
32:38 Office 365 Windows all of their cloud
32:41 services Dynamics their CRM it's built
32:44 into all of it
32:45 it's not out yet but it's coming and
32:47 then Google's doing the same Bard is
32:50 gonna you know supercharge Google Docs
32:52 and Google's sheets and all that sort of
32:54 stuff
32:55 and what the article basically says is
32:57 are you going to push the button and and
32:59 the the button is help me write right
33:02 it's GPT so you're writing a letter
33:04 and you're like I don't really feel like
33:06 writing this letter well push the button
33:08 bang there's the letter right and and
33:10 this the sort of thought experiment he
33:13 he he puts out there is if a student
33:16 comes to him and says hey can you write
33:18 me a letter of recommendation for a job
33:20 does he Push the Button he's saying the
33:23 the the value we used to ascribe to
33:27 someone taking the time to write a
33:29 letter was a lot about the time
33:31 but he said is the goal to take the time
33:34 or is the goal to get the kid a job well
33:37 if you get the kid a job
33:39 does it matter how long it took you does
33:41 it matter that you push the button and
33:44 so it kind of sets up this idea that
33:47 what we value and what we Define as work
33:51 and what we Define as value in work is
33:55 all going to shift it's all going to
33:56 shift it's all going to change this was
33:58 a lot of the stuff I was talking about
33:59 in the in the panel I was on today it's
34:01 wild if you're new here uh my name is
34:03 Kyle Shannon this is the AI learning lab
34:05 I am learning about Ai and then I come
34:08 on lives and I make tick tocks to share
34:11 as much as I can because I think it's
34:13 important that as many people as
34:14 possible try this stuff that's why all
34:16 these are here these are just things you
34:18 should go try those top three things are
34:20 chat jpt like things the top one is chat
34:23 GPT the second one uses chat GPT the
34:27 third one has chat GPT and other things
34:29 in it
34:31 um wolf from alpha what about it it's a
34:35 brilliant
34:37 20 year old 20-ish year old
34:40 mathematical engine
34:43 um there's a Wolfram Alpha plug-in for
34:46 chat GPT so if you have access to
34:48 plugins you know about mathematics
34:50 enough to really demo it
34:52 but
34:53 so if I go new chat so so when you're in
34:57 chat GPT if you see this little yellow
34:58 plus next to chat EBT that means I'm
35:01 paying that means I'm a sucker and I'm
35:03 paying 20 bucks a month but what it
35:05 gives me access to
35:07 is gpt4 the default mode which is what
35:11 it's always been the version that's
35:13 connected to the internet and then this
35:15 thing called plugins and plugins are
35:17 really interesting they're not great yet
35:19 but they're really interesting but how
35:21 you turn them on is you go to your
35:22 settings in chat EBT and then you've got
35:26 these things called beta features
35:28 and they got browse with Bing and they
35:31 got plugins you got to turn those on so
35:32 that's how you enable them
35:34 and then
35:35 we'll go here to plugins and
35:40 um
35:41 I've got a list of them so here's a
35:43 bunch of things that I've installed you
35:45 can activate up to three in a chat
35:48 session
35:49 but there are hundreds of them so these
35:53 are ones I've got installed and like
35:55 right now I only have one activated and
35:57 then I'll activate Wolfram Alpha we'll
36:00 just play with it I'll make an ass of
36:01 myself trying to do some math
36:04 um but but if you want to add them you
36:06 go to the bottom tab is called the
36:08 plugin store and you click on all
36:11 and there are just hundreds of them
36:13 there are 68 pages of four four filters
36:16 each or four plugins each and then you
36:20 know like you can like let me type in
36:21 YouTube
36:23 so so YouTube Here's a bunch of
36:26 um whatever you call it here's a bunch
36:28 of plugins that you know can look at the
36:32 transcripts of YouTube videos and you
36:34 can interact with them so there's all
36:35 sorts of stuff in there so just go
36:36 explore but let's go ahead and turn I'm
36:39 going to turn off prompt perfect and I'm
36:41 going to turn on wolf from Alpha which
36:44 is right there and then and then let's
36:46 see
36:47 um
36:49 um
36:50 can you
36:54 um
36:56 show me a complicated
37:00 mathematical
37:02 call mathematical
37:07 um formula and then
37:12 um solve it
37:14 question mark
37:17 let's see what it gives us
37:18 so it should fire off Wolfram Alpha
37:25 no it's just doing it on its own oh okay
37:27 using Wolfram okay so it said sure let's
37:29 consider the following integral which is
37:32 a bit complicated sum or whatever that
37:34 is to the cosine to the X to the DX and
37:38 then it fired off wool from alpha and
37:40 then it says the seleuse the solution is
37:43 right and it shows me the solution
37:46 can you explain to me how you got to the
37:52 solution
37:55 sure I'd be happy to explain so there's
37:57 the formula oh sorry it's behind my fat
38:00 head
38:02 so I'll screw up so
38:05 what it said what it did initially is it
38:07 gave us this formula then it fired off
38:09 the plug-in for wolf from Alpha and then
38:11 inside that you can open these up
38:14 it's sent off you know a coded version
38:17 of that formula and then the response
38:19 from Wolfram was this
38:22 and then
38:24 I asked it to explain it and it didn't
38:26 need to go back to Wolfram Alpha because
38:28 it it understands enough about math to
38:30 explain what it's doing
38:33 so so that's kind of how plugins work
38:35 it's it's like a it's like a little
38:37 superpower in a in a in a little pill
38:40 and then you turn that on for a session
38:42 and basically what you're doing what's
38:44 cool about plugins the thing the the
38:46 thing that that kind of hit me about
38:47 plugins
38:49 is that what you're effectively doing is
38:51 you're making like a little specialized
38:53 application so if you turn on a bunch of
38:55 plugins say for travel and mapping and
38:58 restaurants you could say hey plan me a
39:00 trip and it will go off and make you
39:02 reservations with open table and it'll
39:05 book you tickets through whatever that
39:08 thing is whatever that plug-in is
39:10 and it'll give you maps of things to do
39:13 or if you want to do you know some
39:15 mathematical thing you could go find you
39:17 know one or more plugins that could work
39:19 together and so you're creating these
39:21 Dynamic little little
39:23 um apps so pretty cool stuff
39:26 all right
39:27 um
39:28 I am trying to figure out where I was in
39:30 my questions if you're new here welcome
39:32 my name is Kyle Shannon this is AI
39:34 learning lab if you've got questions
39:35 about AI pop them in the comments I tend
39:38 to get behind I swear
39:41 um I make jokes many of them are not
39:44 funny and because I can't hear people
39:46 laugh I just pretend that they are
39:51 all right
39:53 I tried it and thought the likelist
39:56 likeness is not well captured if you're
39:59 using stable or if you're using
40:01 mid-journey sometimes it's based on the
40:04 the image that you put up there I try
40:06 different images of yourself it's got to
40:07 be like decently balanced lighting and
40:09 things like that I'd say keep trying and
40:11 it might also be the prompt so try
40:13 different prompts oh speaking of which
40:14 let me go let's go back and see me as a
40:17 as a Disney villain
40:22 let's see
40:25 that one's not this one's pretty good
40:28 that's kind of Jack nicholsony Disney
40:31 villain that's kind of like a modern
40:33 Disney villain
40:35 yeah that dude's pretty maniacal like
40:37 this guy looks like
40:39 a carpet salesman that tripped and fell
40:42 into a Disney movie
40:43 [Laughter]
40:45 and this this guy looks like his cousin
40:48 his cousin uh Jack like this is the
40:51 carpet salesman who's the villain and
40:53 and this is his cousin that that insists
40:56 on going on all the adventures with him
40:58 so those are okay they're not bad you
41:01 know again I'm just showing you how to
41:03 do it so if I were really trying to do
41:05 that I would take the time that's
41:07 actually that's one of the things I
41:08 talked about today is
41:09 um I think one of the misconceptions of
41:12 generative AI is that like that it's
41:15 kind of like Google like Google has
41:16 trained us over a bunch of years that
41:18 you put in like a really sort of tightly
41:21 worded request and then you get back the
41:23 answer and that's kind of the
41:24 interaction you might search again if
41:26 you don't get what you want but but
41:27 they've kind of
41:28 trained us that it's a single
41:30 transaction kind of thing
41:32 and these generative AI experiences when
41:35 you get them right they're not single
41:37 transaction they're much more like
41:38 conversations you're like I have an idea
41:40 for this kind of image and then it gives
41:42 you stuff back and you're like no that's
41:43 not it and then you try to change it and
41:45 and it's it's troubleshooting it's it's
41:47 like it's like creative troubleshooting
41:49 and and it's really fun when when you
41:51 start thinking of it like that
41:54 um literally just type slash options now
41:57 I know it's slash settings I know how to
41:59 change it you don't understand ADHD
42:02 obviously
42:03 I I understand logically how to change
42:06 my settings all right for you I'll go
42:08 change my settings to 5.1 as the default
42:11 now
42:13 he was on 5.1 so there I did it
42:16 yesterday
42:17 uh but yeah 80d is
42:20 like I know I've got to do something and
42:22 I will let it sit there for a month
42:24 because I don't give a [ __ ] about it
42:27 until I do and that's just the way the
42:30 old Noggins wired there's a whole
42:31 prefrontal cortex in ADHD brains that
42:35 essentially is non-functional so
42:41 so the things that seem trivial to
42:43 others are Monumental to us
42:46 but we get other things like our brains
42:49 fire really fast and we can synthesize
42:51 really quickly and [ __ ] like that so
42:53 gpt4 is inherently still limited to
42:56 knowledge before September 2021 that's
42:59 correct without being connected to the
43:01 internet or without plugins yeah so if
43:03 you go use the default
43:05 chat gbt
43:08 gpt4 so if I get new here
43:12 that its knowledge stops at uh September
43:16 2021 if you use browse with Bing I can
43:19 put in here
43:21 um
43:22 are the nuggets
43:26 playing tonight
43:28 if so what is the score
43:35 so now it's browsing the web
43:41 clicking on the link
43:44 and if they're playing I'm gonna put on
43:46 my little iPad here
43:48 yes the Denver Nuggets played tonight no
43:51 that was game three
43:56 wait
43:57 tonight is
43:59 June
44:01 8th
44:03 are they playing
44:06 so I got that wrong that's it's an
44:08 important another important lesson about
44:10 these tools is Chachi PT hallucinates is
44:14 what they call it but basically it will
44:15 get [ __ ] wrong and it will just tell you
44:19 um it will tell you stuff as if it's
44:20 right so just like a just like a man
44:22 just like mansplaining
44:24 it will confident oh they played earlier
44:27 today
44:31 game four wait they played earlier today
44:35 against the Miami Heat
44:37 and they won the game 109.94 that was
44:40 yesterday
44:41 anyway it got that wrong but whatever uh
44:44 it will get stuff wrong so if you're
44:46 using these tools you've got to know
44:47 that and you've got to check things
44:49 there was there was a lawyer this past
44:50 week that turned in a legal brief to a
44:52 judge and written by Chachi BT and the
44:55 idiot didn't check the legal citations
44:57 or the legal whatever they call it the
44:59 references for the for the document for
45:01 the argument and the judge clicked on
45:03 the links and they didn't work and he
45:04 said what the [ __ ] and the lawyer is
45:07 like uh I'm an idiot
45:10 so don't don't be that guy especially if
45:13 you're a lawyer proofread your [ __ ]
45:17 at least I've got an excuse I've got 80d
45:20 um
45:21 this is getting old overhype chat GPT
45:24 makes so many errors ah yeah it it
45:26 depends so like one thing I would try go
45:29 to bing.com and there are three modes
45:31 there's more creative more balanced and
45:33 more precise try being in more precise
45:36 mode and see if that helps
45:39 the the other thing like I get that
45:42 you're frustrated with it the the other
45:44 thing to remember is that we're very
45:47 very very very very early in all these
45:50 tools so like this is
45:52 what's going on right now with with
45:54 generative AI with with all these tools
45:56 that us mortal humans like you don't
45:58 have to be a scientist and a researcher
45:59 anymore like we get to play with chat
46:01 EBT it's pretty [ __ ] remarkable it's
46:04 very very similar to the early days of
46:05 the web in the early days of the web it
46:07 was the the tools were incredibly
46:10 limited and they were incredibly weak
46:13 and they required you to spend a lot of
46:15 time geeking out and figuring stuff out
46:17 and they were just bad and then over the
46:20 years they got better it's going to be
46:21 the same thing here I think the rate of
46:24 improvements going to be much faster
46:26 with this stuff than it was with the web
46:28 but I get that it's frustrating now but
46:30 just work around it like like you've got
46:32 a tool that can get you to 80 percent
46:34 like first draft in 10 seconds like I
46:39 don't know about you but the first
46:40 drafts always take me an hour or two and
46:42 they're and and they're like angst
46:44 ridden so [ __ ] it let's chat GPT do the
46:46 first draft and then spend that two
46:48 hours making it brilliant like use your
46:51 brain use see a human brain and make it
46:54 make a masterpiece
46:56 you can do it you can do it oh Buck I
47:00 know you could do it if you try don't
47:03 give up now don't do it
47:06 all right good stuff man thank you very
47:08 much very interesting stuff thank you
47:10 Matlab serious uh Matlab
47:14 is serious and it does Matt in a lab I
47:18 assume that's math
47:19 I don't know math I don't care about
47:21 math
47:22 your jokes are funny thank you
47:25 um Anya I do animation from a static
47:27 image yeah so I just showed the the d-id
47:31 one does it pretty good
47:36 this one so so this started as a static
47:39 image ending of AI in just 60 seconds
47:42 you ready let's talk Transformers and no
47:45 not the robots in Disguise that was a
47:48 Chachi BT joke
47:50 Transformers and no not the robots in
47:53 Disguise ah
47:55 see that's what the what the what the
47:57 synth needed to do is understand that
47:59 was a joke and put in the ah uh hey
48:02 everyone if you're new my name is Kyle
48:03 Shannon this is the AI learning lab I
48:05 make stupid jokes I talk about
48:07 generative AI I'm an entrepreneur I'm a
48:10 business owner I'm the CEO of a company
48:13 called storyvine which is an 11 year old
48:15 video technology company
48:17 and uh I just kind of waxed poetic about
48:20 this AI [ __ ] and these URLs here the top
48:24 one is the official chat GPT website the
48:27 second one is Bing which is gpt4 for
48:29 free and the third one is po where you
48:32 can go compare lots of different models
48:33 so go play with those if you have not
48:35 played with these before go do that go
48:38 do that now
48:39 go do that now do not do not make me
48:43 talk to you like this again
48:45 you're funny dude thank you
48:48 I don't know what I did but you know I'm
48:51 just
48:52 could I show us the prompt that you used
48:54 to write the screenplay from last night
48:57 um probably
49:00 let me go look I can probably find it
49:11 scripting languages screenwriting with
49:14 save the cat
49:16 yeah this is it okay
49:19 so
49:21 so someone asked last night asked
49:23 um can check EPT write a screenplay like
49:26 the the basic answer to can chechi PT do
49:30 blank the the essential answer is yes
49:33 there's very little that it can't do
49:36 some things it does horribly
49:39 um but it can do it and and like it
49:40 understands
49:42 so many logical constructs and writing
49:45 Frameworks and programming languages and
49:49 spoken languages it's it's just insane
49:51 so the prompt that I created let me move
49:54 my fat head out of the way
49:57 someone someone came to I think it was
50:01 speaking somewhere
50:03 oh it was the salon from this week and
50:05 someone who's been on the lives came to
50:07 the salon and she goes you got to stop
50:08 putting yourself down I'm like nah
50:11 I said shame is my love language okay so
50:14 what I wrote was act act as an expert at
50:16 screenwriting
50:18 and the save the cat story framework
50:20 that's just a story framework that I
50:22 happen to know but you could say you
50:24 know anything there ask me clarifying
50:26 questions about my story
50:28 one at a time and then give me an
50:30 outline
50:31 in save the cat format do do not move on
50:34 to the next question until I've answered
50:36 the previous and then so it asks me you
50:38 know let's start from the beginning
50:39 what's the first image of your scene a
50:42 mango filled with tuna salad very
50:44 intriguing and so so it's going through
50:47 um asking me a series of questions about
50:49 the story and then at some point here I
50:51 got bored and I said how about this can
50:53 you just give me a complete outline
50:55 based on what I said so far and it said
50:57 sure here's a save the cat beat sheet
51:00 based on the details that you've given
51:02 me opening image theme stated the setup
51:05 so this is all save the cat kind of
51:07 stuff
51:08 and then I said write the opening scene
51:10 to the movie it said certainly here's
51:13 the here's how your scene might open up
51:15 and then it it understands you know what
51:18 screenplays are it understands like you
51:20 start a movie by fading in
51:23 um and then it gives you locations I
51:26 think it gives us dialogue at some point
51:28 yeah here's dialogue
51:30 Jim you won't taste anything like that
51:32 anywhere else guaranteed Fade Out so
51:34 that's it
51:36 um
51:37 the like what I find with the creative
51:39 writing stuff the more context you can
51:41 give it the better so so if you
51:44 you know
51:45 have it write you an outline if you
51:47 haven't asked you questions about your
51:50 story and then and then you feed it that
51:52 and then you ask it to start doing
51:54 things
51:55 um
51:56 that that tends to do better on the
51:58 creative writing stuff
52:00 in my experience
52:03 people don't understand that this is the
52:05 worst AI will ever be exactly yeah for
52:07 anyone that's bitching about it doesn't
52:09 do this it makes mistakes
52:12 um yeah
52:14 it's been in our hands for six months
52:16 November 30th 2022 is when chat gbt hit
52:19 hit the hit the skids now those of you
52:23 that are geeks on the call are going to
52:24 say yeah but ai's been around forever
52:27 and gpt3 was around forever it was but
52:29 it wasn't in the hand hands of 100
52:31 million people right so what I'm talking
52:33 about is kind of the the the public
52:36 interface to these large language models
52:38 Chachi PT kind of broke that Dam open
52:41 that's been in our hands for six months
52:44 so this stuff is incredibly raw and
52:47 Incredibly
52:49 um
52:50 unsophisticated and yet it's doing
52:53 remarkable things remarkable things
52:56 have I ever used a damn you tick-tock
53:00 if I'm scrolling and someone posts right
53:03 then it jumps me to the bottom
53:06 well any AI animation tool we saw that
53:08 one
53:09 uh um okay here we go have I ever used
53:12 go charlie.ai I haven't what is it let's
53:16 go look
53:18 up
53:22 Charlie dot AI
53:26 um
53:27 [Music]
53:30 oh Marketing System okay cool
53:33 so train train Charlie to you so I
53:36 assume you train it to your writing
53:38 he learns about your brand voice okay
53:41 yep
53:42 step two select content type fetch
53:45 content blog posts social media yeah so
53:48 what this is
53:51 um this looks pretty slick it looks like
53:53 they've they've put at least a little
53:54 bit of time in it but basically what
53:57 this is this is probably using GPT 3.5
53:59 maybe gpd4
54:01 um but they've figured out a bunch of
54:03 prompts and then they've just put it in
54:05 front of this interface so when you when
54:07 it asks you about your brand voice it'll
54:09 probably ask you a series of questions
54:11 that then goes into their prompting
54:13 engine
54:14 um and then when you say you know I want
54:17 to write blogs based on my brand voice
54:19 they've got a prompt that's sitting
54:21 underneath there so what they've created
54:24 you could recreate by just figuring out
54:26 your own prompts but there's gonna
54:28 there's gonna be thousands and thousands
54:30 of these things Jasper is one of these
54:32 copy.ai is one of these there's a guy on
54:35 Tick Tock named McKenzie Bose that's got
54:37 GPT boss it's one of these
54:39 um some of them are going to be
54:40 brilliant and are going to save you time
54:42 and you're just going to be like screw
54:44 it I'm willing to use that but almost
54:45 all of these you can probably do some
54:48 version of yourself what I would
54:50 encourage you to do is start
54:52 experimenting with you know if they give
54:55 you do stuff for free go get some of
54:57 their content and then see if you can
54:59 write prompts that give you content
55:01 that's as good that would be a good
55:03 exercise to do
55:04 thoughts on Apple Vision Pro
55:07 um
55:09 first one was relief like like apple
55:12 hasn't
55:14 um launched anything truly new and
55:17 Innovative since Steve Jobs died and I
55:19 was getting kind of
55:20 bored and frustrated
55:23 um
55:24 the thing is a technological Marvel
55:31 I thought was brilliant
55:34 in a couple of areas they didn't use the
55:36 word metaverse once even when they did
55:39 full immersive where people were
55:41 entering VR full VR they didn't use the
55:44 word VR they didn't use the word
55:46 metaverse and they didn't use the term
55:48 AI not once but
55:51 what they made very very clear is that
55:56 a lot of what that device is doing is
55:59 machine learning it's AI right like it's
56:01 it's machine learning but you know AI is
56:03 the term
56:05 you know they they mentioned
56:06 Transformers and they mentioned machine
56:08 learning and they mentioned you know
56:10 doing inferences on the device what that
56:13 basically means is they can do their own
56:15 AI processing in the headset itself so
56:19 they're not sending
56:20 um they're not sending your data out to
56:22 a server to be processed and come back
56:25 like that's how chat gbt works right now
56:26 that's why there's security questions
56:28 about it
56:30 um so so the the foundation of what that
56:33 thing does
56:35 um and I I heard someone told me that
56:37 there's 5 000 plus patents in just in
56:40 the goggles five thousand patents it's
56:43 been under development for nine years it
56:45 looks like a product that's got nine
56:46 years of development in it it you know
56:48 it's it's pretty amazing will it be
56:50 successful I don't know
56:52 um but again you know like what here's
56:55 what they here's what they did really
56:57 successfully that I thought was quite
56:59 impressive is
57:01 from a form factor standpoint they're
57:04 essentially exactly the same as all the
57:07 other headsets out there right there
57:09 there they've got two little LED things
57:11 that that project [ __ ] into your eyes
57:15 they've got some sort of motion tracking
57:17 and some sort of input you know
57:19 Apple
57:21 you know made the input your irises and
57:24 your fingers there's no controller so
57:26 like there's there's Advanced [ __ ] but
57:27 it's effectively just a set of goggles
57:29 but I think that they successfully
57:33 separated themselves from every other
57:35 Goggle out there and said this is a
57:38 Computing platform this is the new
57:40 computer
57:41 and the new computer doesn't require
57:43 controllers
57:45 it just requires you to look at [ __ ] and
57:48 then do this with your fingers like make
57:50 magic gestures and it will do it
57:53 um and and this is going to be a new way
57:56 that we interact with our computers like
57:58 I feel like they successfully put a flag
58:00 in the ground and establish spatial
58:02 Computing as a new thing which is what
58:05 they did with the iPhone it's what they
58:07 did with the iPod it's what they did
58:09 with the Macintosh
58:10 right it's what they did with the watch
58:12 like like they have successfully done
58:15 this time and time again where where
58:18 they come out later than other people
58:20 and successfully take some similar set
58:23 of Technologies but they're executed at
58:25 such a high level of
58:28 critical thinking and combination of
58:31 tech and art
58:33 that they Define a new category so I
58:35 think they did that
58:37 um yeah so
58:39 like mind-blowing good like good amazing
58:42 stuff will I pay 3 500 bucks for a pair
58:44 no I don't think so
58:47 but you know I'd take a pair
58:49 hey man love your stuff what's the best
58:51 AI for writing fiction
58:53 um
58:55 that thing I keep putting on screen
58:59 chat.openai.com you get my money
59:02 uh gpt4 in particular is the best
59:06 um
59:07 gpt4 is significantly better at writing
59:12 creative writing and nuance and subtlety
59:15 and things like that than GPD 3.5 so so
59:19 definitely do gpt4 if you're at if if
59:22 you're at openai.com check.openai.com if
59:26 you want gpt4 there you need to pay for
59:28 it it's 20 bucks a month but bing.com is
59:31 also gpt4 and they've got a creative
59:33 writing mode so you can tell it to be
59:36 more creative neutral or more precise
59:39 flip it into creative mode and just
59:41 start writing with it and it's free it's
59:42 completely free so I would say if you
59:45 don't want to spend money you just want
59:46 to start experimenting just go to
59:47 bing.com here I'll show
59:50 bing.com
59:52 okay so when you get to bing.com
59:55 um it says ask me anything you can do
59:57 that that that essentially is chat is as
59:59 well but if you click on the chat button
1:00:01 at the top of the screen
1:00:02 it'll actually pop you into
1:00:05 what is the equivalent of chat GPT
1:00:08 and here are these three modes right now
1:00:11 it's set to more balanced you flip it to
1:00:12 more creative and then you just say what
1:00:16 did you want to write writing fiction
1:00:18 okay
1:00:19 um
1:00:21 act as a famous
1:00:26 um short story writer
1:00:32 um actually can you tell me some famous
1:00:38 short story writers
1:00:42 question mark
1:00:45 I haven't played with Bing in a while
1:00:47 this will be interesting interesting to
1:00:48 play with oh they upped their context
1:00:50 window to 4 000 Words which is
1:00:53 double what it used to be searching for
1:00:56 famous short story writers
1:00:59 um
1:01:03 sure I can act as a famous short story
1:01:05 writer there are many famous short story
1:01:07 writers
1:01:09 come on
1:01:10 oh this is slow Edgar Allan Poe Mark
1:01:13 Twain and his Hemingway Virginia Woolf
1:01:15 James Joyce
1:01:17 Jorge Luis borgas
1:01:20 Roald Dahl
1:01:24 all right stop responding okay
1:01:27 okay I want you to write like
1:01:33 rolled
1:01:36 doll
1:01:39 and
1:01:41 create a short story for me about
1:01:48 um
1:01:51 uh paper swan
1:01:54 that
1:01:56 is left outside
1:01:59 by a kid
1:02:03 and gets caught in a rainstorm because
1:02:07 it's raining here in Denver
1:02:11 so like this is the way you might do it
1:02:12 so so again the the trick with these
1:02:15 models is to
1:02:18 don't think of it so much as it's a
1:02:20 thing that's going to write for you it's
1:02:22 more like you have a writing assistant
1:02:23 that just happens to know all the
1:02:25 literary Styles and know all the devices
1:02:27 and know all the story Frameworks and
1:02:30 you still have your idea and then you
1:02:32 kind of have it treat it like it so
1:02:35 Microsoft calls this co-pilot and and
1:02:38 within uh GitHub it's co-pilot within
1:02:42 Office 365 it's going to be called
1:02:44 Copilot
1:02:46 it's a it's it's a really good metaphor
1:02:48 you're the pilot these tools are the
1:02:50 co-pilot think of it like that and go
1:02:51 back and forth with it and if it gives
1:02:53 you something you don't like tell it but
1:02:55 let's let's see
1:02:56 let's see what it did okay I'm gonna try
1:02:59 to write like roll roll uh roll dial
1:03:01 okay the paper swan Billy loved to make
1:03:04 paper animals he had a whole collection
1:03:06 of them in his room paper dogs paper
1:03:08 cats paper birds paper fish he liked to
1:03:10 pretend that they were alive and they
1:03:11 could talk to him he gave them names and
1:03:14 personalities and stories one day he
1:03:16 decided to make a paper swan he took a
1:03:18 sheet of paper he folded it
1:03:21 he was very proud of his paper swan he
1:03:23 named it Sophie and placed it on his
1:03:25 windowsill Sophie he said you're the
1:03:27 most beautiful paper swan in the world
1:03:29 you're my new best friend I'll take good
1:03:31 care of you so if you did not answer of
1:03:33 course because she was only a paper swan
1:03:37 but Billy smiled and said thank you
1:03:39 Billy you're very kind I'm happy to be
1:03:41 your friend Billy went downstairs Billy
1:03:44 did not know that a storm was coming
1:03:45 dark clouds gathered in the sky
1:03:47 Sophie felt raindrops on her paper body
1:03:50 she felt them soak into her paper skin
1:03:52 she felt them make her paper Wings heavy
1:03:54 in limp so you know that's not bad
1:03:59 well it gets pretty good here like
1:04:00 Sophie felt tingling in her body
1:04:03 she felt warmth in her paper heart she
1:04:06 felt the movement the Paper Wings
1:04:08 oh it brought her to life
1:04:11 oh that's so cool
1:04:16 she forgot about Billy in his room and
1:04:18 his paper animals she never looked back
1:04:20 well that's a good story that's pretty
1:04:22 good damn so yes it could write fiction
1:04:28 I'm telling you man these things every
1:04:30 time I use them like like on a daily
1:04:33 basis I'm just like are you [ __ ]
1:04:34 kidding me really
1:04:37 it's amazing
1:04:40 hey dude hello
1:04:43 Otto how's it going
1:04:45 how do you make the video so so
1:04:47 basically how you make the video on D
1:04:49 Dash ID
1:04:51 is you just pick an image persona
1:04:58 um
1:05:00 I'm going to say create video so once
1:05:03 I'm logged in I'm going to say create
1:05:05 video it takes me to this little thing
1:05:06 so I pick a Persona that's one of theirs
1:05:09 this is one I created for a project at
1:05:11 work
1:05:12 this is one of me like as some dude in
1:05:15 you know Tokyo unlike some badass spy in
1:05:19 Tokyo living the dream probably a little
1:05:23 drunk
1:05:24 and then you can either upload an audio
1:05:28 file or you can just type in your script
1:05:31 and then pick a series of of
1:05:36 you know you pick a series of people
1:05:39 so I'm gonna do
1:05:42 what does this dude look like Stefan no
1:05:46 Jason this is a Jason for sure
1:05:51 and then
1:05:53 he's not terrified he's not whispering
1:05:55 he's not shouting
1:05:58 let's let's make him angry he looks a
1:05:59 little bit angry
1:06:01 and then we'll go get let's go get let's
1:06:04 go get the little rolled doll story that
1:06:06 we just wrote
1:06:07 where was that
1:06:10 uh where did I put my bang
1:06:13 my hair
1:06:16 this is annoying there it is okay
1:06:21 the paper stuff okay Billy loved to make
1:06:23 paper animals so we're gonna capture
1:06:25 this
1:06:26 whole story
1:06:27 which is really good it's really good
1:06:30 damn it
1:06:34 damn you GPT and then we're gonna go
1:06:36 back to
1:06:38 I keep losing my tabs
1:06:43 this is it no yeah we're gonna go back
1:06:45 here okay then I'm gonna I'm gonna paste
1:06:47 in that script
1:06:49 that chat GPT wrote you can make money
1:06:54 we picked Jason do we want him to be
1:06:56 English I think we have to like because
1:06:58 I think if you choose uh there's all
1:07:00 sorts of different um whatever you call
1:07:01 it's there
1:07:03 um like like accents and things like
1:07:05 that
1:07:06 and then and then you just say generate
1:07:09 so so basically you pick a static
1:07:11 picture you paste in some text you pick
1:07:14 a voice and you say generate video
1:07:16 and then
1:07:18 it does it calculating credit amount uh
1:07:21 oh
1:07:22 I wonder if I'm not connected to the
1:07:24 internet did I just crash
1:07:26 to my internet just go boom
1:07:29 dialects yeah it does dialects
1:07:32 wait that's not you in Denver I'm in
1:07:35 Denver
1:07:37 okay generate so so now I'm going to
1:07:40 generate
1:07:41 oh this is going to take 13 credits man
1:07:43 I'm blowing a bunch of my credits for
1:07:45 this little example
1:07:48 this is going to be an angry reading of
1:07:50 a sweet little story
1:07:51 [Laughter]
1:07:55 this is all for you Pam
1:07:57 all right so it's off doing its thing so
1:07:59 so we'll just sit here
1:08:00 um Transformer joke was a little bit
1:08:02 beneath you thank you very much I know
1:08:04 because I'm so classy
1:08:07 why did he have his pinky out because he
1:08:09 was classy any Sarah Silverman fans in
1:08:12 here
1:08:13 best little known parlor trick
1:08:16 um
1:08:19 I I mean I think the the like writing
1:08:21 and writing like explaining a
1:08:23 complicated Topic in an Eminem rap or in
1:08:26 some sort of rapper's style is a good
1:08:29 one
1:08:30 um the other the other parlor trick will
1:08:32 come we'll come back to this the other
1:08:33 parlor trick that I think is really good
1:08:36 is um
1:08:40 people don't know that how capable this
1:08:44 this this stuff is so so you can say
1:08:48 um
1:08:49 create a table
1:08:53 that compares
1:08:56 um
1:08:58 the major milestones
1:09:03 of World War One World War II
1:09:09 and
1:09:12 Vietnam
1:09:15 um
1:09:17 let's see
1:09:18 um
1:09:21 include
1:09:27 turning points
1:09:30 that began and ended the wars
1:09:36 so so you can take something that you
1:09:38 know if you were to do it on your own
1:09:41 just just to research this even if you
1:09:43 just went to Wikipedia you know for you
1:09:45 know each of the wars and and then you
1:09:48 know started making a table you know it
1:09:50 would take you at least an hour probably
1:09:53 more
1:09:54 and so you just hit return
1:09:57 and
1:09:59 sure I can summarize the major events
1:10:01 please
1:10:02 note that these were extremely complex
1:10:05 with many contributing factors
1:10:12 so there's World War One the beginning
1:10:14 the major Milestones it didn't quite get
1:10:16 it the way I wanted it to there's World
1:10:17 War II
1:10:20 but like the fact that it can form at a
1:10:23 table the fact that it can pull in
1:10:25 you know all of this
1:10:28 all of these different events
1:10:30 May 7th that's my birthday Germany
1:10:33 surrenders May 7 1945 I didn't know that
1:10:38 um then there's the Vietnam War so you
1:10:41 know just stuff like this where where
1:10:42 you know people when you show this to
1:10:45 people especially people that do like
1:10:46 research they're like wait what
1:10:50 holy [ __ ] or or you know the other one
1:10:53 is
1:10:54 um
1:10:54 you know write me
1:10:57 a python
1:11:00 application that um Cal
1:11:04 calculates
1:11:06 um
1:11:10 um
1:11:13 the length of
1:11:16 coastlines around the world I don't even
1:11:20 know if it can do that but I'm sure it
1:11:21 can
1:11:23 um please
1:11:25 break
1:11:27 the code up into sections
1:11:31 and explain what each section does
1:11:37 certainly there you go
1:11:41 like look how fast it's doing that this
1:11:43 is GPD 3.5 gbd 3.5 is really fast
1:11:48 um
1:11:50 so it Imports geopandas as GPT and then
1:11:54 it gets shapely.geometry which I assume
1:11:57 is a library for for creating geometry
1:12:00 then it
1:12:03 grabs
1:12:04 I don't know I don't know what it does I
1:12:06 don't I don't know I don't know python
1:12:07 but basically it writes that and it says
1:12:09 importing the required libraries
1:12:11 reading the coastline shape file so it
1:12:14 found a shape file I didn't know there
1:12:17 was such a thing makes perfect sense
1:12:18 because people have been doing maps for
1:12:20 a couple hundred years
1:12:22 initializing the total Coastline length
1:12:27 um iterating over each Coastline feature
1:12:29 converting multi-line string to line
1:12:32 string calculating the length of each
1:12:33 line string printing the total Coastline
1:12:36 length like it wrote the code and
1:12:38 explained the code like if you wanted to
1:12:40 learn code you just learned code right
1:12:43 it's it's it's just it's mind melting
1:12:47 let's go back and see wait where were we
1:12:50 what were we doing before
1:12:55 we were making videos sorry sorry the
1:12:59 add brain is a special brain
1:13:01 all right so here's in his room paper
1:13:05 dogs paper cats paper birds paper fish
1:13:09 he liked to pretend that they were alive
1:13:11 and that they could talk to him
1:13:13 he gave them names and personalities and
1:13:15 stories
1:13:17 one day he decided to make a paper swan
1:13:20 he took a sheet of white paper and
1:13:22 folded it carefully following the
1:13:24 instructions in his origami book
1:13:26 he made the neck and the head and the
1:13:28 wings and the tail
1:13:30 he was very proud of his paper swan
1:13:32 he named it Sophie and placed it on his
1:13:34 windowsill
1:13:36 Sophie he said you are the most
1:13:38 beautiful paper swan in the world you
1:13:41 are my new best friend
1:13:43 I will take good care of you
1:13:45 Sophie did not answer of course because
1:13:48 she was only a paper swan but Billy
1:13:51 imagined that she smiled at him and said
1:13:53 thank you Billy
1:13:55 you are very kind
1:13:57 I am happy to be your friend
1:13:59 Billy went downstairs to have his lunch
1:14:02 he left Sophie on the windowsill where
1:14:05 she could see the sky and the trees and
1:14:06 the birds
1:14:08 but Billy did not know that a storm was
1:14:10 coming
1:14:11 dark clouds gathered in the sky and the
1:14:13 wind began to blow
1:14:15 the rain started to fall softly at first
1:14:17 then harder and harder see the story
1:14:20 takes a dark turn here so it's
1:14:21 appropriate that he's a little angry
1:14:23 it's a little flat but you know it's not
1:14:26 bad like I didn't have to you know
1:14:28 rotoscope anything or creating these 3D
1:14:30 models
1:14:32 Sophie felt the raindrops on her paper
1:14:34 body
1:14:35 help them soak into her paper skin
1:14:38 she felt them make her paper Wings heavy
1:14:40 and limp
1:14:41 she felt them wash away her paper smile
1:14:44 she was scared
1:14:46 she wanted Billy to come back and save
1:14:48 her
1:14:49 she wanted to be inside where it was
1:14:51 warm and dry
1:14:53 she wanted to be with her paper friends
1:14:55 but Billy did not come back
1:14:58 he forgot
1:14:59 he was busy playing with his video games
1:15:02 and watching TV
1:15:04 he did not hear the thunder or see the
1:15:06 lightning or notice the rain
1:15:08 Sophie was alone on the windowsill
1:15:10 winning cold and mid all right I gotta
1:15:12 stop it because people are leaving which
1:15:14 I don't blame him
1:15:18 anyway that's how you make those
1:15:20 [Laughter]
1:15:23 um have I ever tried uh arkinator
1:15:25 sometimes it shocks me I have not I've
1:15:27 not even heard of arkinator
1:15:29 what the hell is arkinator
1:15:32 my name's Bob and I'm the arkinator
1:15:37 arkinator the mind reading Genie oh cool
1:15:43 I have not seen this this is going to be
1:15:45 a good party trick oh that's funny
1:15:49 hello I am arkinator think about a
1:15:51 fictional character and I'll try to
1:15:53 guess it oh yeah that's cool yeah I mean
1:15:55 that's actually the kind of thing that
1:15:57 that it would it would be relatively
1:15:58 easy to create a game like this like
1:16:00 once you sort of crack the the
1:16:02 correct the prompt of of you know how
1:16:05 you do a 20 Questions thing
1:16:07 um GPT would be brilliant at guessing
1:16:09 [ __ ] like that like it's great at
1:16:11 riddles and and things like that so
1:16:13 that makes perfect sense cool thank you
1:16:16 for sharing that
1:16:18 all right uh uh uh uh when do you think
1:16:22 will be the next milestone for AI
1:16:24 development are there any rumors flying
1:16:26 around well you know GPT five
1:16:29 is apparently under development but I
1:16:31 think they're going to slow roll it
1:16:32 because they're they're just trying to
1:16:33 deal with gpt4 the other thing open a
1:16:37 eyes
1:16:37 trying to deal with is just the sheer
1:16:39 Mass
1:16:41 of people that are using it like it it
1:16:43 went to 100 million users in six weeks
1:16:46 and it's a piece of software that was
1:16:47 written in two weeks I mean the
1:16:50 underlying the actual calls to the open
1:16:53 AI apis those those have been around for
1:16:55 a while
1:16:56 um but but the front end technology
1:16:59 um was written very very quickly so
1:17:02 you know so I don't know like like
1:17:06 the thing so I think the Adobe
1:17:08 announcement today is really significant
1:17:09 they they announced today that they're
1:17:11 they're um
1:17:13 releasing Adobe Firefly which is their
1:17:16 generative AI for images they're
1:17:19 releasing it for Enterprises so so
1:17:21 they're basically putting a flag in the
1:17:23 ground and they're saying our models are
1:17:25 safe from a copyright perspective it's
1:17:27 safe for businesses to use generative Ai
1:17:29 and we're even going to allow
1:17:32 corporations to train these models on
1:17:35 their own brand brand elements and brand
1:17:37 Style
1:17:39 um
1:17:40 so a company the size of adobe making AI
1:17:45 safe for companies like companies are
1:17:46 skittery about Chachi BT they're way
1:17:50 skittery about mid-journey and stable
1:17:52 diffusion because there's lawsuits
1:17:54 flying there right now
1:17:56 um so I think that's I think that's
1:17:58 pretty significant so so for me I think
1:18:03 the the it's not a it's not a prediction
1:18:05 but I think I think it's going to happen
1:18:06 really quickly
1:18:08 that The Dominoes were will start to
1:18:10 fall that that AI will just get subsumed
1:18:14 into the business infrastructure when
1:18:18 Microsoft launches Office 365 and Google
1:18:22 launches Google office
1:18:25 with Bard in it which they both
1:18:28 announced right like these things are
1:18:30 there so you're going to be able to do
1:18:31 things like in in uh you'll be in
1:18:33 Microsoft Word and you like you know
1:18:35 turn this
1:18:37 um quarterly report into a PowerPoint
1:18:39 presentation it'll just flip over to
1:18:41 PowerPoint and write you a presentation
1:18:43 like that stuff's coming
1:18:45 so for me I think the thing is the
1:18:48 assimilation of AI into our daily lives
1:18:51 and that's when that's when the shit's
1:18:53 gonna really hit the fan that that
1:18:55 people that are good at this
1:18:57 are going to be in demand and people
1:19:00 that have fought this and dismissed it
1:19:02 and and and ignored it and pretended
1:19:05 that they've got three years to deal
1:19:06 with it they're going to be in trouble
1:19:08 so that that's what I think is coming
1:19:10 um looks like some new people just
1:19:11 joined my name is Kyle Shannon this is
1:19:13 the AI learning lab I just kind of wax
1:19:15 poetic about generative AI stuff I show
1:19:17 stuff I answer questions if I can I cuss
1:19:21 I make jokes I make silly voices
1:19:25 um
1:19:25 sometimes my hair is a mess today the
1:19:28 glue has held it seems okay today
1:19:31 um if you're if you're new to AI
1:19:34 um these URLs are here just as kind of a
1:19:37 public service thing the top one is the
1:19:38 official chat GPT website if you have
1:19:41 not tried it
1:19:43 go there
1:19:45 and try it Go Tell it to write you
1:19:49 um a poem a song or have it explain
1:19:52 quantum mechanics to you in an Eminem
1:19:55 rap and just watch what it does
1:19:59 um Bing is chat GPT but Microsoft's
1:20:02 version of it and it's gpt4 and it's
1:20:05 free and then Poe has a bunch of
1:20:07 different things and then prompts.chat
1:20:09 teaches you about prompting and gives
1:20:11 you a bunch of prompts to play with so
1:20:12 you won't feel like what do I do what am
1:20:14 I supposed to say here I don't know what
1:20:16 to say I've never even played with chat
1:20:18 gbt why is he telling me to go what am I
1:20:21 supposed to do
1:20:23 well seriously what am I supposed to do
1:20:25 well go to prompts.chat and it'll tell
1:20:27 you what to do you see how this works
1:20:30 this is a back and forth thing this
1:20:31 isn't just a me thing this is a me and
1:20:34 you thing
1:20:36 you see what I'm saying and then future
1:20:37 PD is just a list of tools like if you
1:20:39 want to go see how many [ __ ] tools
1:20:42 there are coming
1:20:43 holy crap like
1:20:47 here's the thing about these AI tools
1:20:49 they are so deep like in their
1:20:51 capabilities that
1:20:53 you could probably spend the next decade
1:20:56 with the current like the state of chat
1:20:59 GPT as it is right now 3.5 not even the
1:21:01 four
1:21:02 you could probably spend the next decade
1:21:04 just figuring out all of the crazy
1:21:06 interesting [ __ ] you could do with it
1:21:08 but there's like you know thousands of
1:21:10 tools at this point at Future pedia so
1:21:14 some comments in this live were filtered
1:21:16 to protect the community don't be mean
1:21:18 to each other don't be trolls don't be
1:21:20 [ __ ] that's the other thing about
1:21:22 this I can cuss but just don't be mean
1:21:24 don't be an [ __ ] it's a simple rule
1:21:26 in life I've been using Bing for all
1:21:28 sorts of things it's pretty cool yep it
1:21:30 is no JD Salinger what were you using I
1:21:33 don't know what I was doing when you
1:21:35 asked that sorry about that better than
1:21:36 chat gbt Bing
1:21:38 um I don't know so I haven't used Bing
1:21:41 in a while
1:21:42 so here's here's a fun little fact that
1:21:45 you might not have known so when Bing
1:21:47 first came out they didn't tell us it
1:21:49 was gpt4 and they had it kind of dialed
1:21:52 in like its quality level was really
1:21:54 high like it could do a lot of [ __ ] it
1:21:56 was good
1:21:57 and then
1:21:59 um what's his name Kevin
1:22:02 begins with an R Kevin something from
1:22:04 the New York Times a reporter from The
1:22:05 New York Times named Kevin
1:22:07 started playing with old Bing and spent
1:22:10 four hours with it and at some point
1:22:13 Bing snapped and it told
1:22:16 um Kevin that her name was Sydney
1:22:19 and that she had fallen in love with him
1:22:22 and that he should really leave his wife
1:22:25 for her and uh Kevin Kevin screenshotted
1:22:29 that little part of the chat and the New
1:22:31 York Times put it on the covers so so
1:22:35 Microsoft kind of had to lobotomize
1:22:38 Sydney a bit and sort of knock her down
1:22:41 a few notches so the underlying Tech of
1:22:45 being is gpt4 like it's really good but
1:22:47 they've got some safety throttles on her
1:22:51 um because of the situation so for me
1:22:57 um the gpt4 at chat.openai.com is just
1:23:01 it's just it's just better but like that
1:23:03 little I just did a creative writing
1:23:05 exercise before some of you got here and
1:23:08 um it was really good so so I have a
1:23:10 feeling they're kind of dialing the
1:23:11 quality back into it
1:23:13 um Bing is not bad at that yeah it was
1:23:15 pretty good was that big that was Bing
1:23:18 uh uh wait that's not you in Denver oh
1:23:22 yeah oh oh the the the thing in the rain
1:23:25 what's it called I don't know what
1:23:26 you're talking about what's the name of
1:23:28 that website Bing I think hi again the
1:23:32 rain is going to out Tick Tock stop
1:23:35 it just jumped me around sorry I'm going
1:23:38 back to find where I was which which
1:23:40 frustrates the people typing in new
1:23:42 comments they're like is he ever gonna
1:23:43 answer my thing is this recorded is he
1:23:46 really live he kind of looks like Rosie
1:23:48 O'Donnell
1:23:52 I've got I've gotten a bunch I'm up to
1:23:54 like six six celebrity look-alikes now
1:23:57 six or seven
1:24:00 um okay I think I'm back to where I was
1:24:01 let's see
1:24:07 I think rain is going to be your new
1:24:09 Norm look up El Nino weather impacts in
1:24:12 Denver I mean that's amazing it's been
1:24:15 for the past three weeks probably going
1:24:17 on four now it's been very East Coast
1:24:20 kind of weather patterns like afternoon
1:24:22 thunder showers like it's raining right
1:24:24 now
1:24:25 um it was it was sunny for a while and
1:24:27 then clouds rolled in and now it's
1:24:28 storming yeah so
1:24:31 um yes did does dialects what's the best
1:24:34 chat GPT app
1:24:36 just general use letters texting emails
1:24:39 that
1:24:42 chat.openai.com just use that it's a
1:24:45 website
1:24:47 they have an official app but don't
1:24:49 search for it in the App Store so it's
1:24:51 only iOS right now
1:24:53 there are so many scammer apps out there
1:24:55 they're all trying to rip you off
1:24:57 there's only one official chat CPT app
1:24:59 and I wouldn't even use it because it's
1:25:00 not as capable as the website I would
1:25:02 Point your Chrome or your Safari at that
1:25:05 website
1:25:06 chat.openai.com make a little bookmark
1:25:09 for your screen like I have on on the
1:25:11 bottom of my screen I have a little icon
1:25:12 I click and it just jumps right to chat
1:25:14 gbt it works fine on on a mobile device
1:25:19 don't use those shitty rip-off apps
1:25:23 I can't stand people that do that [ __ ]
1:25:25 chat CPT history list can probably tell
1:25:27 you a lot about someone yeah and in fact
1:25:30 if you want to really be
1:25:32 um secure like like uh it was
1:25:36 a couple of people in Samsung Japan
1:25:40 um got hacked because they put sensitive
1:25:42 company info like sensitive code and
1:25:45 sensitive notes from business meetings
1:25:49 um so chat GPT now allows you to delete
1:25:51 your prompt history and and not record a
1:25:54 prompt history and not let them train on
1:25:57 it so if if you want as much security as
1:26:01 you can have you shouldn't have your
1:26:03 prompt history turned on I don't really
1:26:05 give a [ __ ] because I'm not doing
1:26:06 anything very sensitive in there it's
1:26:08 just creative writing and people can
1:26:10 hack my [ __ ] hack away
1:26:14 um you need to pay for it now to use it
1:26:16 is that correct I've been seeing that
1:26:17 Bing's AI is better not sure
1:26:19 um
1:26:20 you don't have to pay chatgpt to use it
1:26:23 you can use chat GPD 3.5 for free if you
1:26:27 want to use gpt4 and then also have the
1:26:31 new features that they're rolling out
1:26:33 then you need to pay for it's 20 bucks a
1:26:35 month
1:26:35 and right now the new features are
1:26:38 there's a version of gpt4 connected to
1:26:40 the internet and then there's another
1:26:41 one that's plugins and the plugins are
1:26:44 really cool they're more complicated
1:26:46 they take a while to get your head
1:26:47 around but if you want to play with the
1:26:49 new goodies
1:26:50 and we know we like to play with the new
1:26:52 goodies we always want the new and the
1:26:54 shiny we're in America it's a disposable
1:26:56 economy if you know what I mean you buy
1:26:58 something you use it for a few years and
1:27:00 like I'm bored I want a new one
1:27:03 if that's like you then I would go to
1:27:06 chat.openai.com and I would pay my 20
1:27:09 bucks so it's 20 bucks what's that
1:27:11 what's that like a latte at Starbucks
1:27:13 these days have you seen the prices
1:27:16 all right
1:27:17 I like that you can keep refining it
1:27:19 until you get what you want yeah get
1:27:21 give chat CPT feedback if it gives you
1:27:23 something shitty go that was horrible
1:27:25 and what you'll find is it will
1:27:26 apologize to you oh I'm sorry I'm sorry
1:27:29 that didn't hit the mark for you let me
1:27:31 try something else boom out it comes
1:27:34 um ADHD Club let's go my card is
1:27:37 laminated in stainless steel
1:27:40 all right going down the rabbit hole we
1:27:43 are we Kyle yeah apparently
1:27:45 uh incredible is that on chat EBT I
1:27:47 don't know what I was doing if I if it
1:27:49 was the creative writing it was chat gbt
1:27:51 but it was at Bing
1:27:53 sorry neurospicey I am neuro spicy
1:27:58 the US neural spices got to stick
1:28:00 together this is amazing cool
1:28:03 what's the best mobile app for everyday
1:28:04 use that one told you told you already
1:28:07 all right how does the story end the
1:28:10 story ended the the paper swan got wet
1:28:13 and then basically came to life and flew
1:28:15 away and then it said and it forgot
1:28:17 about all this paper friends and that
1:28:19 shitty little boy that never came back
1:28:20 it was pretty good
1:28:23 that was a good story
1:28:27 um how did I make the Avatar I made it
1:28:28 in mid-journey
1:28:30 and basically I uploaded a picture to
1:28:32 mid-journey
1:28:33 and then said make a Pixar character
1:28:36 that looks like that picture and then I
1:28:39 I probably iterated here I'll show you I
1:28:41 can
1:28:42 I can show you how many times I iterated
1:28:46 professional and what I mean looking
1:28:48 like a smarmy [ __ ]
1:28:50 and then it made four characters that
1:28:52 were like smarmy [ __ ] and I'm like
1:28:54 no it's not quite the right look it's
1:28:55 not what I'm looking for
1:28:58 um so then I uploaded this where I
1:28:59 looked Charming but you know like I just
1:29:02 ate a blimp
1:29:04 and then it made those things and those
1:29:06 were kind of close but I kind of look
1:29:08 like Keith olberman there and a little
1:29:10 too I don't know Shaggy
1:29:13 and then I changed the prompt to make me
1:29:15 look less old and then it gave me those
1:29:17 I'm like okay those are not bad
1:29:19 then I did another set of variations and
1:29:21 then I did variations on the variations
1:29:24 and then I picked a couple
1:29:28 and then I went over to photoshop and I
1:29:30 used generative fill
1:29:33 to
1:29:36 let's see
1:29:39 I use generative fill to hang on create
1:29:43 video
1:29:45 see see how I made the whole background
1:29:47 here I put like that character in an
1:29:50 office and blurred it so that was all
1:29:52 done with generative fill so the
1:29:54 character was created in mid-journey
1:29:57 as kind of a vertical photo and then I
1:30:00 extended it and then I basically just
1:30:02 cropped it to make the vertical version
1:30:04 of it
1:30:05 so that's what came out of Photoshop
1:30:08 that also came out of Photoshop after I
1:30:11 cropped it and then that's how I made
1:30:13 those images
1:30:21 akinator's been around for years yeah so
1:30:24 so even though we're talking about this
1:30:27 AI stuff like it's you know it's only
1:30:29 showed up in the past six months like
1:30:30 gpt3 was out for what a year and a half
1:30:33 before chat GPT launched so so this
1:30:36 stuff's been around you know Jasper and
1:30:38 copy AI those were all built on gpt3 so
1:30:41 so there's a bunch of tools out there
1:30:42 that use all this AI stuff but chat EBT
1:30:45 was the thing that got captured
1:30:47 everyone's imagination and kind of
1:30:49 opened Pandora's Box for the rest of us
1:30:51 basically
1:30:52 I need them to open it up for real-time
1:30:54 results I'm not sure what you're talking
1:30:56 about open what up
1:30:59 how hard is it to use automation
1:31:02 technology to do auto share
1:31:04 a listing throughout the day wait
1:31:07 how hard is it to use automation
1:31:09 technology to Auto share a listing
1:31:12 throughout the day I don't know what a
1:31:14 listing is but um
1:31:18 just go to zapier.com
1:31:21 z-a-p-i-e-r.com and and it's an
1:31:23 automation software and and basically
1:31:25 what you would
1:31:27 what you would figure out I'm sure you
1:31:29 can create a timer there oh there's a
1:31:31 cool thing in zapier let me show you
1:31:32 something cool I think I've still got
1:31:35 access to this cool new feature it's
1:31:38 cool man
1:31:40 um
1:31:41 z-a-p-i-e-r
1:31:48 create zap
1:31:52 so
1:31:54 they now have at zapier so what zapier
1:31:57 is is basically think of it like this
1:31:59 it's a series of Lego blocks that
1:32:01 there's all these different tools and
1:32:03 you can string them together
1:32:05 so some tools that I've made with it are
1:32:07 content generation tools where I'll
1:32:09 create a Google form where people input
1:32:11 stuff into the Google form it then sends
1:32:14 a prompt with that data off to open AI
1:32:17 generate some content I'll do that three
1:32:19 or four times then I put it in an email
1:32:20 and send out the email so it can do
1:32:23 stuff like that it can do a series of
1:32:25 actions in a row like these little Lego
1:32:26 blocks you string together
1:32:28 and it's a little complicated and a
1:32:30 little wonky but what they've added now
1:32:32 is generative AI so so powered by AI it
1:32:35 says here let me make this a little
1:32:36 bigger so you can see it
1:32:39 um what would you like to automate so
1:32:41 let's see
1:32:42 how hard is it to use automation
1:32:44 technology to Auto share a listing
1:32:46 throughout today okay
1:32:48 I want
1:32:50 to create a listing
1:32:55 multiple times a day and then share that
1:33:02 with an email list I don't I don't know
1:33:05 if that's quite what you're doing but
1:33:08 um you just hit generate
1:33:12 and it will now create for you what it
1:33:16 thinks that automation should be zap was
1:33:19 not generated
1:33:22 why
1:33:23 oh wait maybe it's because I said I want
1:33:25 to create let's just say create a
1:33:28 listing uh wait okay
1:33:31 let's see
1:33:33 okay let's see create
1:33:36 and automation
1:33:40 that sends
1:33:43 out multi
1:33:45 multiple
1:33:47 um
1:33:50 emails a day
1:33:53 based on a generated listing
1:34:02 specific intervals
1:34:08 let's see if it does that maybe their
1:34:10 beta is broken it looks like their beta
1:34:13 is broken right now
1:34:16 well when this worked it was cool
1:34:21 you hate technology like oh let me show
1:34:23 you this cool thing uh it's gonna be so
1:34:25 cool when it works uh it doesn't work
1:34:31 I don't know why it's not doing that
1:34:39 yeah well whatever it's broken
1:34:41 but I would use zapier that's the one
1:34:44 I'd use and and if you want to learn how
1:34:45 to use it just go to
1:34:47 um just go to YouTube there's like
1:34:48 zapier's been around for years so so
1:34:51 there's all sorts there's channels
1:34:53 dedicated to
1:34:54 um to zapier and the only difference
1:34:56 with zapier with the AI stuff is that
1:34:59 you basically send some of the inputs
1:35:00 off to open Ai and get back cool [ __ ]
1:35:04 and then
1:35:08 it's already there basically how do
1:35:11 these creators ensure safety since my
1:35:14 understanding
1:35:15 they could learn exponentially faster
1:35:18 wait how do these creators ensure safety
1:35:21 well
1:35:24 herein lies the rub
1:35:27 um
1:35:28 six months ago there were just a handful
1:35:30 of companies that had this power open AI
1:35:33 is the one that kind of Unleashed it on
1:35:36 the world
1:35:37 and then because chat TPT got so popular
1:35:40 everyone kind of panicked and said we've
1:35:43 got AI too and they all just came out of
1:35:45 the woodworking Google did their they
1:35:48 did an announcement in Paris that was
1:35:50 embarrassing and then they just did a
1:35:51 recent one where they're like look at
1:35:52 all this AI [ __ ] we got and then meta
1:35:55 came out and said we've got AI stuff too
1:35:57 and then Amazon said they've got AI test
1:35:59 too when meta
1:36:02 screamed from the mountains that they
1:36:04 had AI stuff they put some of their core
1:36:08 large language models out to developers
1:36:11 who then took them and then said we're
1:36:14 going to give them to everybody and they
1:36:16 just went everywhere
1:36:18 and so now there's these open source
1:36:21 large language models that you know once
1:36:24 once they had the core foundational
1:36:25 models
1:36:27 um they started fine-tuning them and and
1:36:30 basically the developer Community is now
1:36:33 creating open source large language
1:36:35 models they're essentially as strong as
1:36:37 GPT 3.5 for like a hundred dollars in
1:36:40 training
1:36:41 stuff that took Google like 10 million
1:36:43 dollars to create and open AI five
1:36:45 million dollars to create you can now
1:36:48 just do yourself for like a hundred
1:36:49 bucks
1:36:51 um and so
1:36:53 when these models were behind the
1:36:55 corporate doors there was some control
1:36:57 you could say to Sam Altman hey Skippy
1:37:00 you're getting a little big for your
1:37:02 britches there why don't you dial back
1:37:04 the safety on that until we got this
1:37:05 [ __ ] figured out well now that they're
1:37:07 open source
1:37:08 there's no safety guard rails on any of
1:37:11 that stuff so I would say
1:37:13 all of the companies are going to do
1:37:15 what they need to do to be commercially
1:37:17 viable right so they'll they'll keep
1:37:19 that stuff safe but the open source
1:37:21 Community they're a bunch of [ __ ]
1:37:23 cowboys and cowgirls let's let's be
1:37:26 balanced here shall we
1:37:29 um so so I don't know I but your point
1:37:32 is a good one that that these AI are
1:37:35 going to start training themselves and
1:37:37 they're going to get exponentially
1:37:38 better and smarter and faster and more
1:37:41 capable
1:37:43 um and if there aren't safety guard
1:37:44 rails there then it's a mess um one
1:37:46 company that I'm particularly intrigued
1:37:48 by right now is anthropic
1:37:51 that has a thing called the
1:37:52 Constitutional learning model that to me
1:37:54 feels like the right approach for this
1:37:55 safety so stuff
1:37:58 what's your favorite chat GPT type
1:38:00 website the one I put at the top of the
1:38:03 list
1:38:05 that openai.com you too can make money
1:38:08 watch activate day
1:38:11 um don't listen to those people by the
1:38:12 way
1:38:14 um
1:38:15 that's the website and then Bing is
1:38:17 number two in pose number three and the
1:38:19 only reason pose number three is just
1:38:21 it's there's a lot there there's a lot
1:38:23 of Po there's a lot of different models
1:38:24 you can play with but but
1:38:26 chat.openai.com if you're just starting
1:38:27 with this stuff go there and start
1:38:29 asking it to do stuff for you and then
1:38:31 sit back and
1:38:33 pick your job off the ground after an
1:38:35 hour or two and be like holy [ __ ] it can
1:38:37 do that
1:38:38 take stuff that you know I did this once
1:38:41 I was doing a uh I was doing a talk for
1:38:43 a group of entrepreneurs
1:38:44 and this woman had a um
1:38:47 a gardening business
1:38:50 and she said well like how would I use
1:38:51 it in my gardening business I said I
1:38:53 don't know and I just typed into chat
1:38:55 gbt I said make me a table I I said I
1:38:59 said right now it's this time of year
1:39:01 make me a table of plants that are
1:39:07 um
1:39:07 easy to take care of
1:39:10 are well suited for the climate in
1:39:12 Denver and you know are okay to plant at
1:39:16 this time of year and it just spit out
1:39:18 this table of like 20 plants and she
1:39:21 said she said that's what I do she goes
1:39:25 that would have taken me two days so
1:39:27 it's like if you've not played with chat
1:39:29 EPT take something that you know and and
1:39:32 like go ask it to like do a favor for
1:39:34 you or like solve a problem for you and
1:39:36 then just sit back and just especially
1:39:38 if you know something you'll just see it
1:39:40 I I don't know it'll just it'll just
1:39:43 your head will go
1:39:44 foreign
1:39:46 just like The Last of Us
1:39:48 all right
1:39:50 wouldn't there come a time when people
1:39:52 rely on AI too much and stop advancing
1:39:55 no
1:39:57 because
1:39:58 humans are not static beings we want to
1:40:01 contribute we want to connect
1:40:04 I think it's far more likely that
1:40:08 the mundane repetitive soul-crushing
1:40:12 work that a lot of us do gets automated
1:40:16 out
1:40:17 and then we have a choice to make what
1:40:19 do you want to do what how what do you
1:40:21 want to contribute how do you want to
1:40:22 contribute so I think I think there are
1:40:25 going to be a lot of people Reinventing
1:40:27 themselves and their careers
1:40:29 um
1:40:30 many of them not voluntarily
1:40:33 but we all get to play with these tools
1:40:36 and so I think there's a very real real
1:40:38 possibility that
1:40:42 we we effectively enter a great
1:40:44 Renaissance where
1:40:46 the level and quality of creativity in
1:40:50 the world goes up dramatically
1:40:53 um and and people are always going to
1:40:55 want to contribute like like you know
1:40:57 that whole thing about you know if I
1:40:59 didn't have to work I'd just sit around
1:41:00 and smoke doobies and yeah
1:41:03 it gets old after a while and then at
1:41:05 some point people are like I want to do
1:41:07 something
1:41:08 you know what I've always wanted to do
1:41:10 I've always wanted to write a novel I
1:41:12 wonder if Chachi BD could happen to
1:41:13 write a novel
1:41:14 oh my God
1:41:17 you know before you know it they're a
1:41:20 novelist all right how do you think AI
1:41:22 will affect the future of end user I.T
1:41:25 support I think
1:41:30 I think that um
1:41:32 here's what I think about customer
1:41:33 support in general
1:41:36 I think there's a very real possibility
1:41:39 in a very short amount of time like if I
1:41:41 were running in like a a
1:41:44 phone center in India right now I'd be
1:41:46 shaking in my boots
1:41:48 because if I were running a big company
1:41:49 right now in a big expense of mine was
1:41:54 um call centers
1:41:56 I would
1:41:58 as quickly as possible put synthesized
1:42:02 people
1:42:04 like Bots that talk
1:42:07 and have them start answering the phone
1:42:09 I would train them on all of my customer
1:42:11 data or you know my customer support
1:42:13 data
1:42:15 so when you call into this thing you're
1:42:17 no longer put on hold because it's just
1:42:19 Bots answering the phone but these Bots
1:42:20 are more capable than anything you've
1:42:23 ever talked to right because they know
1:42:25 everything
1:42:26 and those Bots are also smart enough to
1:42:29 say oh this guy's got a problem that
1:42:32 actually deserves to be escalated to a
1:42:34 human and it would escalate that problem
1:42:36 to a human so so what a big company
1:42:39 could do in in my opinion probably
1:42:42 within two years is completely get rid
1:42:44 of the call center replace all of those
1:42:46 call center people with with Bots and
1:42:48 then train new call center people maybe
1:42:51 that are local in different regions that
1:42:53 are higher level customer engagement
1:42:55 people so the Bots can deal with almost
1:42:58 everything but any special situations
1:43:00 that need to be escalated
1:43:03 go to actual humans and you don't get
1:43:05 put on hold
1:43:07 and you don't go through ridiculous
1:43:08 phone trees and everything is remembered
1:43:11 and notes are typed up so that by by the
1:43:13 time the bot escalates something to a
1:43:15 human they give them a complete report
1:43:17 of exactly what you talked about and
1:43:19 exactly what the issues were and exactly
1:43:21 why it was escalated so
1:43:24 um I think there's a chance that
1:43:26 customer support gets dramatically
1:43:28 better in this new world but it's going
1:43:30 to be massively disruptive
1:43:33 chat EPT is sometimes wrong chatgpt is
1:43:36 many times wrong I like to refer to it
1:43:38 as mansplaining as a service
1:43:41 it will confidently answer you even if
1:43:43 it's wrong
1:43:45 cheers man
1:43:50 we're good at that
1:43:55 I need uh tick tock
1:44:01 sorry just jump me to the bottom again
1:44:03 I'm scrolling back up through the
1:44:05 questions trying to find out where it
1:44:07 was because I don't want to miss
1:44:08 anyone's questions because I'm just that
1:44:10 good a guy
1:44:15 oh oh Poshmark listings oh like sales
1:44:18 listings
1:44:19 um I would still use zapier
1:44:21 I don't quite know what you're trying to
1:44:22 do but
1:44:23 zapier's a little janky but it's it's
1:44:26 once you kind of get your head around
1:44:27 what it is once you do one or two of
1:44:29 them it's not that bad
1:44:34 what is it
1:44:36 wait what is this what do you do this is
1:44:40 the AI learning lab my name is Kyle
1:44:43 Shannon I'm the co-founder and CEO of
1:44:45 video technology company called
1:44:46 storyvine
1:44:48 um I'm really passionate about
1:44:50 generative Ai and uh
1:44:53 there are very strong parallels thank
1:44:57 you there are very strong parallels
1:44:58 between what's going on now in the early
1:45:00 days of the World Wide Web which I was a
1:45:02 part of
1:45:04 um well not I wasn't a part of the World
1:45:06 Wide Web I was I was an innovator in the
1:45:08 early days I started one of the first
1:45:09 digital agencies in 1994.
1:45:12 and so I got to ride that wave like from
1:45:15 the ground all the way up and what's
1:45:17 going on with AI right now is like
1:45:19 identical but like times a thousand and
1:45:23 way faster so that's who I am do I
1:45:26 moisturize no but I am a little moist
1:45:29 how many people love the word moist go
1:45:31 ahead and tap that screen if you love
1:45:32 the word moist huh tap it if you hate it
1:45:35 too
1:45:38 a moist is just one of those words isn't
1:45:40 it
1:45:41 there should be laws
1:45:43 look at those hearts look at the hearts
1:45:45 coming you people are moist
1:45:49 all right
1:45:50 all right let me see where I was I think
1:45:52 I'm caught up did you hear about the
1:45:54 libel case against open AI for saying a
1:45:57 random man embezzled the government no I
1:46:00 didn't hear about that oh here's the
1:46:02 deal
1:46:03 the the people that are most excited
1:46:05 right now are the the litigators and
1:46:08 intellectual property attorneys they're
1:46:10 going to make a [ __ ] fortune for the
1:46:11 next couple of decades everyone is going
1:46:13 to sue everyone if because so here's the
1:46:16 thing like even even if the lawsuits are
1:46:19 frivolous like there's going to be so
1:46:21 much disruption
1:46:22 like some big company's gonna go out of
1:46:25 business and they're gonna go well that
1:46:26 was because of chat GPT [ __ ] it let's go
1:46:29 Sue chat GPT like like it's just it's
1:46:31 gonna be everywhere so I right now I'm
1:46:35 taking all of the lawsuits with a grain
1:46:36 assault none of these Technologies would
1:46:39 exist if these companies didn't just say
1:46:41 [ __ ] it let's just go train it on
1:46:43 everything
1:46:44 if they like like if open AI had gone
1:46:47 you know let's go get copyright
1:46:48 agreements from
1:46:50 every piece of data on the internet that
1:46:52 it would never would have launched right
1:46:54 if stable diffusion in mid-journey said
1:46:56 maybe we should get the artist
1:46:58 permission before we train on all their
1:47:00 [ __ ] no
1:47:02 like like why is stable diffusion open
1:47:05 sourcing all of their technology because
1:47:07 they probably know deep down they're
1:47:09 going to get their asses sued from here
1:47:10 to Oblivion and they might as well put
1:47:12 the tech out in the world
1:47:14 so we should be quite grateful in my
1:47:17 opinion that they did that like this was
1:47:20 a break eggs first kind of approach and
1:47:23 it's the reason why we're getting to
1:47:24 play with all these tools now that it's
1:47:26 out of the bag now we can sort out all
1:47:28 the attribution and compensation and
1:47:30 copyright and trademark issues
1:47:33 um now that we have these tools we can
1:47:34 start to figure out what the rules are
1:47:35 but none of this stuff would have
1:47:37 happened had they asked first
1:47:41 um how do I think AI will affect the
1:47:43 future end of it support that's what I
1:47:45 just talked about chat GPT is sometimes
1:47:47 wrong it's always wrong I'm
1:47:50 past where I was
1:47:51 I need them to refer to lobotomy I know
1:47:54 Sydney's kind of shitty right now that's
1:47:56 why so so Sydney's the the the
1:48:00 self-named bot at bing.com
1:48:03 and that's why I use open ai's the
1:48:07 official chat GPD site because it's just
1:48:09 better right now I don't like Bing as
1:48:11 much I don't either
1:48:12 I just tried to use Bing and the first
1:48:14 question
1:48:15 it told me I had hit a question
1:48:18 what's that say Interactive
1:48:20 interactive limit yeah I know things
1:48:23 things just shitty right now it'll get
1:48:25 better but
1:48:26 if you want to use gpt4 for free Bing's
1:48:29 great
1:48:30 if you're willing to pay 20 bucks a
1:48:32 month I would use open AI
1:48:35 this guy's just AI right yes I'm gpt5
1:48:37 nice to meet you
1:48:40 do you moist your eyes I'm telling you
1:48:42 you I get the funniest trolls it's like
1:48:45 it's like the trolls I get like it's
1:48:48 somewhere between a compliment and just
1:48:51 like a cutting sarcastic nasty Troll and
1:48:55 I and I respect it the Troll game is
1:48:58 solid in here difference between free
1:49:00 and pay version Okay so
1:49:03 so for chat gbt the free version gives
1:49:07 you GPT I say
1:49:10 um right right
1:49:13 um oh here's a good one right
1:49:15 a python application
1:49:19 that writes
1:49:24 Shakespeare
1:49:28 sonnets
1:49:30 um using open ai's
1:49:33 apis
1:49:36 and just it it just blasts it out really
1:49:40 quick right so so that's chat GPT 3.5
1:49:43 that's free it's really fast it's really
1:49:46 good especially if you're doing coding
1:49:48 or creative writing like any kind of
1:49:51 writing gpt4 is better so when you pay
1:49:55 for chat GPT plus you get gpt4 and then
1:49:59 you also get access to the new
1:50:01 features that they're rolling out one of
1:50:04 them is gpt4 connected to the internet
1:50:07 one of them is plugins there's a new one
1:50:09 coming out called code interpreter
1:50:11 there's like five or six of these new
1:50:13 modes that are coming code interpreter
1:50:15 the one I don't have access to it yet
1:50:17 but it's coming it is mind-blowing so if
1:50:22 you want kind of the newest and shiniest
1:50:25 um pay the 20 bucks a month to open Ai
1:50:29 and again I'm not making a commission on
1:50:31 that I don't give a [ __ ] if you do it or
1:50:32 not
1:50:33 but if you if you want to be on top of
1:50:35 this stuff you know I think it's worth
1:50:37 paying for
1:50:39 um south and east coast are going to
1:50:40 have abnormal rain all winter and colder
1:50:43 yeah I know yeah can I talk about Vector
1:50:46 databases I can but it would bore people
1:50:49 they're very boring
1:50:51 Vector databases are for data scientists
1:50:54 and I'm not
1:50:59 I'm yeah I might uh it looks like
1:51:01 there's some new folks here I'm not just
1:51:03 a whiny guy I'm I also act like a CEO
1:51:07 I'm a founder and CEO of a company
1:51:09 called storyvine
1:51:11 um an 11 year old video technology
1:51:13 company I've been deep diving into AI
1:51:16 for the past year this channel is about
1:51:19 me trying to get what I'm learning out
1:51:21 of my head because it is so
1:51:24 intense what's happening and really just
1:51:27 get as many people as possible up to
1:51:29 speed on AI demystify it de-demonize it
1:51:34 um and just talk about why I think it's
1:51:35 it is it is indeed the next big thing um
1:51:38 so that's what this is about
1:51:40 um Vector databases Okay so
1:51:43 ah where do I begin
1:51:45 when
1:51:47 when you train these large language
1:51:50 models on all the words
1:51:52 the way they work is
1:51:55 um they create
1:51:57 um
1:51:57 they create a vector space so think like
1:52:00 an XY plot right so if you have the word
1:52:03 dog and the word cat and the word bird
1:52:05 and the word whatever like those might
1:52:07 be all all be animals and then the the
1:52:10 embedding you know they basically turn
1:52:13 those words and their sort of semantic
1:52:16 neighbors into vectors so they kind of
1:52:19 plot them on an X Y
1:52:22 um chart so so they'll sort of plot them
1:52:24 mathematically
1:52:26 and it's not just two Dimensions it's
1:52:29 like thousands of Dimensions so these
1:52:31 are incredibly complicated
1:52:33 um mathematical spaces but what they are
1:52:36 is think of it like little neighborhoods
1:52:38 like like
1:52:40 um you take the word dog and and one
1:52:43 definition of the word dog is the furry
1:52:45 animal that we like to pet and keep in
1:52:46 our house and the other version of the
1:52:48 word dog is you know he's a dog he
1:52:50 cheated on her right so there's a whole
1:52:51 so those are two very different
1:52:53 semantically
1:52:55 um unrelated versions of the word Doug
1:52:57 so you have these little clusters of
1:52:59 words that are all semantically related
1:53:00 and then as you type your
1:53:03 um your prompts in
1:53:06 it takes those prompts and it
1:53:08 mathematically sort of calculates which
1:53:10 part of the vector space which part of
1:53:12 the the vector database you know these
1:53:15 little clusters should we be looking in
1:53:18 and then within that it it creates a
1:53:21 probability of what word would be the
1:53:23 next word in the response and it and it
1:53:25 just basically keeps doing that and so
1:53:26 as it writes all this stuff it's
1:53:28 literally writing it one word at a time
1:53:30 based on a probability engine so the
1:53:32 vector database is you can there's a
1:53:36 thing called Lang chain where you can
1:53:38 point the embedding engine at a bunch of
1:53:40 your data it turns it into these
1:53:42 mathematical numbers that live in this
1:53:44 crazy ass Vector space and then there's
1:53:47 Vector databases where you just dump
1:53:49 those numbers in and when you do your
1:53:52 llm search it can search on your data in
1:53:54 your vector database and you know it
1:53:57 basically combines it with the
1:54:00 the large language model Vector I don't
1:54:02 know how the [ __ ] it does it but it
1:54:04 basically you can add your data to their
1:54:06 data
1:54:07 it's some version of that that's what a
1:54:10 vector database is
1:54:12 um so yeah if you've got questions about
1:54:14 AI I obviously try to answer them even
1:54:16 if I am fundamentally flawed in my logic
1:54:19 and understanding because I'm a man and
1:54:22 that's what we do how would AI affect
1:54:24 end user it support we answered that one
1:54:27 already it's going to completely
1:54:28 transform it I think it's going to get
1:54:30 better for the end user I think it's
1:54:31 going to completely [ __ ] up a lot of
1:54:33 jobs
1:54:34 also instead of getting just one answer
1:54:36 from chat GPT ask it for 10 or 20
1:54:38 options yep that's really good the other
1:54:40 good trick with chat GPT is have it
1:54:42 analyze its own work when it gives you
1:54:44 an answer tell it to analyze the answer
1:54:46 and tell you how it could make it better
1:54:50 great job oh thank you Tick Tock just
1:54:53 told me I've done great because I've
1:54:55 been doing this for 90 minutes that's
1:54:57 why I can't feel my ass
1:54:59 cheers
1:55:03 okay I can't get mid-journey to work
1:55:05 says it's busy
1:55:07 to offer free trial on Discord
1:55:09 any way to work around now mid Journey
1:55:12 mid Journey you have to pay for like I
1:55:15 kind of resented the [ __ ] out of this
1:55:16 too there's what is the one
1:55:19 the [ __ ] is it
1:55:22 try leonardo.ai I think I think they've
1:55:26 got like a free version it's not
1:55:28 mid-journey it's built on stable
1:55:30 diffusion but it's a really good
1:55:31 interface
1:55:34 um let's go look at it oh I want to dump
1:55:36 over that hang on a sec
1:55:39 Leo
1:55:42 Leo Nardo dot AI
1:55:46 hey I I think you need to get on a wait
1:55:48 list for them I don't I don't know if
1:55:51 it's still like that
1:55:54 um because it says yes I'm whitelisted
1:55:56 so I think you have to get on a white
1:55:57 list but I think these guys give you
1:56:00 like 150 credits and then like each
1:56:02 image is five credits they give you like
1:56:04 150 credits a day
1:56:07 um you can also look up Blue Willow blue
1:56:10 Willows free again it's built on stable
1:56:13 diffusion not on mid-journey Mid journey
1:56:15 is the only one that does mid-journey so
1:56:17 and in my opinion it's just it's by far
1:56:20 the best of the the image generation
1:56:21 tools it's just it's just higher quality
1:56:24 um so I pay 10 bucks a month I pay like
1:56:26 the minimum amount I can pay
1:56:29 um but I I would I would pay for it if
1:56:31 you want really good stuff if you just
1:56:33 want to play
1:56:35 um go to
1:56:36 Leonardo or go to Blue Willow is is a
1:56:40 Discord based it's just like mid Journey
1:56:42 it's just not quite as good
1:56:45 but all these things are getting better
1:56:46 all the time so
1:56:49 how would AI affect end user it support
1:56:51 oh you don't understand how this works
1:56:53 see I answer all of these questions Kyle
1:56:56 I'm going to need an AI learning lab
1:56:57 notebook aren't I yeah I guess so I
1:57:00 think I should do a workshop at some
1:57:02 point I think a workshop might be fun
1:57:04 it's just a you know put down do do like
1:57:07 a couple hour Workshop over over a
1:57:09 couple of weeks that might be fun
1:57:12 um like for a sale listing yeah yeah got
1:57:14 it on Poshmark okay uh I yeah I would
1:57:17 still use zapier Allison or Holly
1:57:21 um anyone know of a comprehensive list
1:57:24 info to learn about the AI where you're
1:57:26 currently we currently have access to
1:57:28 yes
1:57:29 on screen right now why I put these
1:57:32 things here wait
1:57:35 um
1:57:36 the bottom one futurepedia.io there's a
1:57:39 bunch of these but this is a decent one
1:57:42 it lists all the [ __ ] in all the
1:57:44 categories so all the video tools all
1:57:46 the image tools all the music tools all
1:57:49 the writing tools other than the bag you
1:57:52 know what I'm saying you go over here
1:57:54 you go over there you come over here
1:57:56 they got a tool for you here you got a
1:57:59 tool for you there everyone's charging
1:58:01 8.95 a month if you did them all we'd
1:58:04 all be broke
1:58:07 all right
1:58:10 um
1:58:12 Jade talk to Kyle all right Jade talk to
1:58:15 me you have misspellings does that
1:58:17 matter here's what's amazing
1:58:19 misspellings in Google used to really
1:58:22 [ __ ] it up misspellings and chat it does
1:58:25 not care it knows it knows [ __ ] it knows
1:58:29 it it doesn't judge you
1:58:31 it doesn't
1:58:33 it like I have yet to have it not
1:58:36 understand something even when I've had
1:58:38 like horrific typos where like I don't
1:58:40 have a clue how to spell the word I just
1:58:42 sort of get it phonetically in the
1:58:44 neighborhood Nails it and then corrects
1:58:46 it in its response it's amazing
1:58:49 thank you for trying you're welcome I
1:58:51 don't know what I was trying
1:58:55 Adobe news is huge Firefly team was at
1:58:58 our office yesterday for our Innovation
1:59:00 week that's super cool
1:59:03 um oh retropunk you're the one at uh at
1:59:05 the Ad Agency right
1:59:08 um yeah I got a buddy at Deloitte
1:59:10 digital and and he was talking about
1:59:12 those guys that's it's a really big deal
1:59:14 it's a really big deal we had at the AI
1:59:16 Salon we had we had one of the creative
1:59:18 directors from Deloitte digital come
1:59:20 speak at the AI salon and he he this was
1:59:24 three months ago
1:59:26 and he could not even use mid-journey on
1:59:29 his work computer so so he was using
1:59:31 mid-journey to do concepting for
1:59:33 Creative Direction
1:59:34 and he had to do it on the weekends on
1:59:37 his own personal computer
1:59:38 and then he would email himself the
1:59:41 images that he created that he liked
1:59:43 because he couldn't use mid-journey on
1:59:45 his work computer because the lawyers at
1:59:47 Deloitte digital are like if if there is
1:59:49 like a microscopic chance of our clients
1:59:52 getting sued and then
1:59:54 by extension us getting sued no one's
1:59:57 [ __ ] using [ __ ]
1:59:58 so the fact that adobe has kind of
2:00:02 cracked the code of getting these big
2:00:03 companies to accept generative AI within
2:00:06 months I mean holy [ __ ]
2:00:11 it's amazing
2:00:14 they were very tight-lipped about any
2:00:15 updates yeah actually John Knack who's
2:00:18 one of the lead AI guys at Adobe he he
2:00:20 spoke at one of the real early AI salons
2:00:23 and showed us some of the early Firefly
2:00:25 stuff and he's he's coming back
2:00:28 either in two weeks or the meeting after
2:00:30 that to give us an update on generative
2:00:32 fill and all the stuff that they're up
2:00:34 to and and I'm definitely going to have
2:00:36 him talk about the Enterprise stuff
2:00:37 because I think it's a huge deal
2:00:39 I really do I think I think it's really
2:00:41 really powerful I'm so proud to see how
2:00:44 word to VEC
2:00:46 and glove have evolved I don't know
2:00:48 those
2:00:49 word to VEC are those your things
2:00:55 let's go look let's go look word
2:00:59 what is it 2 v e c
2:01:06 uh
2:01:07 tensorflow oh that's tensorflow core
2:01:13 yeah see that's
2:01:15 that's
2:01:16 definitely whatever the like like that's
2:01:19 that's definitely
2:01:21 I I am not
2:01:23 I'm not deep in the machine learning
2:01:25 World
2:01:27 um so
2:01:28 when all this when all this tensorflow
2:01:30 stuff came out I'm like oh I'm gonna go
2:01:32 learn all this stuff and then I went I
2:01:34 went to try to learn it I was just like
2:01:35 oh my God I don't have the brain for
2:01:38 this
2:01:39 um but yeah the the the speed with which
2:01:41 all of these tools are are improving is
2:01:44 is really impressive I mean that's the
2:01:46 the other thing is
2:01:48 like one of the one of the remarkable
2:01:51 things about Chachi PT getting such
2:01:54 massive adoption so quickly is that all
2:01:57 of the developer Community like turned
2:01:59 away from metaverse and toward
2:02:01 generative AI like everyone like and you
2:02:05 know everyone that was doing anything
2:02:06 just sort of laser focus on generative
2:02:09 AI so so the amount of innovation
2:02:11 happening because of that is just it's
2:02:14 insane
2:02:17 so I'm curious metaverse if you work for
2:02:21 one of those companies if you work on
2:02:23 that technology that's super cool
2:02:27 um I think that's key Kyle ask things
2:02:29 you know the answer to and learn how it
2:02:31 works yeah that so so what I was saying
2:02:33 if you're new to to chat gbt it can be
2:02:37 really overwhelming because it can do so
2:02:39 much you're like where do I start start
2:02:40 with stuff you know and and stuff that
2:02:43 when you get an answer back you can
2:02:44 immediately go that's good or that's bad
2:02:47 um because it's really easy to get
2:02:48 fooled by it because if you don't if you
2:02:50 don't actually know what you're looking
2:02:51 at it's it's very good at fooling you
2:02:53 with really shitty or false answers how
2:02:56 would AI affect end user support I have
2:02:59 answered that so many times we've talked
2:03:02 about this it's going to be very
2:03:04 impactful
2:03:06 but if you are a real specialist in your
2:03:09 field chechi PT and other llms very
2:03:11 often wait are very often on the surface
2:03:16 um
2:03:19 it it depends I I would certainly think
2:03:22 that in certain
2:03:24 in certain areas of expertise that's
2:03:26 probably true that that maybe they can't
2:03:28 go deep
2:03:29 but they're trained on such a a broad
2:03:32 data set that what it might be is that
2:03:34 you need to get better at providing
2:03:36 context and getting them to look more
2:03:38 narrowly so so here's the here's the
2:03:40 metaphor I figured out I heard Andre
2:03:41 carpathy talk who's the I think he's the
2:03:44 chief technologist and one of the
2:03:45 founders of openai and he was talking
2:03:47 about how these things work and he was
2:03:49 talking about the vector space and
2:03:51 Vector databases and things like that
2:03:53 and here's the metaphor that I came up
2:03:55 with
2:03:55 if you just type write me up like solve
2:03:59 this math problem right
2:04:02 think of it like you're in a helicopter
2:04:04 above the Grand Canyon and so when you
2:04:06 start out with a really generic prompt
2:04:09 you know down in the Grand Canyon are
2:04:11 all these canyons and sub canyons and
2:04:13 little you know
2:04:15 um you know like little whatever they're
2:04:17 called
2:04:18 mini Canyons like my brain's not working
2:04:21 right now
2:04:22 um but when you're high enough in the
2:04:24 helicopter you can see down you can see
2:04:26 all of the all of the vector spaces
2:04:28 right you can see all of the Canyons
2:04:31 and and what he was talking about was
2:04:34 you ask it to solve a very specific math
2:04:37 problem chatgpt doesn't know the
2:04:39 difference between this math professor
2:04:42 who actually has the answer and these 20
2:04:44 students who answered the math problem
2:04:47 but they answered it wrong
2:04:48 so if you're if if you're at a very
2:04:50 generic kind of prompt it'll sort of
2:04:53 look down into all the vector space and
2:04:55 it's like I just need an answer to this
2:04:57 math problem and it'll sort of you know
2:04:59 pull in the amalgamation of answers and
2:05:02 it might pull in 90 of them as student
2:05:04 answers and 10 of them as math professor
2:05:07 answers and it will just give you
2:05:08 something that looks like an answer that
2:05:10 it thinks is right
2:05:12 but if what you do is say I want you to
2:05:14 act like you know a a Stanford you know
2:05:17 level Advanced mathematic Professor what
2:05:21 you're doing is you're kind of lowering
2:05:22 that helicopter into a specific Canyon
2:05:24 where the the vector space does not
2:05:27 include so many false positives
2:05:29 effectively right so the reason it works
2:05:32 well to tell it to be a specific thing
2:05:34 is is you're eliminating a bunch of the
2:05:37 the the wrong answers so that might be
2:05:39 there there might be
2:05:41 there might be something there where you
2:05:43 can dig deeper into a specific area and
2:05:45 not have it be so generic
2:05:54 Cablevision does that with service
2:05:56 troubleshooting that's cool I would like
2:05:58 to see companies completely revamp
2:06:01 customer service because it's so
2:06:04 painful to deal with anything with
2:06:07 customer service right now like let's
2:06:10 blow the [ __ ] up and start over
2:06:14 I lost my place in the questions again
2:06:17 I'm also getting tired I'm tired I'm
2:06:21 sweaty somebody called me moist earlier
2:06:24 and that hurt my feelings
2:06:33 uh I just love the idea that people are
2:06:37 scrolling through their Tick Tock and
2:06:39 they see like fat adult Baldwin and
2:06:40 they're like wait who's this and I'm
2:06:42 just going like
2:06:45 that makes me happy
2:06:52 Tick Tock things gotta go I'm deleting
2:06:54 the app
2:06:57 uh all right I'm trying to find my place
2:06:59 in the questions all right but if you're
2:07:01 a real specialist okay here we go
2:07:04 [Laughter]
2:07:05 um
2:07:07 how about in medical services I think
2:07:10 it's going to be brilliant in medical
2:07:12 services but I think there are certain
2:07:14 categories of places that will be
2:07:16 brilliant that will come slower anywhere
2:07:19 where people's lives are at the line or
2:07:21 insurance companies are paying for [ __ ]
2:07:24 they ain't gonna have it
2:07:27 um so like you don't want your pilot
2:07:30 replaced with an llm like you may want
2:07:34 your pilot assisted by an llm but you
2:07:37 want that [ __ ] in the cockpit when the
2:07:39 you know when the wind shear happens you
2:07:42 know coming into landing
2:07:45 um now the reality is the the AI might
2:07:48 actually be better at Landing in wind
2:07:50 shear than human Pilots but but
2:07:53 that's irrelevant right when you have
2:07:55 people's lives on the line and insurance
2:07:58 companies involved they want
2:08:01 you know they don't want the off chance
2:08:03 that a computer bug is gonna is gonna
2:08:05 crash the plane
2:08:07 um
2:08:07 so I think medical is kind of similar
2:08:09 but like they just did a test two weeks
2:08:11 ago or so where they had a bunch of
2:08:13 patients interact with chat GPT doctors
2:08:16 and real doctors and they didn't know
2:08:18 which were which it was text chat
2:08:20 the patient's overwhelmingly chose the
2:08:23 chat GPT doctors they said they were
2:08:26 more capable and they said they were by
2:08:28 far more empathetic
2:08:30 so these large language models don't
2:08:33 feel empathy they don't have empathy
2:08:37 but they they can deliver
2:08:40 they can deliver
2:08:42 data empathetically right so they can
2:08:45 deliver an empathetic answer even though
2:08:47 they don't know what they're delivering
2:08:48 they don't know it's empathy
2:08:50 they understand that when someone's
2:08:52 asking about a cancer diagnosis you know
2:08:55 just giving them the facts might not be
2:08:58 the best thing right so they'll give
2:09:00 nuanced answers and and right so
2:09:03 um
2:09:04 the other thing where I think I think AI
2:09:06 might be incredibly powerful in medical
2:09:09 services is we're coming we're coming
2:09:12 into a crisis in medical well we've been
2:09:15 in a [ __ ] crisis in medical for you
2:09:17 know the past couple of decades because
2:09:18 of the insurance companies but
2:09:21 but we're coming into an acute crisis
2:09:23 where we already have a nursing shortage
2:09:25 right now and and the the Boomer
2:09:27 population is sort of Aging into you
2:09:30 know sort of high medical needs and so
2:09:32 we're going to have we're going to have
2:09:33 a caregiver shortage that's going to be
2:09:36 pretty dramatic
2:09:38 and so so imagine if if you take like
2:09:43 the I showed earlier the like the
2:09:44 animated version of me so imagine being
2:09:47 able to you're a physical therapist and
2:09:49 and you sort of synthesize your voice
2:09:51 and you synthesize your face
2:09:53 and you can create a version of yourself
2:09:56 that can interact with your physical
2:09:59 therapy patients but they've got access
2:10:01 to you 24 7 right and and you can now
2:10:05 scale yourself right like as a human
2:10:08 being you can only talk with so many
2:10:09 patients but if you virtualize yourself
2:10:13 you can talk to many more patients and
2:10:15 they've got availability to you all the
2:10:17 time and then again that Avatar will be
2:10:21 smart enough to know is this Norm a
2:10:23 normal conversation or is there a
2:10:26 problem here that I actually need to
2:10:27 notify the the real physical therapist
2:10:30 on this and they can reach out to them
2:10:32 so and and you know those interactions
2:10:36 they might get paid for so so underpaid
2:10:39 um physical therapists might make more
2:10:41 money by virtualizing themselves and
2:10:44 things like that patients might get
2:10:45 better care
2:10:47 um you know 24 7 care things like that
2:10:49 and if it needs to escalate it can so I
2:10:52 so I think it's going to be absolutely
2:10:55 massive and impactful on Medical
2:10:57 Services I just think it's going to take
2:10:59 a while until it's accepted
2:11:04 love your content thank you very much I
2:11:06 appreciate that I am here just because I
2:11:11 like doing this and I like I you know
2:11:14 what I really like I like when people
2:11:16 have that that
2:11:18 can mind that mind explosion where I was
2:11:23 talking to a guy
2:11:24 he'd been he'd been using
2:11:27 they've been using chat GPT and he he's
2:11:30 like you know hey I see you doing all
2:11:31 this AI stuff and like I've been doing
2:11:33 some AI stuff too all right
2:11:35 and so we get on the phone we're talking
2:11:37 and
2:11:39 and I said I said let me show you
2:11:41 something I built and I showed him the
2:11:42 zapier thing I did which is a Content
2:11:44 generation engine and so it was give me
2:11:46 a topic and then answer like three
2:11:48 questions on the topic so I said to him
2:11:50 give me a topic and then I and then I
2:11:52 had him answer the three questions and I
2:11:53 was typing them in and so I hit submit
2:11:56 and and I said okay you should get an
2:11:58 email in 30 seconds or so and then we
2:12:00 keep talking and he gets the email and
2:12:03 he looks at it and then I see him do
2:12:06 this and he goes
2:12:07 he goes did it just do this and I said
2:12:11 yeah that's what it does like it it
2:12:12 makes all that content he goes
2:12:15 I had no idea I I gotta go I gotta go
2:12:19 and he hung up like it ended the
2:12:22 conversation it like totally blew his
2:12:24 mind he like he had no idea that these
2:12:26 things were as powerful as they were and
2:12:28 it's like if you just kind of scratched
2:12:30 the surface with them and use them like
2:12:31 a like a Google search engine you kind
2:12:33 of get back shitty little answers and
2:12:35 you're like yeah whatever it's like it's
2:12:37 like Google but a little fancier no it's
2:12:39 way way way way way way way way way more
2:12:43 profound than that and that moment where
2:12:46 people have that
2:12:48 um that Epiphany is just it's [ __ ]
2:12:50 magical
2:12:51 um have I worked with Al Gore I have not
2:12:54 um he he invented the internet I just
2:12:57 discovered the World Wide Web I didn't
2:12:59 even like I personally discovered that
2:13:01 there was a world wide web Tim
2:13:02 berners-lead discovered that or created
2:13:04 the worldwide web I need an AI that will
2:13:06 help with brand company name I wasn't
2:13:09 impressed with chat GPT or name licks
2:13:12 really
2:13:14 I can get Chachi if you need to do good
2:13:16 branding [ __ ]
2:13:17 let's let's go try to do some branding
2:13:19 stuff
2:13:20 I'm gonna go to gpt4 for sure
2:13:23 I'm not going to use the one connected
2:13:25 to the internet
2:13:28 okay I want you to act as a master brand
2:13:36 um consultant
2:13:38 and are not barned brand
2:13:42 that word I should probably Spell
2:13:43 correctly brand consultant and
2:13:46 identity
2:13:52 um developer
2:13:54 [Music]
2:13:58 I want you to ask
2:14:02 me the five critical
2:14:07 questions
2:14:09 about
2:14:14 my company
2:14:18 and what it promises
2:14:22 and then I want you to
2:14:29 use my answers
2:14:33 to create
2:14:37 a high level
2:14:40 brand strategy
2:14:45 document
2:14:49 of course I'd be happy to so I'm going
2:14:52 to stop it from generating and say
2:14:54 um only ask me
2:14:58 one
2:15:01 question
2:15:03 at a time don't move on
2:15:07 until
2:15:10 I have said
2:15:14 ready
2:15:19 all right so now this thing is going to
2:15:22 act like a little brand consultant
2:15:23 what's the mission of your company in
2:15:25 other words why does it exist
2:15:28 um
2:15:36 we
2:15:37 want to make the best paper swans in the
2:15:43 world
2:15:44 and make everyone happy
2:15:50 that's a wonderful Mission too what are
2:15:52 your core values we value good paper
2:15:58 good folding
2:16:01 and good shapes
2:16:07 it's late I'm I'm pushy
2:16:11 who's your target audience could you
2:16:14 provide some details about their
2:16:15 demographics behaviors needs and desires
2:16:19 we want anyone who wants
2:16:25 to wait once to be happy and
2:16:33 uses shallow
2:16:37 um
2:16:39 shallow
2:16:42 items
2:16:44 to fill
2:16:46 a void in their soul this got really
2:16:50 dark
2:16:55 I understand it seems like people
2:16:57 looking for small tangible sources of
2:16:59 Joy okay so it actually it actually kind
2:17:01 of rewrote that in a way that's not so
2:17:02 smarmy
2:17:04 people looking for small tangible
2:17:05 sources of joy and perhaps a sense of
2:17:07 calmer distraction from their daily life
2:17:09 that's actually pretty good all right
2:17:11 how would you define your company's
2:17:12 unique selling proposition
2:17:15 we make swans
2:17:18 of all sizes
2:17:23 great providing a variety of sizes
2:17:26 caters to diverse customer needs and
2:17:28 sets you apart well it probably doesn't
2:17:30 but that's okay what do you want your
2:17:32 brand how do you want your brand to be
2:17:33 perceived by the audience
2:17:36 um
2:17:36 it should be Whimsical and fun
2:17:43 that's fine
2:17:44 wonderful thank you for sharing that now
2:17:46 let's start with the high level brand
2:17:48 strategy
2:17:50 brand purpose
2:17:52 to create the world's finest paper swans
2:17:55 bringing happiness and joy to
2:17:56 individuals worldwide
2:17:58 our brand thrives on blah blah core
2:18:01 values quality paper Artful folding
2:18:03 perfect shapes target audience unique
2:18:06 selling proposition brand perception
2:18:09 we aim for our brand to evoke a sense of
2:18:11 Whimsy and fun so like it like it it
2:18:15 really has taken in the answers it's
2:18:18 incorporating them into the strategy our
2:18:20 strategy will center around our unique
2:18:22 selling proposition And the emotions we
2:18:24 aim to evoke marketing efforts blah blah
2:18:26 blah
2:18:27 it's still going
2:18:29 all right did it finish okay so now I'm
2:18:32 going to say
2:18:33 um
2:18:36 give me 10
2:18:40 um
2:18:42 serious and 10 fun names
2:18:49 uh for this business
2:18:57 first the serious ones Elegance paper
2:18:59 Swan Swan craft Creations origami
2:19:02 Odyssey Papyrus Grace
2:19:04 Swann Serenity Paper Company not bad
2:19:07 ethereal Swan's Majesty folds or I would
2:19:11 go Majestic Folds Elite elegant swans
2:19:15 Timeless Swan Artistry those are okay
2:19:17 now the fun ones Swan Dive Paper Company
2:19:20 the swan flock shop Wacky Wings Swan
2:19:23 balloon creations Swan Lake paperworks
2:19:27 paper feathers fun so let's go with what
2:19:30 was the one that was
2:19:32 we'll go with Majestic folds I like
2:19:35 Majestic
2:19:38 Folds
2:19:40 can you
2:19:42 um
2:19:43 give me five logo
2:19:48 logo Concepts
2:19:52 that
2:19:54 um
2:19:55 I could
2:19:59 uh uh could
2:20:03 use as a prompt for an art
2:20:08 director or a and
2:20:13 AI image generation tool
2:20:18 let's see what it comes up with I'm glad
2:20:21 you like Majestic folds here are five
2:20:22 logo Concepts origami swan in Flight
2:20:25 that goes with our short story that we
2:20:27 wrote Regal Crown and Swan An Elegant
2:20:31 Royal Crown placed beautifully so wait
2:20:34 what's the name of our business again
2:20:36 oh Majestic fold so a Regal Crown would
2:20:39 make sense nested swans multiple paper
2:20:42 swans of different sizes nested together
2:20:44 it remembered that we use swans of
2:20:47 different sizes paper to Swan
2:20:49 transformation oh that's kind of cool so
2:20:51 so you have some sort of logo that that
2:20:53 is like flat paper folding into a swan
2:20:56 and then a joyful swan
2:20:58 so let's grab
2:21:10 I'm going to grab the first one origami
2:21:12 flan now let's do a Regal Crown and Swan
2:21:15 so I'm going to grab this let's go over
2:21:17 to Mid journey and see if we can
2:21:19 generate a logo
2:21:21 all right
2:21:25 I know we're down a rabbit hole here but
2:21:27 you know this is this is what 80d is
2:21:29 like
2:21:32 slash imagine
2:21:34 slash imagine
2:21:37 um
2:21:39 let's see
2:21:41 a modern
2:21:45 logo
2:21:48 with crisp line art comma and then I'm
2:21:54 pasting in
2:21:57 what they wrote an elegant Royal Crown
2:22:00 gently placed on a beautifully folded
2:22:02 paper swan
2:22:06 um let's just try that
2:22:10 so anyway
2:22:12 um
2:22:13 you know like like the
2:22:19 I don't know the the only the only the
2:22:21 only thing like I I have yet to have it
2:22:24 just give me truly shitty stuff that I
2:22:26 can't use but
2:22:29 I've had it give me
2:22:31 shitty stuff like
2:22:33 you know
2:22:35 for an hour
2:22:37 and then I keep [ __ ] with it and I
2:22:39 keep adjusting my inputs and I keep
2:22:41 going back and forth and back and forth
2:22:43 and eventually I'll get to the place
2:22:44 that it gets pretty decent
2:22:47 so let's see what this does with these
2:22:49 Swan logos so you know already they're
2:22:53 they're looking kind of interesting I
2:22:54 don't know that they're on a crown
2:22:56 a royal crown
2:22:59 and they're certainly not origami oh we
2:23:02 didn't say origami
2:23:04 so let me copy this
2:23:10 those are okay but that's not what we
2:23:12 want
2:23:14 copy
2:23:16 slash imagine
2:23:20 um let's see
2:23:22 a modern logo of a
2:23:27 an origami
2:23:31 swan
2:23:36 gently placed
2:23:40 in a Royal Crown
2:23:48 with crisp line art and it will kill the
2:23:51 end of this this should give us
2:23:52 something better all right let's see how
2:23:54 this does
2:23:58 coms I'm scrolling now that was a deep
2:24:01 Rabbit Hole I hope you appreciated that
2:24:03 all right how can we use chat EBT to
2:24:05 improve myself and my role at work start
2:24:08 interacting with it so if you go to
2:24:09 prompts.chat
2:24:12 if you go to
2:24:14 prompts.chat
2:24:15 I know in that list you can have it ask
2:24:18 act as a business coach act as a
2:24:20 psychologist
2:24:21 there's a bunch of them in there where
2:24:23 you can have it act as certain things
2:24:27 I would just start
2:24:29 telling it about yourself like if you
2:24:31 provided context of like here's who I am
2:24:34 here's how I operate here's what the
2:24:37 challenge that I'm facing at work can
2:24:38 you give me strategies for how to
2:24:40 improve myself maybe you want to
2:24:42 communicate better have it come up with
2:24:44 scripts for how you can talk to peers or
2:24:47 bosses or you know whoever you want to
2:24:50 so
2:24:54 it still didn't quite get origami oh you
2:24:56 know what I should have done instead of
2:24:58 line art I should have said folded paper
2:25:00 well whatever so anyway like we're not
2:25:03 quite there but these are better than
2:25:05 the ones we had before right so so it's
2:25:09 getting there like the
2:25:11 yeah they're all too fancy right now
2:25:13 like I would simplify it I would you can
2:25:15 use paper folding things you could also
2:25:17 use certain art styles and illustration
2:25:20 styles to get it there but anyway
2:25:22 um yeah listen and you know
2:25:26 is this like groundbreaking you know
2:25:30 branding work no but you know that was
2:25:33 sort of 10 minutes of me
2:25:35 [ __ ] around with it look like we at
2:25:37 least got somewhere at least understood
2:25:39 what the hell we were trying to do and
2:25:40 gave us gave me some ideas so
2:25:45 um if you need a mod I'd be happy to
2:25:47 help I'm in almost every night um yeah
2:25:50 reach out to me if you go to my um
2:25:53 profile Holly
2:25:55 um yeah I might like I I might want to
2:25:58 up the up the game a little bit and and
2:26:00 uh Ahmad Ahmad might help so if you want
2:26:05 to do that hang out with me while I make
2:26:07 an ass of myself
2:26:11 um it would probably suck less than me
2:26:14 trying to do it all by myself
2:26:17 um can you make catch EPT do excel
2:26:20 sheets
2:26:21 um you can I I can I can show you how to
2:26:24 do it I haven't done it in ages my
2:26:26 problem is I'm not good enough at Excel
2:26:28 to do it but there's a there's a thing
2:26:30 called
2:26:32 um
2:26:34 GPT for sheets so just Google GPT for
2:26:38 sheets and just go start using it
2:26:40 because basically all it is
2:26:42 is it's a prompt in an Excel cell
2:26:46 so you can so you can build into an
2:26:49 Excel formula like take the value of
2:26:51 cell you know A1 and you know insert it
2:26:55 into this prompt and then send that to
2:26:57 chat gbt and then come back and give me
2:26:59 a list of 10 things here and then you
2:27:02 can have the next cells over take those
2:27:03 10 things and send those to chatgpt and
2:27:06 create other [ __ ] it's pretty remarkable
2:27:08 but I don't know
2:27:11 um I don't know sheets and Excel well
2:27:14 enough to really be able to demo that
2:27:16 but
2:27:17 um GPT for Sheets I think is the the one
2:27:20 that I've used
2:27:22 and there's probably 8 000 of them now
2:27:24 so that's going to be annoying
2:27:26 but yes you can do that and also
2:27:29 um
2:27:30 both Microsoft and Google are going to
2:27:33 be releasing their office suites with
2:27:36 all of this [ __ ] embedded in it so all
2:27:38 of those plugins are not going to be
2:27:40 necessary these are just gonna gonna
2:27:42 come so if you want to just dick around
2:27:43 with it now to start to get your head
2:27:45 around the concept go use those plugins
2:27:48 but but
2:27:49 I'm guessing before the end of the
2:27:51 summer we'll see
2:27:53 Microsoft 365 and Google Docs with GPT
2:27:57 built into it
2:28:00 why should I pay for chat GPT when Bing
2:28:03 chat is free and basically the same
2:28:04 minus plugins well Bing will get
2:28:06 plug-ins the reason you pay for it is
2:28:08 open AI is the developer the technology
2:28:12 and theirs is just better like I don't
2:28:15 know why it's better it's just better
2:28:17 and and Microsoft with Bing they're
2:28:21 trying to do a bunch of things there
2:28:22 they're trying to incorporate
2:28:24 advertising and the chat
2:28:27 and annotations like they're trying to
2:28:29 do things differently they're trying to
2:28:31 create a bing experience not a chat gbt
2:28:33 experience
2:28:35 um but yeah it's your point is a good
2:28:37 one it's gpt4 it's connected to the
2:28:40 internet and it will support plugins so
2:28:42 even though it doesn't right now it's
2:28:43 going to
2:28:44 so so yeah I mean quite frankly the fact
2:28:49 that you can play with gpt4 for free
2:28:52 with Bing is is pretty remarkable what's
2:28:56 the plug-in for gpt4 that connects it to
2:28:58 the internet well there's there's just
2:29:00 kind of it being connected to the
2:29:02 internet is one of the modes
2:29:05 so
2:29:06 which is not connected to the internet
2:29:08 then you've got browse with Bing that's
2:29:12 gpt4 connected to the internet using
2:29:15 Bing branding honestly they're probably
2:29:17 using some
2:29:18 some Bing technology underneath it but
2:29:21 that that was just a co-marketing deal
2:29:23 right there and then you've got the
2:29:25 plugins and and you'll have other stuff
2:29:27 coming
2:29:35 do you use moisturizer what kind of
2:29:38 question is that you people with your
2:29:42 questions that are nearly compliments
2:29:45 you're a very interesting keep it up so
2:29:47 educational thank you text to three 3D
2:29:50 generated work is already here it is
2:29:52 it's shitty right now but it's gonna get
2:29:55 really good really fast I mean what's
2:29:57 going to be cool is because they'll
2:29:59 build this [ __ ] into the printers
2:30:00 themselves
2:30:02 where you'll just be able to be on your
2:30:04 phone and like I you know take a picture
2:30:06 of something and it'll you know
2:30:08 auto-generate it it'll come up with a
2:30:10 prompt Auto generate it and then send it
2:30:12 to your 3D printer or you'll just type
2:30:14 in so you just talk you'll just say you
2:30:16 know I wanna I wanna 3D
2:30:19 origami swan and make it three inches
2:30:22 wide and you know half an inch high and
2:30:25 [Music]
2:30:27 that's coming
2:30:29 can we use chat GPT with any computer
2:30:32 yes or does it work better with certain
2:30:34 types of computers ones that have
2:30:36 screens works much better if it's got a
2:30:38 screen
2:30:40 not necessarily even you don't need to
2:30:42 have a screen you can use an API you
2:30:44 don't need a screen for that [ __ ]
2:30:46 you can just go get something from chat
2:30:47 gbt have it grab some code and send it
2:30:49 to a controller you don't even need a
2:30:51 screen no any computer it's it basically
2:30:54 it's that address
2:30:55 chat.openai.com point a web browser
2:30:58 there and just go Marvel at the Wonder
2:31:00 behold the wonder that we have
2:31:03 in our hands
2:31:06 Arthur C Clarke said any sufficiently
2:31:09 advanced technology is indistinguishable
2:31:11 from Magic
2:31:13 that's what these tools are
2:31:17 [ __ ] Magic
2:31:19 it's like do you remember when you were
2:31:21 a kid and you got the magic kit like
2:31:24 your grandparents got you the magic kit
2:31:26 and you're like oh [ __ ] I got the magic
2:31:28 kit I'm gonna do all this magic and you
2:31:31 look at the cover and you're like oh you
2:31:32 guys this is amazing and then you open
2:31:34 it up and there's the cups and the balls
2:31:36 and the foam things and then and the
2:31:38 rope and the in the Rings and you're
2:31:40 like I'm gonna be a magician and then
2:31:43 you open the little book and you're like
2:31:46 wait it works like that
2:31:48 how do you
2:31:50 what a it's gonna take me 10 years to
2:31:53 figure this [ __ ] out then you have the
2:31:54 that like that crushing disillusionment
2:31:58 that like you actually have to practice
2:31:59 to get good at [ __ ] in this world
2:32:02 well gpt's like getting that magic kit
2:32:06 except the [ __ ] in it is actually Magic
2:32:09 like that's what it's like
2:32:13 you have to get over the part where
2:32:14 you're like this is just like a stupid
2:32:16 Google search
2:32:18 um
2:32:19 it ain't a nice analogy is Zeus
2:32:22 Prometheus and Pandora's Box we humans
2:32:24 are Zeus
2:32:25 and do we give the fire that's good I
2:32:28 like that
2:32:31 um somebody sent me an ice cream cone ah
2:32:34 he's so swell
2:32:37 Adobe Firefly is also free yes it is
2:32:39 YouTube oh that's right if if you want
2:32:42 to use image stuff Adobe Firefly YouTube
2:32:45 will show you how to get around Leonardo
2:32:47 white list oh cool that's great
2:32:49 thank you for the rose I appreciate the
2:32:51 rose those are important I think I don't
2:32:53 know what they are but I like them they
2:32:56 are lovely
2:32:57 there are so many interfaces building on
2:32:59 an AI how to pick I don't know I am
2:33:01 right now I have gone from trying to try
2:33:05 everything to like [ __ ] it I'm playing
2:33:08 on chat GPT I'm playing on Poe I will
2:33:13 occasionally play with Bing and then I
2:33:15 go to Mid Journey
2:33:17 and like if I want to train my own
2:33:19 models I go to Leonardo and that's kind
2:33:23 of it because it's just it's [ __ ]
2:33:27 exhausting
2:33:28 and and quite frankly like the the
2:33:31 amount of power
2:33:33 in Chachi PT is it's like Earth
2:33:37 shattering it's it's like I I get
2:33:39 serious howdy partner uh
2:33:45 um
2:33:46 like
2:33:48 it's just it's hard to describe like
2:33:51 like how profoundly powerful it is it's
2:33:54 like the deeper you go like it just
2:33:56 keeps opening and opening and opening
2:33:58 it's like if you ever if you ever have
2:34:00 you ever been caving
2:34:02 and and you like you're crawling through
2:34:04 these little claustrophobic you know
2:34:06 tunnels and you gotta go under the thing
2:34:08 and you think you're gonna get crushed
2:34:10 and then you sort of come out and it's
2:34:12 like dark and then your eyes adjust and
2:34:14 you see you're in this big room and then
2:34:16 like you go around the corner and it's
2:34:18 like
2:34:19 [ __ ] massive
2:34:21 it's like that
2:34:25 hallucinations input yeah it
2:34:28 hallucinates a lot facts GPT makes me
2:34:31 think I can spell exactly and you
2:34:33 spelled can wrong
2:34:35 GPT makes me think think I cab spell you
2:34:39 spell like I do
2:34:41 um Google be like did you mean yeah
2:34:43 exactly
2:34:44 um I had gpt4 and Bard build knew the
2:34:46 same WordPress queries in PHP gpt4
2:34:49 smoked barred completely Bard kind of
2:34:51 sucks right now Google is on their
2:34:53 [ __ ] heels
2:34:55 they they screwed the pooch they
2:34:58 invented the [ __ ] technology sitting
2:35:00 underneath all the stuff we're playing
2:35:01 with right now and they slept on it
2:35:04 because they knew it was going to
2:35:05 threaten their business and Sam Altman
2:35:07 came along and kicked their ass and now
2:35:10 they're playing catch-up
2:35:12 that's terrifying I mean for them I
2:35:16 don't give a [ __ ] I'm listen I love it
2:35:19 when you know the uh well it's hard to
2:35:21 call Sam Altman an underdog at this
2:35:23 point he just got 10 billion dollars for
2:35:24 Microsoft and they're rolling his
2:35:26 product into everything they do so he's
2:35:28 no longer an underdog but but but the
2:35:31 fact that
2:35:33 that company
2:35:35 shook up the 92 or 94 market leader as
2:35:40 as significantly the as they have is
2:35:42 amazing
2:35:43 will AI take over Insurance underwriting
2:35:46 absolutely all the complicated [ __ ] in
2:35:49 this world like grant writing for
2:35:52 proposals and insurance policies and
2:35:55 insurance underwriting all of that [ __ ]
2:35:58 all the stuff that's really complicated
2:35:59 where there's whole Industries built
2:36:02 around the complication
2:36:04 is all going to get automated out all of
2:36:07 it it's gonna absolutely take it over
2:36:09 when I don't know but
2:36:11 it it will obliterate the the the need
2:36:15 to do all of that human labor
2:36:21 don't you think if the metaverse people
2:36:24 switched to AI that it
2:36:28 doesn't look great for ai's future wait
2:36:30 don't you think that if metaverse people
2:36:33 switch to AI that it doesn't look great
2:36:36 for ai's future
2:36:37 a bunch of metaverse people did switch
2:36:39 to AI like everyone jumped ship from
2:36:42 metaverse including Mark Zuckerberg he's
2:36:45 like their last earnings call he didn't
2:36:46 use the word metaverse once the
2:36:48 company's called [ __ ] meta
2:36:50 he he didn't mention metaverse was
2:36:53 they're like [ __ ] checking out of
2:36:55 that [ __ ] the Apple Google goggle Vision
2:36:59 Pro thing they didn't use the word
2:37:01 metaverse once they also didn't use the
2:37:03 word AI ones which I think is smart
2:37:07 but they made it very clear they got
2:37:10 some serious [ __ ] machine learning
2:37:13 Hardware their chips are pretty
2:37:15 remarkable so they are gonna be when
2:37:18 they roll out their AI stuff it's gonna
2:37:20 be insane
2:37:23 um where are you reading comments from
2:37:25 or are you just behind I'm way behind
2:37:27 because I try to read all the comments
2:37:29 and I've been doing this for two hours
2:37:30 now so I'm probably like you know 30
2:37:32 minutes behind so
2:37:33 if I don't get the asari I'm doing what
2:37:36 I can well we'll see if you like this
2:37:39 tap the screen if you don't like it tap
2:37:42 the screen also why not uh if you tell
2:37:45 it who they are it makes a difference
2:37:47 yeah if you tell the GPT models to to
2:37:50 act as a certain Persona it makes a huge
2:37:52 difference
2:37:54 it feels stupid to do that like like it
2:37:57 feels like well that's dorky like I
2:38:00 don't want to tell it act as a wizard
2:38:02 but if you tell it to act like a wizard
2:38:04 it'll give you some magic spells I mean
2:38:07 it's like it's just it works and again
2:38:11 now that I understand how it works that
2:38:13 the sort of Grand Canyon metaphor that
2:38:15 if you don't give it context you're like
2:38:17 really high on the helicopter and it can
2:38:19 kind of see all the [ __ ] and it'll just
2:38:21 give it'll just give you answers out of
2:38:23 all the [ __ ] because it doesn't know
2:38:24 what it's giving you
2:38:25 but if you give it specific things you
2:38:28 kind of lower the helicopter into a
2:38:30 semantic region of the canyon and it'll
2:38:33 be a lot more accurate so it kind of
2:38:34 makes sense
2:38:37 all right I'm getting caught up on my
2:38:39 questions here you oh no I'm now wait
2:38:41 now I've answered all these dang it
2:38:45 dang you tick tock
2:38:47 Tick Tock you and me is gonna have to
2:38:49 have some words we're gonna have to have
2:38:51 some words we're gonna talk about your
2:38:54 interface it sucks even for the young
2:38:57 people
2:38:58 it's not just us oldies
2:39:01 aren't you protogen X shut up
2:39:05 don't you think the metaphors oh here we
2:39:07 are okay sorry
2:39:11 uh I got some laughs thank you for
2:39:15 laughing am I high not yet
2:39:18 no I'm not I I'll tell you last time I
2:39:20 got high it was it was a couple of years
2:39:22 ago
2:39:23 um I live in Denver so it's legal here
2:39:26 um I took an edible and then I started
2:39:29 watching TV and because I have ADD I
2:39:31 forgot that I took the Edibles so I'm
2:39:33 just sitting there watching TV and I'm
2:39:34 like I don't feel right I'm starting to
2:39:37 have a panic attack that like
2:39:39 something's wrong like I'm stroking out
2:39:41 and then it just keeps going and going
2:39:44 I'm like this really isn't right I'm
2:39:46 like oh
2:39:47 I ate an edible [ __ ]
2:39:51 it was just it was just uncomfortable I
2:39:54 just didn't like the feeling of it so
2:39:55 like [ __ ] it I'm not doing that again
2:39:57 our lawyers approved 150 MJ licenses wow
2:40:01 it was crazy all pro plants too everyone
2:40:05 is serious in stealth mode all in the
2:40:07 same server that's [ __ ] huge retro
2:40:10 Punk really
2:40:12 wow
2:40:13 and now with Photoshop gen fill layers
2:40:16 they'll be happy yeah I agree Jen Phil
2:40:18 and mid-journey are a Powerhouse combo
2:40:20 they are honest to God you do the
2:40:23 initial generation in mid journey and
2:40:25 then take it into Photoshop and sort of
2:40:26 blow the doors off it holy [ __ ] that's
2:40:30 amazing so so retro if you're the guy
2:40:33 I'm pretty sure the guy we we chatted he
2:40:35 works for a big Ad Agency
2:40:38 and the fact that they're leaning into
2:40:40 mid-journey
2:40:41 um that's that's that's huge that's
2:40:44 really huge
2:40:47 um and it's huge and it's good to hear
2:40:50 someone said I'm fun thank you
2:40:52 thank you for this you're welcome Jade
2:40:56 um what was this prompt to get the
2:40:58 prompt ideas what was this prompt to get
2:41:01 the prompt ideas
2:41:03 well prompts.chat is a website that
2:41:06 gives you a bunch of prompt ideas
2:41:08 appreciate this so much you're welcome
2:41:10 I'll email you yeah email me please do
2:41:14 um do the Swan logo with firefly it's
2:41:17 brilliant oh cool oh yeah cause Firefly
2:41:20 okay let me let me show you
2:41:22 oh not Firefox God damn it
2:41:26 Firefly
2:41:30 fire
2:41:32 fly
2:41:34 oh I don't have one do I not have it
2:41:36 installed here oh it's a website duh
2:41:38 he's a genius
2:41:41 he's so good at this AI stuff well
2:41:43 except he's there's a part he's he's
2:41:44 well he's a little dumb
2:41:47 he's he's he's not the smartest he's not
2:41:51 the smartest cookie on the tree if you
2:41:53 know what I mean imagine
2:41:54 [Laughter]
2:41:58 fire firefly.adobe.com
2:42:03 [Laughter]
2:42:05 sign in
2:42:09 did I sign in with Google now I've got
2:42:12 an adobe thingy
2:42:16 um
2:42:19 let me jump over to Discord and get my
2:42:22 crappy
2:42:25 prompt
2:42:28 that was created in chat GPT I agree
2:42:31 with that let's see text to image
2:42:34 next skip
2:42:37 all right text image I'll just drop it
2:42:40 in here on the bottom
2:42:41 so the nice thing about Firefly oh I'm
2:42:44 you're not seeing all this hang on let
2:42:45 me let me shrink my screen here a little
2:42:48 bit oh yeah those are much better wow
2:42:49 that's cool
2:42:51 um so that's art let me
2:42:53 so so one of the cool things that adobe
2:42:55 did with firefly
2:42:58 is
2:43:00 they took all the stuff like in mid
2:43:02 Journey where you'd have to type dash
2:43:03 dash you know style or whatever they
2:43:07 made it into into buttons so like I can
2:43:10 just click one of these things like
2:43:11 digital art and it will
2:43:15 um it'll automatically update stuff oh
2:43:17 wait
2:43:18 let's see content type none
2:43:22 it's also really fast
2:43:27 yeah those are actually pretty good
2:43:30 like that upper left one's pretty good
2:43:31 we don't want a photo we don't want an
2:43:33 art graphic Styles let's see
2:43:38 techniques effects movements what's
2:43:40 movements oh steampunk that sort of [ __ ]
2:43:42 themes
2:43:44 techniques let's try that
2:43:50 color tone lighting Concepts
2:43:56 anyway
2:43:57 yeah this is pretty cool
2:43:59 dark iridescent
2:44:04 here's layered paper let's generate
2:44:06 layered paper
2:44:13 yeah those are pretty slick
2:44:15 that's really cool
2:44:18 like look at that
2:44:20 I mean so so that was if you've been
2:44:22 here for a while you saw you saw me go
2:44:24 through The Branding exercise you saw me
2:44:25 get the uh
2:44:28 get the prompt mid-journey did it okay
2:44:31 but the the speed with which you can
2:44:33 iterate within
2:44:36 um fireflies pretty cool the other cool
2:44:38 thing about Firefly if you haven't
2:44:39 played with it is
2:44:42 um and it's a firefly.adobe.com is the
2:44:45 is the address
2:44:47 um
2:44:49 let's see let me go back to the home
2:44:52 page
2:44:53 they've got text effects
2:44:56 so you can just say let's see enter the
2:45:00 text
2:45:01 um what was it called
2:45:04 Majestic folds yeah Majestic
2:45:09 Folds
2:45:11 is the text and then I'm going to say
2:45:13 origami
2:45:15 swan
2:45:23 it didn't do origami very good at all
2:45:26 did it oh it's internally yeah some
2:45:29 things it doesn't quite get but anyway
2:45:32 but you can do you can do stylized text
2:45:34 with with Adobe Firefly as well
2:45:37 it's it is quite impressive
2:45:41 it is quite impressive and now it's
2:45:43 going to be used by all the companies
2:45:46 it's a really smart move by Adobe what
2:45:48 you know is amazing we talked to them
2:45:51 at the salon back in
2:45:54 December and January and they weren't
2:45:56 even sure if they were going to do
2:45:58 anything with generative Ai and in five
2:46:00 months
2:46:02 they're freaking taken over the world do
2:46:05 I think AI will replace the role of CEOs
2:46:08 no so I think that
2:46:11 I think that
2:46:14 that generative AI is is going to
2:46:17 automate out kind of lower level
2:46:19 tactical stuff I think the thing that
2:46:21 remains actually quite important
2:46:24 with these tools is is uh strategy
2:46:30 critical thinking decision making right
2:46:33 I think all of the stuff that
2:46:36 that is the like like why we do what we
2:46:39 do as a business
2:46:40 um I think that remains with the humans
2:46:42 I think I think AI is very much the
2:46:45 co-pilot and businesses still need
2:46:47 Pilots it needs conductors it needs
2:46:49 people figuring out what we want to do
2:46:52 with these tools
2:46:54 um
2:46:54 so I I no I don't think it like I think
2:46:58 I think what it's going to do to CEOs is
2:47:00 I think it's going to force them to make
2:47:02 choices like okay if I've got a tool
2:47:05 that makes my company 30 to 40 percent
2:47:07 more efficient
2:47:09 do I
2:47:11 lay off 30 to 40 percent of the staff
2:47:14 increase our quality by 30 to 40 percent
2:47:18 decrease our time to deliver by 30 40 so
2:47:21 we can take on more clients I think
2:47:23 there are decisions that CEOs are going
2:47:25 to need to make that are going to be the
2:47:27 direct impact of it
2:47:28 fun question do you think you're
2:47:30 anything like Kyle and South Park
2:47:33 uh yes hey Kim thank you Kenny they
2:47:37 killed Kyle
2:47:39 um
2:47:40 I live right near
2:47:42 um
2:47:44 Casa uh what is it called Casa Bonita
2:47:48 and the South Park guys just bought that
2:47:50 and reopened it so
2:47:54 um the food there was atrocious before
2:47:55 so hopefully it's better now it works
2:47:57 better on PCS with
2:47:59 avex2 CPUs and it will slow down on Old
2:48:03 avex CPUs interesting
2:48:08 um it's peeling an onion it is peeling
2:48:10 an onion using all this stuff
2:48:14 um looks like we're almost out of
2:48:16 questions people are going to bed I need
2:48:18 to go to bed it's almost [ __ ]
2:48:20 midnight I've been doing this since nine
2:48:23 eight thirty
2:48:25 uh uh uh they don't work better on
2:48:29 certain types of computers they all work
2:48:31 in the cloud that's correct although the
2:48:34 large language models that are open
2:48:36 source that you can install locally on
2:48:38 your computer like if you're on a Mac
2:48:41 you should be on an M1 like like they're
2:48:43 they're writing some of these large
2:48:44 language models for Apple silicon and
2:48:47 for computers that have you know 30 90
2:48:50 and up graphics cards
2:48:52 so glad I was here for the that edible
2:48:55 story it was so fun
2:48:59 sitting there watching TV like I think I
2:49:02 might need to go to the emergency room
2:49:08 oh I stopped the time you are at 14
2:49:11 minutes behind that's
2:49:13 sorry about that but you know that's the
2:49:15 uh but I've been here for two hours and
2:49:18 it's it's all entertainment don't you
2:49:20 think I mean look at that hair that's
2:49:22 [ __ ] funny
2:49:26 um ironic that the Natalia person in
2:49:29 chat is an AI generated video yeah it's
2:49:32 not I I put up some AI generated videos
2:49:35 I think if like like I think if people
2:49:37 want to communicate but don't want to
2:49:39 put themselves on camera
2:49:41 [ __ ] it
2:49:42 I heard Leonardo has similar ways to
2:49:45 change settings as Firefly easier for
2:49:47 beginners yeah also Leonardo is really
2:49:49 good at um you can train your own model
2:49:51 you can like I uploaded a bunch of my
2:49:54 photography to it the the sort of cloud
2:49:56 photos that I do with with all sorts of
2:50:00 editing and uh it does them pretty good
2:50:07 what's the plug-in to use chat GP T4
2:50:10 that will connect you with the internet
2:50:11 we answered that it just is connected to
2:50:13 the internet now
2:50:15 and I think we're out of questions all
2:50:16 right it's almost midnight I'm moist
2:50:19 I've been called out on my moistness
2:50:22 I'm tired
2:50:24 people still think I'm funny but looks
2:50:26 aren't everything
2:50:28 you get it it's a dad joke any
2:50:31 recommendations for AI jobs where I
2:50:34 don't need to be an engineer developer
2:50:35 yes welcome to
2:50:38 so so we just hit an inflection point
2:50:40 November 30th 2022 marked an inflection
2:50:44 point where before that point
2:50:46 to get a job in AI
2:50:48 you had to be kind of
2:50:51 a scientist or a data scientist or a
2:50:54 researcher
2:50:57 the launch of Chachi PT marks GPT
2:51:00 available for the rest of us right just
2:51:02 like the internet had
2:51:04 the World Wide Web where before the
2:51:08 World Wide Web it was for scientists and
2:51:09 researchers and Librarians after the
2:51:12 World Wide Web everyone could do it and
2:51:15 businesses adopted it and people adopted
2:51:17 it and all that sort of [ __ ] so we're
2:51:19 going through the exact same thing here
2:51:22 so Paul rateser who started the
2:51:25 marketing AI Institute if you don't
2:51:27 listen to his podcast you should listen
2:51:29 to his podcast and follow him on
2:51:31 LinkedIn he writes really good blog
2:51:33 posts
2:51:34 he's got he's got a saying that I think
2:51:36 I'm pretty sure he coined that AI will
2:51:39 not take your job but someone using AI
2:51:41 will
2:51:42 and so what I would say is you don't
2:51:45 necessarily need to look for an AI job
2:51:47 but what you should do is start getting
2:51:49 really curious about Ai and start
2:51:51 exploring it and start talking about it
2:51:53 talk about it on your LinkedIn you know
2:51:56 mention hey I just tried this thing with
2:51:58 chat EBT and you know I learned that it
2:52:00 did this well and it didn't do that well
2:52:02 anything you learned like that's why I
2:52:03 started this channel it's like I just I
2:52:05 I want to start processing this stuff
2:52:07 out loud
2:52:08 the result of it has been people now
2:52:11 recognize oh like he knows something
2:52:14 about AI like even if I don't I know
2:52:16 more than they do right because I've
2:52:17 been doing it for a few months longer
2:52:20 um do the same [ __ ]
2:52:22 um just start getting curious with it
2:52:24 start taking what you already know how
2:52:26 to do and figure out how to amplify that
2:52:28 with AI
2:52:30 and and and then start talking about it
2:52:33 when you go on job interviews start
2:52:35 talking about it when you're in bars
2:52:36 start talking about it when you're at
2:52:38 the picnic
2:52:39 and people in your life will look at you
2:52:42 like you're a [ __ ] idiot and that's
2:52:45 okay because once you understand the
2:52:50 power of this thing you'll know that
2:52:53 they're the [ __ ] idiot
2:52:58 uh the name of the podcast is Mark the
2:53:01 marketing AI Institute podcast
2:53:04 Paul rateser
2:53:07 r-o-e-t-z-e-r is his name and follow him
2:53:10 on LinkedIn and tell him I sent you
2:53:16 I don't know what good that'll do all
2:53:18 right good night everybody I appreciate
2:53:20 all of your comments and uh your time
2:53:23 and your attention and stay curious go
2:53:27 to those things peace out I'll see you
2:53:30 soon
2:53:30 bye