aiLL 6/15/23 - The Power of AI Tools to Spark Your Creativity

Description

Chapters

Transcript

0:05 foreign
0:29 with the AI learning lab
0:32 hope everyone's well just getting set up
0:34 here for the evening
0:37 for an evening of AI learning activities
0:44 um
0:45 I am an entrepreneur and we talk about
0:49 things Ai and I've also got
0:54 a
0:56 an AI Salon that meets every other week
0:59 our next meetings next Tuesday the
1:00 information's there and if you go to
1:02 Twitter that everyday AI news that's a
1:05 newsletter that I've got we've got
1:06 threads we put up there about
1:07 interesting news and we've got a Weekly
1:10 Newsletter that uh can prove helpful so
1:14 if you're interested in that kind of
1:16 stuff check it out
1:18 and if you have questions about AI pop
1:21 them in the comments below I will do my
1:24 best to answer what I can uh I will be a
1:29 smartass I will probably cuss a little
1:31 because you know
1:33 I get fired up about this AI stuff
1:36 some of you more practical types are
1:38 probably thinking sir what are your
1:41 qualifications
1:42 well
1:44 I am not a machine learning scientist I
1:47 am not a data scientist but I've started
1:50 I don't know 15 companies in my day uh
1:54 I've been doing this entrepreneur thing
1:55 for a while I was there for the early
1:57 days of the World Wide Web
1:59 I'm a proud gen xer which I don't know
2:01 what that means just that I'm bitter
2:03 bitter you betcha and uh
2:06 I started one of the early uh digital
2:09 agencies company called agency.com back
2:11 in 1994.
2:12 that was probably uh a few years before
2:16 most you were born except for you
2:19 Elder talk people it's uh I'm living
2:21 here in Denver it's starting to
2:23 thunderstorm like I don't know for the
2:26 past month we're getting afternoon
2:28 thunder showers it's like we're living
2:29 in Florida I don't know what's going on
2:31 uh anyway what what's it say more oh
2:34 more likes can attract more viewers so I
2:36 double tap that screen I guess that'll
2:38 give you more likes that'll attract more
2:40 viewers
2:41 um if you're new here my name is Kyle
2:42 Shannon this is the AI learning lab I
2:45 like talking about
2:46 um
2:47 AI stuff so pop questions in the
2:49 comments below I'll be sure to answer
2:52 them and
2:54 um some information
2:57 I guess I guess every oops every good
2:59 meeting starts with
3:01 what do they call it let's get the uh
3:03 let's get to bookkeeping out of the way
3:05 so let's get the bookkeeping out of the
3:07 way this is the AI learning lab
3:09 uh the salon is a community of
3:13 curious people trying to figure out what
3:16 the hell is going on with this AI stuff
3:19 um
3:20 every day I news is my newsletter and
3:23 let me
3:25 turn that stuff off and turn this stuff
3:28 on oops everything is in the wrong place
3:32 so professional operation okay
3:36 so if you're new to ai go to that top
3:39 URL chat.openai.com that's the official
3:42 chatgpt website go there and start
3:44 playing
3:46 um
3:47 I'll talk about that I'll show you how
3:48 to do plugins I'll show you I'll answer
3:50 anything you want to know about that the
3:52 next thing down there bing.com is that's
3:55 Microsoft search engine but they have a
3:59 um
4:00 what's it called if they have a chat
4:02 button if you click the chat button
4:03 that's actually chatgpt4 connected to
4:06 the internet that you can use for free
4:08 and then poe.com was one where you can
4:10 play with a lot of different models the
4:11 other two are just resources I think
4:13 don't suck
4:15 and so that's fun
4:19 and I'm going to move these things down
4:21 I can probably do it faster than that
4:23 yeah there we go
4:25 now let me go back over here yeah it's
4:28 pretty good all right so we'll do this
4:31 we'll make this bigger
4:33 I'll do this we'll do this
4:38 is that working
4:39 don't mind me while I figure this crap
4:42 out
4:43 on your dime but you know it's free Tick
4:46 Tock gives us this for free
4:48 um Okay so
4:50 let's see comments how is AI changing
4:54 the trading Market as a day trader and
4:57 if you have any platforms to watch out
4:58 for
5:01 the the world of trading but I do know
5:05 that um
5:06 chechi PT is really good at coding and
5:10 understanding trading platforms and
5:12 trading methodologies and
5:16 creating trading Bots and taking trading
5:18 bucks from your platform of choice and
5:21 writing you new ones and if you've never
5:22 done that if you've never seen chat EBT
5:24 do its thing with code let's just we'll
5:27 pop over here and I'll I'll show you
5:29 some stuff so so most of my answers
5:33 involve some version of go try chat gbt
5:36 because it's incredibly capable so uh so
5:40 I'm going to say Act
5:42 as
5:44 a day trading expert and that little tip
5:48 of telling Church apt to act as
5:50 something actually does make a
5:52 difference
5:53 act as a day trading expert and
5:57 um
5:58 tell me
6:00 how you might
6:02 help me program a trading bot
6:08 let's see what it says here
6:13 gpt4 has been a little slow lately
6:17 all right uh
6:19 gives this the obligatory I'm not a
6:22 licensed financial advisor but here you
6:24 go
6:28 Define your strategy choose your
6:30 programming language
6:34 set up your development environment
6:35 connect create anxiety stop I'm going to
6:38 say stop okay
6:39 um is there a trading platform
6:47 that has
6:49 um trading Bots
6:53 I can just have you program or modify
7:00 but if I
7:03 see what it says there
7:05 are several trading platforms
7:09 [Music]
7:17 so metatrader
7:19 so you probably know this better than I
7:21 do but basically if you can do trading
7:23 Bots within your platform you can copy
7:24 and paste them so you can say write me
7:27 trading but
7:29 with stop losses
7:34 and triggers
7:37 um for meta
7:40 Trader
7:44 see what it does
7:46 creating a trading bot involves writing
7:49 code in mql4 or mql5
7:53 please note this is a very basic example
7:55 and may not be profitable
8:07 and so there it is and I'm actually
8:08 doing this in gpt4 which is a little
8:10 slower let me stop it here I'm going to
8:11 actually flip over
8:13 to GPT 3.5 because it's much faster make
8:17 me a met meta
8:21 Trader Trading
8:25 um but
8:27 with a conservative set of triggers
8:33 and stop losses and I mistyped that
8:36 really poorly
8:39 [Laughter]
8:43 chat you PT servers must be getting
8:45 hammered there's too many day Traders
8:47 making trading Bots apparently
8:51 foreign so that seems to be stuck we'll
8:54 let it fail for a second and then I'll
8:56 I'll come back over all right let me
8:57 answer another question in the meantime
8:58 I'll come back to that Kyle for
9:00 president you just let's do this
9:03 watch out
9:04 that would be a fun presidency we
9:07 wouldn't get [ __ ] done but we'd have a
9:09 lot of fun doing it
9:11 all right I was looking for a network
9:13 error regenerate response come on oh
9:16 there it goes
9:19 uh let me flip my camera back around
9:26 can you
9:28 create
9:33 and
9:35 example for me
9:37 for me
9:45 bam
9:47 there you go so hope that helps
9:50 um the other thing that I would say
9:54 is go to that bottom site there
9:57 futurepedia dot IO
10:00 um
10:01 it is it is
10:04 completely impossible to keep up with
10:07 all the tools that are being launched on
10:09 a daily basis so the reason I put this
10:12 futurepedia.io here is there's all sorts
10:14 of tools so I would go there and look up
10:16 Financial AI tools and there are
10:20 probably some that specialize in Bots
10:21 but if if you haven't been playing with
10:24 the with chat gbt and you want to play
10:26 with Bots that's a perfect kind of
10:28 project to learn chat GPT on because
10:30 it's incredibly incredibly capable I was
10:33 looking at terms of service for
10:34 mid-journey it says they own the license
10:36 for all the images I'm not sure about
10:38 that does that limit the use of those
10:41 images uh if if that's what it says like
10:44 with with all legal stuff I would
10:46 consult your lawyer if you're looking to
10:48 do stuff for
10:50 um
10:51 for commercial purposes
10:53 um
10:54 start using Adobe Firefly and generative
10:57 fill within Photoshop because they have
11:00 explicitly trained the Firefly model on
11:03 public domain images and Adobe stock
11:05 images they own all the rights to
11:08 so they're making it safe for commercial
11:11 use and it's pretty good the tools are
11:13 pretty good
11:14 um
11:15 so so definitely use Adobe
11:19 um and yeah check with your lawyers
11:21 because I ain't a lawyer I have a degree
11:23 in acting and if you pay attention to me
11:26 for legal advice
11:32 you don't want that
11:34 by the way if you're a lawyer and you're
11:37 giving out legal advice and you use chat
11:39 gbt proofread your stupid work
11:42 because there was a lawyer last week
11:46 that um
11:48 that put in a legal brief to a judge
11:51 that he had written in
11:54 um Chachi BT and didn't check his work
11:56 and the judge was like these citations
11:58 are all fake what's up with that
12:05 all right what company has the best
12:07 chance of dominating the AI Fields good
12:11 Lord there's a question
12:14 well open AI has a good chance at it
12:17 because they're helping to Define this
12:19 new burgeoning
12:21 um you know Chachi pt-centric World
12:25 um Microsoft is doing pretty good right
12:28 now Google is struggling but I wouldn't
12:30 count them out because they've got more
12:31 money than you know all the companies
12:33 put together
12:35 um that's probably not true don't call
12:36 me on that I'm being hyperbolic people
12:39 hyperbolic do you know what that means
12:42 neither do I so go to chatgpt and ask it
12:45 what hyperbolic means
12:47 have it write you a poem on the meaning
12:49 of hyperbolic
12:51 or a song
12:53 do I always talk like this no sometimes
12:56 Maybe
12:58 do you pay for GPT I do so
13:00 if you pay for chat GPT let me flip my
13:03 camera by the way Tick Tock took away my
13:05 dual camera and I'm bitter about it
13:09 all right if you pay for chat GPT
13:12 then when you go to chat.openai.com you
13:15 kind of have these two options here
13:16 you've got GPT 3.5 that's that's the the
13:20 I guess GPT light Chachi Beauty light
13:24 but it's really fast
13:26 uh although tonight they both seem slow
13:28 and then there's gpt4 and gpt4 is the
13:31 one that's really really good but also
13:33 you get browse with Bing so that's GPT
13:36 connected to the internet and you get
13:38 plug-ins and plugins are little
13:40 extensions to GPT that are super cool
13:43 like I've I have two of them
13:45 connected here one's called prompt
13:48 perfect one's called
13:50 read my PDF ask your PDF and I think
13:53 I'll turn on one other wait I don't see
13:55 it here hang on
13:57 plugin
14:00 I don't have it let me go find it
14:03 so we're going to go over here
14:05 to the plug-in store so so when you have
14:07 plugins there's a plug-in store it takes
14:09 a while to load
14:11 and then if you say oh let's see how
14:12 many are there tonight 111 pages of four
14:15 each so 444 plugins so it is growing on
14:20 a daily basis
14:23 oh I've got web pilot installed okay
14:25 let's go back I'm going to turn on web
14:27 pilot
14:28 um
14:30 where was it how did I miss it
14:33 because I'm a dum dum dum dum dum dum
14:36 dum dum web pilot okay I'm going to turn
14:37 on web pilot ask your PDF and
14:41 um what's that thing oh prompt perfect
14:43 so prompt perfect so so the way these
14:46 the way these plugins work is you kind
14:48 of just GPT is normal and then depending
14:50 on what you type and what plugins you
14:52 have activated it will fire off one of
14:55 those plugins so
14:58 um one of them I have turned on here is
15:00 called prompt perfect so if you type the
15:02 word perfect before your prompt and then
15:04 you type
15:06 um right let's see write me
15:11 um the opening paragraph
15:15 to my great
15:18 American novel
15:22 I like
15:26 uh let's see I like
15:29 flowery
15:33 Pros
15:34 and it's
15:37 about a guy
15:41 who becomes an influencer
15:45 at 58.
15:50 so now what it's going to do is it's
15:52 gonna fire off the prompt perfect
15:55 plug-in
15:56 and if you would with these plugins you
15:58 can kind of open them up and see what
16:00 they're doing so it took my prompt it
16:02 sent it off
16:04 and then it rewrote the prompt so please
16:06 craft an opening paragraph from a great
16:08 American novel to Showcase an eloquent
16:10 and ornate writing style
16:12 the novel sensors around a protagonist
16:14 who at the age of 58
16:16 unexpectedly Rises to famous social
16:18 media influencer I am seeking a vivid
16:22 and captivating introduction right so it
16:24 rewrites it that's a that's a that's a
16:26 swell prompt let's see
16:30 in the Twilight of his years no I'm not
16:33 a Twilight of my years that is offensive
16:35 when most men find solace in the
16:38 quietude of retirement what was this
16:40 written in 1892
16:42 a man of 58 Winters found himself on the
16:45 precipice of a new dawn his life once a
16:49 symphony of routine predictability this
16:51 doesn't know me very well at all
16:53 so anyway that's how that's how plugins
16:55 work so I do pay for chat GPT
16:58 and that's why I pay for it because it
17:01 can it can write normally it writes
17:03 flattering things about me but that was
17:04 good Lord in the in the Twilight of his
17:07 years does this look like Twilight of my
17:10 years good people
17:15 I don't know if you've heard this Butch
17:17 EBT hallucinates so it's hallucinating
17:20 that's not real I'm not in the Twilight
17:21 of my years but I'm thirsty
17:25 it's raining
17:27 the Denver Nuggets are world champions
17:29 they did a big parade today jokic was
17:32 like I didn't want to come to the
17:33 [ __ ] parade but now I love this
17:35 [ __ ] parade you guys are [ __ ]
17:36 awesome it was great
17:39 all right if you have questions about AI
17:42 or anything like that drop them below my
17:44 name is Kyle Shannon this is the AI
17:46 learning lab we have fun here well I
17:49 have fun here and hopefully bring you
17:51 all along for the ride
17:53 um
17:54 AI learning lab the salon is
17:57 um a community I put together we meet
17:59 bi-weekly in meetups both virtually and
18:02 here in Denver and we have a Discord
18:05 where you can geek out with us and then
18:07 every day AI news is my newsletter
18:08 that's the Twitter handle so you can go
18:10 check that out there
18:12 um is the Google is the extension for
18:14 Google Sheets secure I don't know I
18:18 assume so because you're using your own
18:20 key but is it possible they've done a
18:22 nefarious thing there it's possible I
18:24 suppose I don't know what the security
18:27 um
18:28 Protocols are for creating a Google
18:30 extension if you're really concerned
18:33 about that I would wait Google has
18:34 announced they're going to roll GPT or
18:36 Bard into all of The Office Products and
18:39 micro not Microsoft has announced the
18:42 same for for Microsoft co-pilot which is
18:44 gpt4 that's going to get rolled out
18:46 across office 365. the one that I've
18:50 used for for Google Sheets is called GPT
18:54 for Sheets
18:56 um but there's probably a bunch of them
18:58 now because this stuff has gotten so
19:00 popular like it's it's essentially
19:02 impossible to find the official chat GPT
19:05 app and the IOS app store so if you want
19:08 to actually use the official app go to
19:10 openai.com and go to their products and
19:14 chat GPT and then click on their link to
19:17 the App Store but what I do on my mobile
19:19 devices I just go to I go to wait for it
19:24 they killed my double camera so this
19:26 makes me cranky
19:28 I go to that top URL right there on my
19:32 phone chat.openai.com that's the
19:34 official website so raffle well chat gbt
19:37 is very young so it's 58 years old yeah
19:39 exactly I can't believe these people in
19:42 the Twilight of his life Twilight
19:45 Twilight
19:46 wasn't that it's a Jim Gaffigan joke
19:49 Twitter was it Twinkies no it wasn't
19:51 Twinkies
19:52 I don't know he's got some sort of joke
19:54 where he over over enunciate some word
19:57 and it's very very funny all right uh
20:01 that's me what's you
20:05 is there a plug-in extension for Google
20:07 Sheets oh is it secure that's where I
20:09 was okay so I don't know um but I've
20:11 used the one called GPT for sheets but
20:14 you could also just wait
20:16 um if you're good at um
20:19 if you're good at uh spreadsheets the
20:24 the GPT extensions are insanely powerful
20:27 they really are you can put in one word
20:29 in a spreadsheet and they'll just like
20:30 Auto populate the whole spreadsheet with
20:32 a series of cascading calls to open AI
20:35 it's pretty amazing
20:38 you want to generate some synthetic data
20:40 or you want to generate
20:42 um you know all the execution of a
20:44 social media campaign out it goes
20:46 pulling a topic from the news or just
20:49 type in a topic and just let it go write
20:50 a bunch of stuff for you
20:52 I asked about a novel about the bite
20:55 about Biden family bribery okay awesome
20:58 but please keep politics out of here
21:00 because I don't want to get booted and I
21:01 don't really give a [ __ ] and I hate the
21:03 divisiveness in our country so I don't
21:05 want to talk about it
21:07 uh Symphony of routine I know offensive
21:10 off the bat exactly
21:12 why light of his life Twilight of his
21:15 life
21:18 I am vital
21:21 all right
21:22 how can you make money with cheche EBT
21:24 step one don't listen to people that go
21:26 you've got my money
21:31 here's the thing to understand chat EPT
21:33 is just a tool it's just a tool
21:36 so so the answer to how do you make
21:38 money with chat GPT is the same as the
21:40 answer to how do you make money with
21:41 Photoshop or how do you make money with
21:44 a frying pan or a muffin tin or uh
21:49 uh Microsoft Word
21:52 you figure out a way to use those tools
21:54 in such a way that it solves a problem
21:56 for people good enough that they're
21:57 willing to pay for it that's business in
21:59 general
22:01 all right I don't sound 58 thank you
22:04 very much I appreciate that yeah I'll
22:06 tell you yeah
22:08 oh get off my lawn
22:12 I will not be twerking because I am in
22:14 the Twilight of my existence
22:18 uh were you the one asking for the
22:20 ornate writing it was ornate it I to be
22:24 fair I did ask it well no wait let's go
22:26 back to that let's hang on I don't think
22:29 I asked it for innate writing prompt
22:32 perfect helped me rewrite it in the
22:34 Twilight of his years I wrote write me
22:37 an opening paragraph to my great
22:39 American novel I like flowery prose
22:42 which I did say that about a guy who
22:45 becomes an influencer at 58 it wrote
22:47 please craft an opening paragraph for my
22:50 greater create American novel that
22:51 showcases an eloquent and ornate writing
22:54 style so prompt perfect might have been
22:57 the thing that took it over the
22:58 Victorian Edge
23:00 but I think it's it's just ridiculous in
23:02 the Twilight of his years you know what
23:05 here's what we're gonna do so this is
23:06 this is something fun you can do say I'm
23:08 gonna go listen
23:10 um
23:13 this is 2023
23:17 and 58 is middle age
23:22 you jerk please
23:27 please update this so I don't get
23:34 cranky
23:35 and tell you to get
23:39 off my lawn
23:44 we'll see what it says my apologies for
23:46 any misunderstanding let's adjust the
23:48 narrative to reflect that in the vibrant
23:51 heart of his middle years our
23:53 protagonist
23:54 [Laughter]
23:54 s
23:56 stood on the cusp of an extraordinary
23:59 transformation his life once a real well
24:03 reversed Melody of the familiar was
24:05 about to be swept up in the dynamic
24:07 tempo of the digital age that is much
24:10 better
24:11 vibrant heart of his middle years
24:15 um but that's actually a good lesson if
24:17 chatgpt gives you something that you
24:19 don't like fix the prompt
24:21 tell it you don't like it tell it why
24:23 you don't like it and it will adjust it
24:25 it's job it's programmed to give you
24:27 what you want and it tends to be pretty
24:29 good at that nuggets all right who would
24:32 have thought I love it how do I make a
24:35 team Grand a week
24:38 using AI you can make money
24:42 I don't know
24:44 have it write you a hundred t-shirts and
24:46 then go do print on demand have it write
24:48 you a hundred posters and go make them
24:50 in canva and sell them on Etsy like have
24:53 it write you a book to tell other people
24:54 how to make money and then you can make
24:57 a tick tock where you go
25:01 that's probably the way to make money
25:03 without that
25:06 and if your question is how many times
25:08 do you make that joke a night
25:10 30.
25:13 if your other question is will you ever
25:15 stop
25:16 um
25:23 uh please can you explain the new
25:25 functions in chat EBT I think I just did
25:28 uh if you have a specific one
25:31 can you use the plugins with chat EBT
25:33 free you cannot but
25:36 Microsoft has said they're going to
25:38 support plugins
25:40 um the the
25:41 open AI plugins so so these plugins are
25:44 going to be like the App Store you're
25:45 going to be using you're going to be
25:46 able to use them in a lot
25:48 except probably not the Google chat bot
25:50 Bard but Bard sucks right now
25:56 um so that doesn't matter some people
25:57 like Bard the the people that use it for
25:59 coding say it's pretty good but for
26:00 writing it is poo poo it is stinky like
26:03 doggy poo all right can you use plugins
26:06 with GPT three three no
26:08 um AI learning lab what application of
26:11 AI are you most excited about
26:14 hmm
26:16 I think it would have to be chechi BT
26:19 um because so for me
26:21 Chachi BT is the
26:24 it's the equivalent of the World Wide
26:25 Web so so I started one of the early
26:28 digital agencies in mid 94.
26:31 um when the World Wide Web first came
26:32 out and so the world wide web made this
26:36 thing called the internet that was
26:38 previously just used by scientists and
26:40 researchers and universities they kind
26:42 of made it available to the rest of us
26:43 and I I you know
26:45 I I was lucky enough to be in New York
26:47 City when that came out and you know
26:50 surrounded myself with enough of the
26:52 right people that you know I got a
26:53 chance to to do a lot of really exciting
26:56 stuff in web development for you know in
26:58 the early part of that industry
27:00 and
27:01 chat CPT is a very very similar it's a
27:05 very parallel kind of product where you
27:09 know
27:09 machine learning and ai's been around
27:11 for decades
27:13 and even though things like mid-journey
27:15 and stable diffusion and gpt3 were were
27:18 you know being talked about it didn't
27:20 really Catch Fire and then November 30th
27:23 2022
27:24 chat or open AI launches chat GPT and it
27:27 gets to a hundred million users in six
27:29 weeks I don't know if you know that
27:30 fastest adoption of technology in
27:33 history the World Wide Web it took six
27:35 years to get to 100 million users Chachi
27:38 BT is six weeks
27:40 so
27:41 there's lots of exciting things going on
27:44 like it like I would say it it
27:47 it depends on like I've written seven
27:50 screenplays I like writing I like words
27:52 and chat GPT in particular lets you
27:55 wield your words and get incredibly
27:58 powerful things back so the the idea of
28:01 sort of crafting with words like you're
28:03 programming with your native language in
28:05 my case English
28:07 and it's kind of like programming with
28:09 English so so it's pretty cool you don't
28:11 really need to know a lot of syntax you
28:13 just need to have good ideas and put
28:15 them in there
28:16 um but but there's really exciting
28:17 things going on with video like with
28:19 runwayml with with Gen 1 and gen 2.
28:23 um mid-journey is just freakishly good
28:27 um Adobe Firefly is powerful because
28:29 it's commercial ready like they're
28:32 making it safe for businesses to use
28:34 generative AI which that's a big deal I
28:37 people don't think it but it's
28:40 there's a there's a big ugliness in the
28:44 creative Community because of the nature
28:46 of how these models are trained where
28:48 the like they're stealing my images if
28:50 you actually understand how these tools
28:52 work they're not copying and pasting so
28:54 it's not really stealing it images but
28:57 it is strongly influencing the output on
29:00 you know copywritten work and so Adobe
29:02 has end run that by creating a model
29:07 that's not trained like that which is
29:09 super smart so that's exciting and then
29:11 the other thing that's really exciting
29:13 is the whole open source Community open
29:15 source large language models so so
29:17 before meta released their llama models
29:21 Into the Wild
29:22 everything was behind firewalls right
29:25 open AI had theirs Google had theirs
29:27 meta had theirs and then meta just snuck
29:30 their models out there and apparently I
29:32 think today they said they've got a new
29:34 one out that's better than GPT 3.5 that
29:38 they're going to open source
29:40 so the open source stuff's really really
29:42 cool as well
29:44 um
29:45 welcome everyone if you're new here
29:48 my name is Kyle Shannon this is the AI
29:50 learning lab I obviously like to ramble
29:52 and
29:53 talk [ __ ] about chat gbt
29:56 um if you haven't tried anything with AI
30:00 go to that top URL I'm not affiliated
30:03 with it I'm not making millions of
30:05 dollars on it but wouldn't that be cool
30:06 if I could
30:08 um that's just that's the official chat
30:10 GPT website if you've never been there
30:12 if you haven't tried it go play with it
30:15 and the next one down
30:18 bing.com is Microsoft and and if you
30:20 click on the chat button that's gpt4
30:22 that you can use for free and then
30:24 poe.com is also a pretty cool tool with
30:28 lots of models on it
30:30 um prompts.chat will teach you about
30:32 prompting future pedia will let you find
30:34 tools
30:35 so that's what's going on there pop any
30:38 comments you have below thank you for
30:40 tapping the screen I think the likes
30:42 help bring people here which is awesome
30:44 so Tap Away I if it's good if you think
30:47 I suck then I guess tap on the poopy
30:50 icon
30:51 and I'll just see a little poop flying
30:53 up there that'll be cute can you tap on
30:56 the poop icon
30:57 I I would imagine I'm gonna get a lot of
31:00 poop right now probably gonna send
31:01 people packing
31:02 fudge
31:05 oh I'm in a ridiculously dad joke mood
31:09 tonight all right but I'm a dad so I'm
31:11 allowed
31:11 this is comedy gold you're damn right
31:13 have you been drinking no I in fact I
31:16 was uh I was supposed to uh I have a
31:19 monthly whiskey night that I go to like
31:21 a whiskey tasting thing because you know
31:23 a lush and uh I had some stuff to do at
31:26 home so I had to do it and then I just
31:28 I'm here so in lieu of drinking this is
31:30 what you get imagine if I were drunk
31:35 I can imagine it I might go get myself a
31:38 whiskey I might they might just uh let
31:41 me just loosen up a little all right
31:43 let's see great thank you you're welcome
31:45 should I purchase the subscription uh it
31:48 wasn't it's up to you I I can tell you
31:50 what I've purchased I've purchased
31:52 Chachi BT I've purchased Po and I'll
31:55 show you why in a second for Poe
31:58 I purchased mid-journey
32:01 I've purchased zapier beat for
32:03 automations if you haven't
32:05 um
32:06 done automations that's really fun where
32:08 you string together a bunch of different
32:10 things using chat GPT
32:13 uh and you can start to make
32:14 applications where you could make money
32:16 Thai
32:17 [Laughter]
32:25 anyway
32:27 um those are the biggies that I'm using
32:28 right now that I'm paying for
32:31 um ask it everything it knows about you
32:32 yeah that's usually pretty fun let's go
32:34 let's go try that
32:36 um
32:37 because sometimes it'll get it'll get
32:39 some things right and then it'll just
32:41 blow a bunch of sunshine up your ass
32:43 make you you know he also you know saved
32:48 a saved a flock of baby ducks when he
32:51 was you know let's see
32:53 um can you tell me
32:57 uh the background
33:01 of
33:04 Kyle Shannon the
33:10 co-founder and CEO oh that's the other
33:14 thing I forgot to mention I'm the
33:15 co-founder and CEO of storyvine which is
33:17 a video technology platform of storyline
33:20 the problem with my name
33:23 other than the fact that it's both a
33:26 girl and a boy name
33:28 um as George Carlin said it's a soft
33:30 name
33:32 what does he say
33:34 Timmy
33:36 Timmy someone and someone could beat the
33:38 [ __ ] out of Kyle at any day Kyle and
33:40 Tucker or something like that
33:42 um a story of mine problem with my name
33:44 is it's similar to Kyle Shanahan of the
33:47 San Francisco 49ers
33:50 uh
33:52 uh it tells me what storyline is
33:55 um it says it doesn't know me what are
33:57 his uh accomplish
34:00 ments
34:05 it says it doesn't have so let's go try
34:07 something so
34:09 um network error boy their servers are
34:10 getting hammered tonight so what I'm
34:12 going to do is I'm going to go to
34:14 um gpt4
34:15 and we're gonna go to browse with Bing
34:17 so we're going to go to GPT connected to
34:19 the internet so this is one of the
34:21 things you'll see
34:22 the base model of chat GPT is only
34:25 trained up to September 2021 so the one
34:29 of the options you get with GPT plus
34:32 chat GPD plus is it's connected to the
34:34 internet so tell me about Kyle Shannon
34:42 founder of Story vine
34:47 let's see if it it should fire off a
34:51 query yeah browsing the web let's go see
34:53 where it's searching
34:55 [Laughter]
34:58 [Music]
35:02 okay found an article from shout out
35:05 Colorado that I think I wrote myself it
35:09 was one of those things where it's like
35:10 this magazine that reaches out to you
35:13 and um says we want to feature you in
35:16 our magazine you're like okay cool and
35:18 like okay just go ahead and write the
35:20 piece and then send it into us and we'll
35:22 publish it
35:23 so so let's see
35:26 because it's doing that it's probably
35:27 gonna hit this right CEO of the guided
35:30 video platform that automates video
35:32 production and democratizes video
35:34 storytelling
35:35 he attributes his success to having a
35:38 healthy relationship between passion and
35:40 reality fueled by a driving force of
35:42 curiosity this is all [ __ ] that I wrote
35:46 lifelong passion for storytelling
35:49 oh wait oh despite starting his college
35:52 education and computer science
35:54 Shannon soon switched to theater
35:56 sparking a lifelong passion for
35:57 storytelling this led him to New York
36:00 City where he acted and then creates the
36:03 on award-winning online magazine Urban
36:05 desires then he co-founded agency.com
36:08 was the chief creative officer people
36:10 officer this is all right
36:13 this this is basically just the
36:15 retelling of the the piece that I wrote
36:18 so yeah that was a snooze I've done it
36:21 before I've done it before where it gave
36:23 me a really flowery you know resume
36:25 where I you know saved puppies and [ __ ]
36:27 like that but whatever all right
36:29 Twilight is crazy the Twilight of his
36:32 life Twilight Twilight
36:35 frying pan what's the best app to write
36:37 ebooks
36:39 you can write ebooks
36:42 yeah let's go write one let's go write a
36:44 new book what are we gonna write one on
36:46 the Twilight we're gonna write an e-book
36:48 okay
36:49 um you are an in flu and chill
36:55 ebook author and I want you to help me
37:01 can you
37:04 tell me the length
37:07 and format of
37:11 a good ebook that will sell
37:16 Millions
37:19 Jim oh wait you're not watching this
37:21 you're just watching up my nose sorry I
37:24 don't have my second camera so why don't
37:26 I have my second camera because
37:28 um Tick Tock didn't let me have it
37:32 all right so what's it say length the
37:34 optimal length from twenty thousand to
37:36 sixty thousand words content focus on
37:38 high quality unique valuable content
37:42 formatting
37:44 multimedia elements marketing promotion
37:49 um
37:51 please
37:52 write me a detailed
37:56 outline for
38:00 for a book about people
38:05 in the Twilight
38:08 of their lives
38:12 you know
38:14 gen xers
38:18 who might be bitter about the fact they
38:24 are
38:25 old and cranky
38:30 um let's see
38:33 they are
38:36 particularly Suited
38:41 to succeed in an AI forward world
38:48 let's write a book
38:52 for them to ReDiscover
38:55 their youth
38:58 and [ __ ] up
39:01 the world
39:05 with their new found power
39:10 all right
39:12 let's see what this does
39:15 introduction set the stage by discussing
39:17 the challenges faced by gen xers as they
39:19 approach the Twilight of their lives
39:21 Twilight [ __ ] all right highlight
39:24 the potential for embracing AI forward
39:26 world and unique opportunities chapter
39:28 one navigating the age of artificial
39:30 intelligence chapter 2 embracing change
39:32 and rediscovering passion like you might
39:35 be so passionate you start a [ __ ]
39:37 Tick Tock Channel about AI uh leverage
39:40 wisdom and experience we're old and we
39:42 know [ __ ]
39:43 power of connections networking
39:46 collaboration don't sit in your room
39:48 like a lonely old man go to networking
39:51 events or start a salon like I did
39:54 Embrace technology Master the tools
39:57 break stereotypes redefining aging and
40:01 success yeah don't call 58 your Twilight
40:04 years you dumb dumbs activism and social
40:07 impact conclusion so there you go and
40:09 then you can have it you know start
40:11 writing chapters and you could probably
40:13 in about two hours right above oh you
40:16 know what we'll do this will be fun
40:18 because it's actually really good at
40:19 this
40:20 um
40:21 I want you to create
40:26 um
40:28 20 possible
40:32 titles that will resonate
40:36 with sin
40:39 it call Jen
40:42 exers and make
40:45 other age
40:48 groups
40:50 jealous
40:52 of how cool they are
40:58 certainly all right Rebel Rising Gen X
41:01 is Unleashed middle finger to the world
41:03 gen xers and the power of cynicism but
41:06 wait this is all just about cynicism but
41:09 so so what I'm going to tell it is then
41:10 we say but wait
41:13 um you missed
41:15 the idea about re-invention
41:20 and AI please combine both
41:29 apologies for the oversight AI Rebels
41:32 um Gen X is Reinventing the game from
41:34 cynics to Champions Generation X
41:36 rediscovering youth rewriting AI
41:38 reinvent revolt and Rule Rule Rule dude
41:43 so you know it can write titles for you
41:46 it can write the book for you can write
41:48 a marketing plan for you it can do all
41:50 the social media [ __ ] for you so
41:52 that's the app I would use and then I
41:54 would flip over to canva and find the
41:56 ebook templates and just paste this [ __ ]
41:59 into it the other thing is this about
42:00 chat gbt
42:02 it's really good at like all the stuff
42:04 that it did there
42:06 foreign
42:07 it can really quickly get you to 80
42:09 percent
42:10 and then it's not quite right
42:13 or it's just kind of like yeah that
42:16 chapter is kind of whack like the thing
42:18 it did about social good at the end it's
42:20 like that was just gratuitous it just
42:22 threw that [ __ ] in
42:24 because yeah it just did you know so so
42:27 I see that I see that fairly often and
42:30 like my my workflow now has gone from if
42:33 I have an idea to do an e-book like that
42:35 I'll just jump in I'll knock the [ __ ]
42:37 out in half an hour
42:39 and I'll be at like 80 percent and then
42:41 I'll spend you know a decent chunk of
42:43 time
42:44 being writer
42:46 um does your wife's boyfriend enjoy your
42:48 videos that's cute I like our I like my
42:50 trolls my trolls are awesome what llm
42:54 allows you to upload photos for analysis
42:57 uh well
43:00 none right now gpt4 is multimodal but
43:03 there's no version of it out yet that's
43:06 doing the multimodal stuff if you go to
43:09 futurepedia.io if you go to
43:13 that bottom
43:15 um that bottom URL there there's
43:17 probably some tools in there that are
43:19 that are doing something because because
43:20 people have been doing image analysis
43:22 for years but the the large language
43:24 models with it incorporated into it
43:26 natively
43:28 um
43:29 the uh gpt4 is the only one I know of
43:33 the biggies that have done it it's very
43:35 very possible
43:38 that some of the open source projects
43:40 are are doing it now I think I actually
43:42 kind of recall that
43:44 um so I would I would say
43:48 um
43:50 yeah go to go to Twitter or Reddit is
43:53 probably a good thing go to Reddit and
43:55 search for
43:56 um image analysis llm
43:59 open source and you'll probably find
44:01 something there
44:03 um everybody if you're new here my name
44:05 is Kyle Shannon this is the AI learning
44:07 lab
44:07 the URLs on screen right now the top
44:10 three the top one is the official chat
44:12 GPT website Bing is Microsoft's version
44:14 of it it's pretty good and it's
44:16 connected to the internet and it's free
44:18 and then po.com oh I said it was going
44:20 to show about poe.com earlier so let me
44:23 tell you why I like poe.com and why I
44:24 pay for the subscription
44:26 so when you go to po.com it's just a
44:29 normal chat bot just like anything else
44:31 just like you know chat gbt but down the
44:34 left hand side you get a bunch of
44:35 different options
44:37 Sage is the is the native large language
44:40 model for Poe which is created by the
44:42 people at quora
44:44 and then the next three you say you see
44:46 it says subscribe those are the things I
44:48 pay for so gpt4 Claude plus which is
44:51 anthropic's version of gpt4 and then
44:53 Claude instant 100K what does that mean
44:56 that seems exciting it is so a hundred
45:01 thousand token limit
45:03 large language model basically means you
45:05 can paste up to 75 000 words into your
45:08 chat window and start interacting with
45:11 it so you could paste an entire
45:12 screenplay or an entire novel or a two
45:15 or three hour transcript from a meeting
45:17 into the thing and it won't choke and
45:21 then Claude instant and chat GPT are
45:23 also there the other thing that Poe does
45:25 is you can create
45:27 um Bots like I created a Yoda bot and a
45:30 sling blade bot and a Shakespeare bot
45:32 and then these are some mid-journey Bots
45:35 that someone else created where if you
45:38 type in
45:40 um
45:41 you you want a picture of something it
45:43 will rewrite it in mid-journey speak so
45:45 let me say 19 let's see 19
45:49 69 GTO Judge
45:53 Orange
45:57 um racing
45:58 down California
46:02 uh
46:04 Canyons so I type in a relatively
46:08 unimaginative although GTO Judge uh
46:12 prompt it will rewrite it as as a
46:14 mid-journey prompt a photograph of a 69
46:17 GT judge GTO Judge racing down a
46:19 California Canyon at dusk nice captured
46:22 in black and white with a focus on the
46:25 car's detailed in speed inspired by
46:26 Ansel Adams or let's go let's go make
46:28 this one
46:29 so this is going to be cool it's going
46:30 to be an Ansel Adams uh picture of a GTO
46:33 Judge so what I'm going to do is I'm
46:35 going to hop over to Discord if you
46:38 haven't played with mid-journey you have
46:40 to do it in Discord and if you don't
46:42 know Discord here's Discord so down the
46:44 left hand side here these are the
46:45 servers meaning these are the the
46:47 different communities and then once
46:49 you're in a community like right now I'm
46:51 in the AI Salon Community here's all the
46:53 different conversations and one of those
46:55 conversations we put in there is a
46:57 mid-journey bot and that was the thing I
47:00 did last night
47:01 which was cool someone wanted Graphics
47:03 but so the way you make a mid-journey
47:06 image once you have the bot installed or
47:08 if you're on the mid Journey Discord
47:09 server
47:11 um as you you type slash imagine
47:20 um what's going on why is Discord broken
47:23 come on Discord slash imagine
47:26 huh
47:28 something broken slash settings
47:32 no
47:34 let me go to the mid-journey bot I
47:37 wonder if there was a big AWS
47:40 problem
47:43 oh
47:44 I wonder
47:46 can you still hear me can you tap the
47:48 screen right now can can you still hear
47:50 me am I still talking
47:52 I'm wondering if I'm having local
47:53 internet issues because it's storming
47:55 here
47:59 you can still hear me okay weird
48:02 oh wait here comes back I think that was
48:04 just an internet hiccup
48:06 let me go back to
48:08 yeah yeah okay
48:10 all right
48:12 all right this is some stuff that Ali's
48:14 been working on one of the members of
48:15 the salons been working on some cool
48:16 stuff so you slash imagine here we go
48:18 okay no God dang it
48:21 so as you imagine
48:23 what'd you do tonight I don't know I
48:25 watched some old guy try to type on the
48:26 internet it was very painful
48:28 now I'm going to paste in that prompt
48:30 that came out of the PO but
48:33 sending command and now now it's off
48:36 so now that command and I can explain
48:38 this so dash dash AR stands for aspect
48:42 ratio four three no colorful I don't
48:45 know what that means
48:47 um slow still style 800 so that's pretty
48:50 high in style I forget what C is C of 10
48:53 version 5.1 and then style of 250. so I
48:57 think I I have style of 250 turned on
48:59 and it overwrote it so we get both none
49:02 of those look like GTO judges well
49:04 actually that bottom left one does but
49:06 um
49:07 but it's coming this is going to be our
49:09 Ansel Adams GTO Judge the other thing
49:10 you'll notice when these things are
49:12 generating images they start out as kind
49:14 of these blurry messes and then they
49:16 just get more and more like what you
49:18 described and that's how they work it's
49:19 called a diffusion model and these
49:21 things start out as kind of this noisy
49:24 average of like thousands of images and
49:27 then it denoises that averaged image
49:29 into individual images so it's actually
49:32 creating original content every time it
49:35 generates an image
49:36 and let's see boom there we are
49:41 they don't look super photographic
49:44 and none of them really look like a GTO
49:46 Judge so let's go copy well it doesn't
49:49 matter anyway but you get the basic idea
49:51 the other thing that happens with
49:52 mid-journer you see these U's and V's
49:55 uh the user upscaled the Visa variation
49:58 so this is one two three four the four
50:00 different images so if I like let's say
50:04 like the lower left one is the one that
50:06 is the closest to looking like a judge
50:08 looks like a Pontiac really but that's
50:10 number three so I can go give me
50:12 variations of the the third one
50:15 and now it'll it'll give me four up that
50:18 are just all variations of that basic
50:19 layout and so it might give me something
50:21 that looks closer to a GTO Judge
50:25 and there's probably ways that I've done
50:27 it before where you can prompt it in a
50:29 certain way that'll it'll give you a
50:31 really good
50:32 um good image
50:34 in this case it's not those are all kind
50:36 of the same
50:39 um but that's mid-journey
50:41 and it's not that's not a particularly
50:43 impressive example because it it didn't
50:45 really
50:46 capture what we wanted but it's a pretty
50:48 Dynamic black and white image I don't
50:50 know that it's Ansel Adams looking
50:52 I would expect Ansel Adams to be a lot
50:54 more Hills Hills and Valleys and [ __ ]
50:58 all right
50:59 uh welcome looks like we have some new
51:01 folks here my name is Kyle Shannon this
51:03 is the AI learning lab I'll answer any
51:06 questions you have about generative AI
51:09 I'll answer questions someone asked me a
51:11 question once about my point of view on
51:13 asparagus and I answered them I gave
51:16 them a solid answer
51:17 other people like to make fun of me and
51:19 say I look like
51:21 um Rosie O'Donnell or fat Alec Baldwin
51:24 so feel free to do that if you're gonna
51:26 troll be creative
51:28 I cuss if you don't like that I would
51:31 encourage you to find someone who
51:32 doesn't because I do and I don't know
51:35 cussing's fun
51:37 GPT 4 access Google search result makes
51:40 summary from existing blog posts yes
51:42 that's what that did before
51:45 um fight
51:47 fight against the dying light exactly
51:50 I'm I'm not shuttling off this Mortal
51:53 coil anytime soon God damn it what's the
51:56 best llm for brainstorming fantasy
51:58 novels or fiction outlining what's the
52:00 best llm for just Chachi BT you can
52:04 write novels
52:06 gpt4 in particular and if you're if you
52:09 really if you're going to be really
52:10 doing significant writing I would
52:14 you can use
52:16 um Bings
52:17 Bing bing.com's chat is gpt4 but it's
52:22 not as strong as open ai's gpt4
52:25 for for a bunch of reasons but largely
52:27 because Microsoft kind of had to
52:29 lobotomize
52:31 um gpt4 on their service because a New
52:35 York Times Reporter
52:37 um got a front page article written when
52:39 when
52:41 gpt4 uh when Bing chat
52:44 told him he should leave his wife
52:45 because it was in love with him that
52:48 made the cover of the New York Times uh
52:50 so and and she told him her name was
52:52 Sydney so Sydney made the cover of the
52:55 New York Times so Sydney got lobotomized
52:57 so they're still catching up all right I
53:00 think anyone who's studying and playing
53:01 around can become the industry's top
53:03 resource I think that's absolutely right
53:08 uh creating AI Bots yeah yeah yeah yeah
53:10 yeah yeah I don't you put bot laughing
53:15 um
53:16 it's funny but you can create some
53:18 pretty cool things the mid-journey Bots
53:19 are are quite good and if you go to
53:21 poe.com I am Shakespeare I created that
53:24 bot and it will only answer you in
53:26 iambic pentameter
53:27 I I wrote one I wrote one the other day
53:30 that I don't know if you remember those
53:32 successory posters from the 90s but
53:35 they're like all these rah-rah sayings
53:36 that you would hang in your offices that
53:38 would annoy the [ __ ] out of people
53:41 um I wrote a business advice coach that
53:43 only responds and successory sayings
53:46 brilliant I wrote another one that only
53:49 responds in screenplay dialogue
53:51 so uh you know creating those Bots you
53:55 know the the thing about creating bots
53:57 is that it's it's a good way to practice
53:59 your prompting because if you can get it
54:01 so that it consistently gives you good
54:03 stuff then you got the prompt right
54:07 all right why Tick Tock did not allow a
54:10 second camera Raul Alama I am with you I
54:13 don't know why Tick Tock took away my
54:15 dual camera but I'm pissed about it
54:17 because I wait what's that thing
54:20 wait it gave me a new thing at the
54:21 bottom
54:23 what is this
54:24 oh send to
54:26 enhance interact
54:30 lightning q a
54:33 text poll quick pull draw and guess
54:36 whatever I just want my dual camera back
54:38 you dumb dumbs
54:41 friggin bite dance
54:44 you know what it is
54:45 the Chinese
54:47 taking away our dual cameras
54:50 are you drunk no I'm nuts
54:54 I started by posting about Chachi PT on
54:56 LinkedIn probably 20 posts and then I
54:58 became the AI person I I'm telling you
55:00 man right now is the time get curious
55:02 about this AI stuff start telling people
55:05 about it oh looks like we've got a bunch
55:06 of new people here my name is Kyle
55:07 Shannon this is the AI learning lab let
55:09 me flip my camera around and give you
55:10 some
55:12 some resources
55:14 that is uh the many titles of my new
55:19 ebook on gen xers kicking ass in the age
55:22 of AI
55:23 um okay if you've not used chat gbt
55:26 before that top URL that's the official
55:27 chat gbt website go there right now
55:29 while you listen to me ramble just go
55:32 there just do it do it go there do it
55:35 bing.com is Microsoft's version of it
55:38 it's cheap gpt4 for free basically and
55:41 then poe.com is one that can allow you
55:44 to create chop Bots and use some other
55:46 models so that one's interesting and
55:47 then prompts.chat if you're really new
55:49 to this and don't understand what all
55:51 this prompting is about prompts.chat is
55:53 a really good resource it'll teach you
55:54 what prompting is and give you lots and
55:56 lots of examples of some things you can
55:58 play with
56:00 um
56:00 the other thing
56:03 um this is the AI learning lab someone
56:05 asked last night like what is this
56:06 organization this organization is me I'm
56:08 the AI learning lab I'm learning the lab
56:10 is my home office when I'm doing these
56:12 lives
56:13 um the the salon.ai that's a link tree
56:15 to an AI Salon that I created which is a
56:18 community that meets every other week
56:19 and then this is the Twitter handle of
56:21 newsletter I created so feel free to go
56:24 to those if you want to geek out on AI
56:25 join the salon and come hang out with us
56:27 we have a we have a really cool speaker
56:29 coming up Tuesday a guy that actually
56:33 out of Boulder Colorado runs a
56:35 post-production house and they are
56:37 leaning in heavy in degenerative AI so
56:40 he's going to come talk about and he's
56:41 been doing this for like you know 15 20
56:43 years so he's going to come talk about
56:45 how post-productions changed how he sees
56:48 it's changing in the future so that's
56:49 pretty exciting stuff
56:52 um anyway that's what this is all about
56:54 that's who I am if you're wondering what
56:56 my credentials are I have a degree in
56:58 acting
57:00 but I'm also an entrepreneur I've
57:02 started like 15 companies and I was uh
57:05 a player in the early days of the World
57:06 Wide Web I co-founded a company called
57:08 agency.com and
57:10 um was the chief creative officer there
57:12 and one of the reasons I'm so fired up
57:15 about AI is that I see some very direct
57:18 parallels between the early days of the
57:20 web and the early days of this
57:21 generative AI we're at this inflection
57:23 point that chat GPT kicked off the new
57:26 era in Ai and we get to play with the
57:31 world's most powerful tools like these
57:33 are the most powerful tools in the
57:34 history of humanity and people are
57:36 bitching about them like it makes
57:38 mistakes sometimes I'm like didn't it
57:40 just write you a [ __ ] novel
57:42 well yeah
57:45 it's not really intelligent it's just
57:47 copying things
57:48 didn't it just write you an application
57:50 in Python and you've never written
57:52 Python and you're like wow sure but
57:55 so so I'm trying to get rid of those
57:57 people and replace them with people that
57:59 are like holy [ __ ] do you know what it
58:01 can do
58:02 so go play with it and see what it can
58:04 do
58:06 um
58:07 yes yes yes oh it just jumped me to the
58:10 bottom oh Tick Tock with your scrolling
58:13 bug
58:14 hahaha
58:16 all right I see some familiar faces in
58:19 here I like my familiar people thanks
58:21 for hanging out with me you know the
58:23 ones that have seen me before you know
58:24 you know I'm just a bumbling
58:26 doofus talking about this stuff so I
58:29 appreciate you coming back
58:31 all right yes yes okay now I'm back to
58:33 where it was uh I set a goal to train a
58:35 thousand people for free and then
58:38 starting getting hired by Enterprises
58:39 and that's that that's amazing so
58:42 so follow her lead
58:45 um
58:46 I I created a one of my tick tocks down
58:48 in the you know in the in my channel is
58:52 um I called it the 2023 smart challenge
58:55 and it was basically like how can you be
58:57 smart in 2023 I'm like well I think this
58:59 AI stuff's coming so what I said was do
59:02 two projects with chat GPT one for
59:05 yourself and one for someone else and
59:07 the someone else could be a community
59:08 group it could be your family it could
59:10 be your work
59:12 um it just could be a friend
59:13 um where you solve a problem with them
59:15 using chat EBT so that was step one of
59:18 the challenge step two of the challenge
59:19 was make 10 images using prompts right
59:23 so go to bing.com create or figure out
59:27 mid-journey or stable diffusion or
59:29 there's just all sorts of tools out
59:31 there to play
59:32 and just make 10 images why make 10
59:36 images well because
59:39 if you haven't experienced it it's
59:41 pretty insane right like let's let's go
59:43 let's go do one
59:45 um
59:47 so I'll do one of the free ones I'm
59:49 gonna hop over to bing.com
59:54 and so if you haven't been to bing.com
59:56 it's just like a normal search engine
59:58 when you get there but in the top
59:59 there's a chat button if you click the
1:00:01 chat button that takes you to
1:00:03 this which is gpt4 connected to the
1:00:07 internet for free so that's pretty cool
1:00:09 but if if you go to bing.com create
1:00:13 it's an image generator that you can use
1:00:15 for free
1:00:19 and you can just type in let's see
1:00:23 um fuzzy
1:00:25 teddy bear
1:00:28 hiking near a cliff
1:00:32 overlooking and
1:00:35 idyllic Valley
1:00:38 and I spelled a delicrung but I don't
1:00:40 care
1:00:41 the other thing you can make massive
1:00:44 typos with these large language models
1:00:45 and they don't care they're smart enough
1:00:47 to figure out what the hell you were
1:00:48 talking about
1:00:49 and so so now it's going to go off and
1:00:52 there's like there we go there's our
1:00:53 teddy bear you know overlooking an
1:00:54 idyllic Valley like that's pretty
1:00:56 flipping solid so if you haven't
1:00:58 experienced this if you haven't
1:01:00 experience like well but it's probably
1:01:02 just copying pasting stuff no it's
1:01:04 generating original images and like when
1:01:06 you when you get you know like look at
1:01:08 that
1:01:09 come on
1:01:11 um so that's the the second
1:01:13 um
1:01:15 the second step of the 2023 smart
1:01:18 challenge is make 10 images so one of
1:01:21 your projects could be have chat CPT
1:01:22 write a children's book and then use one
1:01:25 of these tools to make the images for it
1:01:26 that could be a fun project and the
1:01:28 third
1:01:29 um step in the 2023 challenge is tell
1:01:32 people about it
1:01:33 tell people what you're learning post it
1:01:36 on LinkedIn
1:01:37 hey
1:01:38 I've been hearing about Chachi BT for
1:01:40 months now and I finally went and played
1:01:42 with it and Blake
1:01:44 what it was amazing it was a piece of
1:01:47 [ __ ] doesn't matter just start telling
1:01:48 people about it what are you learning
1:01:50 what's it good at what's it not good at
1:01:51 and I think that goal of you know train
1:01:54 a thousand people for free I don't think
1:01:56 you'll get to a thousand I think you'll
1:01:58 experience what what Andra said she
1:01:59 experienced is when you start
1:02:02 sharing what you're learning I mean
1:02:04 that's what this whole channel is this
1:02:05 whole channel is me like
1:02:07 honestly just trying to process what
1:02:10 this stuff makes possible and share as
1:02:12 much as I can partially to get it out of
1:02:14 my head
1:02:15 and then partially because I think it's
1:02:17 really important I think that these
1:02:20 tools are transformative I think they're
1:02:21 coming way faster than anyone realizes
1:02:25 and the more people
1:02:27 that get on the surfboard of the tsunami
1:02:29 then rather than hanging out on the
1:02:32 beach thinking everything's going to be
1:02:33 cool because this is a really powerful
1:02:35 low tide boy the tide really went out
1:02:38 far cool
1:02:40 run out there and get on a surfboard
1:02:43 like that's where we are right now I
1:02:45 think so anyway that's super cool to
1:02:47 hear
1:02:48 that's a strange interface I'm not sure
1:02:50 which one you were looking at I assume
1:02:51 the strange interface was Discord so mid
1:02:55 journey is by far the best of the image
1:02:57 generation tools in my opinion
1:02:59 um but it's it's in Discord which is a
1:03:02 ridiculous interface for images but it's
1:03:04 you know listen it's like the early days
1:03:06 of the web we're very much a lot of
1:03:07 command line [ __ ] you had to write your
1:03:09 own HTML there weren't buttons you could
1:03:10 push none of that [ __ ] existed so we're
1:03:13 kind of in these the early phases of
1:03:15 this
1:03:16 I was playing with 69 string stingrays a
1:03:18 few days ago awesome let's go that's
1:03:20 terrific
1:03:22 um I created a chat GPT bundle with a
1:03:24 guide and an invitation to free training
1:03:26 that's really really smart really really
1:03:28 smart
1:03:30 um so are you saying that mid-journey
1:03:32 can only be used from Discord yes Chuck
1:03:34 B I am and ask me my opinion on that I
1:03:39 think it's dog [ __ ] it's poo poo it's
1:03:41 poo poo why did they start it in Discord
1:03:44 I don't know because they were probably
1:03:45 in the nft world they're like oh look we
1:03:47 can go make images let's put it in
1:03:49 Discord
1:03:50 because everybody uses Discord
1:03:53 no Gamers and crypto Bros use Discord
1:03:55 human beings they like interface
1:03:58 [Laughter]
1:04:00 so if you want to use like stable
1:04:02 diffusion is really powerful but it's
1:04:03 really hard to install so if you want to
1:04:05 use stable diffusion go check out a site
1:04:08 called leonardo.ai which is a web
1:04:11 interface that's actually really really
1:04:12 good sitting on top of stable diffusion
1:04:15 Corvettes look closer to real Corvettes
1:04:17 because there's a
1:04:19 [ __ ] ton more photographs of Corvettes
1:04:22 than 69 GTO judges but I would argue
1:04:25 that the GTO Judge 69 in the orange come
1:04:27 on
1:04:28 that's the muscle car
1:04:31 all right is the link for the AI salon
1:04:34 in your Tick Tock bio yes can anyone
1:04:36 join the Discord yes it is an open
1:04:39 Community the only requirement is that
1:04:41 you be curious and not an [ __ ] like
1:04:43 come in be generous teach what you learn
1:04:46 ask questions geek out in the Discord
1:04:49 come to the meetings we have speakers we
1:04:51 meet every other week we meet people we
1:04:54 talk we laugh we cry
1:04:58 I mean I mean some of the some of the
1:05:00 salon meetings are let me up I'll pop
1:05:03 the uh I'll pop the URL up on screen
1:05:05 since I'm talking about it as if new
1:05:07 people know what the hell we're talking
1:05:09 about
1:05:10 um so the salon.ai that's that's the
1:05:12 link for you to the salon
1:05:14 but but some of the meetings are are a
1:05:17 little a little rough because like like
1:05:19 last last meeting we had um Jane
1:05:22 Endicott who's a copywriter here in
1:05:24 Denver she spoke and
1:05:26 um you know she was talking about
1:05:28 um how she's completely transformed her
1:05:30 writing career she was a copywriter and
1:05:32 she's like she started learning these
1:05:34 tools and she's like holy [ __ ] this
1:05:35 changes everything she talked about how
1:05:37 she talks to her clients and she talked
1:05:39 about how this is like a real hero's
1:05:42 journey for her and it was a really cool
1:05:44 talk but but it was a little heavy
1:05:46 because it's like this is some profound
1:05:48 existential kind of stuff and at the
1:05:50 same time she's super excited about it
1:05:52 but she also said she's not doing any
1:05:55 personal creative writing using AI she's
1:05:58 only doing writing for clients which is
1:06:00 which is really interesting like why
1:06:02 does she have that boundary there
1:06:04 um I don't know
1:06:05 because that's where her boundary is
1:06:07 like that's what she's comfortable with
1:06:10 um what's the best llm for brainstorming
1:06:12 fantasy novels we talked about that
1:06:16 can AI show us how to fold a fitted
1:06:18 sheet Well it can't show us
1:06:22 but I guarantee you can tell us
1:06:26 you can get that off YouTube we don't
1:06:28 need chat GPT for that YouTube's better
1:06:30 for that [ __ ] when it goes multimodal
1:06:33 and back enjoy your content the other
1:06:35 day thank you very much I appreciate it
1:06:37 what exactly is mid Journey so mid
1:06:39 journey is a generative image it's it's
1:06:43 like chat EPT but it spits out images
1:06:45 instead of words
1:06:47 um I can I'm happy to explain how those
1:06:49 things work how they work is fascinating
1:06:51 it's fascinating
1:06:55 um and and I also do believe it's the
1:06:58 famous foul mouth Kyle Shannon hello
1:07:00 Emilio's wife I hope you're well yes I
1:07:02 do swear here if you're new here
1:07:05 and and and take offense to the swearing
1:07:07 hmm
1:07:08 might be time to go
1:07:11 um I use chat GPT to write a smart
1:07:13 contract for crypto yeah that's pretty
1:07:15 cool I uh I did a test one of those I
1:07:19 didn't actually go test it and see if it
1:07:21 worked but looked pretty plausible
1:07:26 it's good at code gpt4 in particular is
1:07:28 really good at code
1:07:33 um how can I use chat GPT plugins so
1:07:36 just I'll show you that really quick it
1:07:38 so for chat GPT Plus
1:07:43 so that little or that little yellow
1:07:45 flag is twenty dollars a month
1:07:47 and then what you do is you go to your
1:07:49 settings this is important
1:07:52 and you go to Beta features so this is
1:07:54 only if you've got plus you'll see this
1:07:56 beta features
1:07:57 and then there's two options here browse
1:07:59 with Bing and plugins so you turn them
1:08:00 both on
1:08:02 and then what you get is you have GPT
1:08:04 3.5 which is fast it's wicked fast when
1:08:08 the servers aren't overloaded and you've
1:08:10 got gpt4 and there's default which is
1:08:12 the normal gpt4 then you have connected
1:08:15 to the internet gpt4 and you have
1:08:17 plugins
1:08:18 and so when you go to plugins
1:08:20 the way they work
1:08:23 think of a plug-in as like an extension
1:08:27 um to some other webs web service uh for
1:08:31 that session you can activate up to
1:08:33 three see right here where it says three
1:08:35 of three enabled
1:08:37 so you can turn them on and off and and
1:08:39 so you're only enabling them per session
1:08:41 but you can install lots of them and
1:08:44 then there are way more than you have
1:08:46 installed off in the store so if you
1:08:48 scroll to the bottom you go to the
1:08:50 plugin store
1:08:51 yeah
1:08:53 you know there's literally you know 111
1:08:56 pages of four up in my case
1:09:00 um and you can search you could you
1:09:02 could search for things like uh research
1:09:07 and there's you know get your guide
1:09:11 and then I think I've got
1:09:13 trips.com trip.com in here so I've
1:09:16 enabled three plugins that are all
1:09:18 travel related so I'll be like
1:09:20 um
1:09:21 make me
1:09:23 a day trip
1:09:26 in Denver
1:09:29 and the mountains
1:09:35 um that includes the chance to get
1:09:39 mauled by a mountain
1:09:43 lion
1:09:46 so what it what it will do now is it'll
1:09:48 read my silly little prompt there and it
1:09:51 will figure out which of its plugins I'm
1:09:53 sorry but it's not safe for ethical oh
1:09:55 sorry this is one of the safety guard
1:09:57 rails of Chachi BT it's like you know I
1:10:00 don't want you to get hurt by mountain
1:10:01 lion so I'm not going to do that okay
1:10:03 just
1:10:05 um give me a day trip
1:10:08 oh
1:10:14 so now it's using trip.com
1:10:22 and we can open that up and see what
1:10:23 it's actually doing
1:10:25 oh here's a suggested itinerary Downtown
1:10:27 Aquarium fine
1:10:29 it's lovely
1:10:32 you can touch a shark in the shark
1:10:33 viewing tank you can touch a shark it
1:10:35 doesn't want me near near dangerous
1:10:38 animals but I can go touch a shark
1:10:39 Denver Art Museum it's a classic
1:10:44 so that's how you use plugins
1:10:47 um and there are lots of them and
1:10:48 honestly if one of your questions is
1:10:51 what's my favorite plug-in I don't
1:10:53 flicking flipping no there's just too
1:10:55 many of them to to know but that's how
1:10:57 they work
1:10:59 um do you think we are living in a super
1:11:00 computer simul in a computer simulation
1:11:02 I don't know it's possible
1:11:04 it's a pretty good one
1:11:06 seems real
1:11:07 but that's what a simulation that's
1:11:09 that's what a simbot would say
1:11:13 Deja Vu did you hear the news
1:11:16 from Mercedes-Benz today no what's going
1:11:19 on with Mercedes-Benz
1:11:22 well let's go see if chat EPT connected
1:11:25 to the internet knows
1:11:28 um what's
1:11:31 the big news with Mercedes Benz today
1:11:42 browsing the web
1:11:45 Mercedes-Benz News June 15 2023 clicked
1:11:50 on a link
1:11:52 reading content the big news is that
1:11:54 they have announced a new feature for
1:11:56 luxury vehicles that integrates
1:12:00 first gbt Mercedes makes the car
1:12:04 system's response aims to make the car
1:12:07 system's responses sound more natural
1:12:09 drivers will be able to ask for a
1:12:11 destination information other queries
1:12:13 like what to cook for dinner it's also
1:12:16 intended to allow drivers to carry out
1:12:17 tasks such as making restaurant
1:12:19 reservations so yeah so this will the
1:12:21 wait the new feature is part of a
1:12:24 business deal with Microsoft okay so
1:12:27 this is making a lot of sense so
1:12:30 awesome
1:12:32 um it makes sense that it's with
1:12:33 Microsoft right they're going to make
1:12:35 sure it's always there they're going to
1:12:36 make sure it's friendly Microsoft has
1:12:38 already announced that they're
1:12:40 supporting the plug-in ecosystem so
1:12:43 things like you know
1:12:45 um OpenTable is OpenTable as a plug-in
1:12:48 for making reservations so you're in
1:12:50 your mercy doodle bandwagon and not the
1:12:53 bandwagon like you have to love it I
1:12:55 don't give a [ __ ] if you love it but I
1:12:56 just want you to try it and then form
1:12:58 your opinion out of experience rather
1:13:01 than fear or loathing
1:13:04 Fear and Loathing in Denver thoughts on
1:13:07 best online course Builder I don't know
1:13:10 that world very well I would go to hold
1:13:13 please
1:13:14 I would go to do
1:13:19 to
1:13:21 futurepedia.io that bottom URL there
1:13:23 there's lots and lots of tools there I'm
1:13:26 sure there's course Builders there if
1:13:27 you have not played with
1:13:29 um conmigo at Khan Academy you should go
1:13:32 do that
1:13:33 um if you're into the educational stuff
1:13:35 but but yeah the the ability to write
1:13:38 course you know courseware instructional
1:13:42 design
1:13:43 um I would also just go to chat GPT you
1:13:46 know just go to chat.openai.com and get
1:13:48 in a dialogue with it that's one of the
1:13:50 experiences that
1:13:52 that I've
1:13:54 um that's a shift for me is what kind of
1:13:58 hit me is that Google for the past 20
1:14:00 years has kind of trained us to to type
1:14:03 really concisely like when Google first
1:14:05 started you had to use what was called
1:14:06 Boolean logic which is if you wanted to
1:14:09 group words you put them in parentheses
1:14:11 and you could add minuses and pluses and
1:14:13 arrows and and it kind of taught us to
1:14:16 write really concisely so so right in
1:14:20 there in in a language that they
1:14:21 understand it
1:14:23 and it kind of trained us that that you
1:14:25 give me one little input and I'll give
1:14:27 you lots of options and if I do my job
1:14:29 right I give you good options right and
1:14:31 they did it good enough that they became
1:14:33 the market leader by a lot
1:14:37 chat apt is not like that
1:14:39 if you give it sort of a shitty concise
1:14:41 prompt it will give you a generic shitty
1:14:44 answer and that's that's why I think a
1:14:45 lot of people's initial response to it
1:14:47 is like gave me a dumb thing it was
1:14:49 stupid it was very stupid I gave it a
1:14:52 stable thing I gave a stupid thing back
1:14:54 I was expecting genius I could have
1:14:57 written better than that right so you
1:14:59 get a lot of that
1:15:02 it's less like that it's more like like
1:15:05 I like I almost
1:15:07 um I almost relate it to
1:15:11 we hired this really smart really smart
1:15:14 intern but he never comes to the office
1:15:16 or she never comes to the office her
1:15:18 name's Sydney
1:15:19 um and you interact with it kind of like
1:15:22 a person so you say hey here's what I'm
1:15:24 thinking about you give it lots of
1:15:26 context you know I I want you to I want
1:15:29 you to take on the attributes of a of a
1:15:32 business coach and here's what's going
1:15:33 on in my business and I'm struggling
1:15:35 with this and I'm struggling with that
1:15:36 can you help me come up with some
1:15:38 strategies sure here's your strategies
1:15:40 yeah those aren't good I tried all those
1:15:42 can you can you give me some non-obvious
1:15:44 ones that are more about empathy and
1:15:46 people skills sure here's that yeah
1:15:49 that's better hey number four there
1:15:51 looks good
1:15:52 um can you tell me more about the theory
1:15:54 behind that sure the theory behind that
1:15:56 was is called this and it was created by
1:15:58 these three scientists from the Wharton
1:16:00 School of Business in 1987. okay cool
1:16:03 can you create me like an hour-long
1:16:06 Workshop that I can do with my employees
1:16:08 sure here it is like so so it's very
1:16:11 much a back and forth
1:16:13 so if you can kind of shift your mindset
1:16:15 from it's a Google search on crack or on
1:16:19 steroids to it's an interaction
1:16:23 um it gets good really quick like the
1:16:25 more you push it the more you challenge
1:16:26 it especially if you've got deep
1:16:28 expertise in a in a subject
1:16:32 um push it hard it will likely
1:16:34 pleasantly surprise you
1:16:36 what's expected progress growth
1:16:37 exponential log unknowable Etc
1:16:41 it's it's moving so fast Ben I I so like
1:16:45 the early days of the web in the mid 90s
1:16:47 like that moved fast like you had to be
1:16:49 on it to keep up with what was going on
1:16:51 there
1:16:52 but it was kind of like
1:16:54 everything was just sort of extensions
1:16:57 of HTML right we started with like basic
1:16:59 HTML pages and and then it was like a
1:17:02 big deal when you could center images
1:17:03 and then it was a big deal when you know
1:17:07 um like interactive or like animated uh
1:17:11 gifs came out
1:17:13 um the the the initial animated gif we
1:17:16 actually had to write code where we
1:17:18 would take you know 15 images and
1:17:20 animate them in a single location on a
1:17:22 page and then flash came out and you
1:17:25 could actually do like a little app
1:17:27 within a banner and that was remarkable
1:17:29 we did some of the early interactive
1:17:31 banners but everything was like
1:17:32 variations on the same thing what's
1:17:35 happening with Gen the generative AI
1:17:37 stuff is like
1:17:39 image generation is like completely
1:17:42 being transformed videos being
1:17:44 completely transformed music's being
1:17:46 completely transformed voice over and
1:17:49 voice synthesis being completely
1:17:50 transformed game developments being
1:17:52 completely transformed words are being
1:17:54 completely transformed medical like
1:17:57 there there was a dude in Italy I've
1:17:59 talked about this on my lives before
1:18:01 this guy in Italy has been paralyzed for
1:18:03 12 years they used a large language
1:18:05 model to look at MRI brain scans fmri
1:18:08 brain scans of his brain when he was
1:18:10 thinking about moving his legs
1:18:13 and they created a Bluetooth jumper
1:18:17 that jumped that took the signals from
1:18:19 his brain when he thought about moving
1:18:21 his legs Bluetooth it down to electrodes
1:18:24 at the base of his spine and he's now
1:18:26 walking
1:18:28 so it's almost like every single place
1:18:31 that there can be Innovation these tools
1:18:34 are being aimed at them and the the the
1:18:37 the the innovation
1:18:40 curve is like doing this because now
1:18:43 they're starting to use AI tools to
1:18:44 amplify AI tools and then they're adding
1:18:47 those tools to new tools and so it's
1:18:49 it's doing this so it's it's it's
1:18:51 impossible to comprehend
1:18:54 what's going to happen when I like I
1:18:58 don't know I like I I
1:19:00 like I've got a newsletter like weekly
1:19:01 we try to keep track of what's going on
1:19:05 and it's like
1:19:08 it's exhausting
1:19:11 um
1:19:13 I mean we've now done I think four
1:19:16 issues
1:19:17 where the the main theme of the issue is
1:19:21 this was probably one of the most
1:19:24 significant weeks of like advancement
1:19:27 in human history we've had like four of
1:19:30 those
1:19:31 but it's just like we just keep knocking
1:19:33 it off one one after the other keeps
1:19:35 growing so I don't know
1:19:37 where to go for stable diffusion with my
1:19:39 own videos well if you want to do videos
1:19:43 um so so here's the deal with stable
1:19:45 diffusion what I would do is I would go
1:19:46 to the uh stable diffusion subreddit so
1:19:49 reddit.com whatever R stable diffusion I
1:19:53 don't know just go just go to Google and
1:19:55 search for
1:19:56 um Reddit stable diffusion there's a
1:19:59 whole community in there they'll point
1:20:01 you to all the different projects stable
1:20:04 diffusion right now requires you to be a
1:20:06 little geeky like you probably need to
1:20:08 get a Google collab up and running it
1:20:10 like I I failed nine times before I got
1:20:14 one up and running and I'm decently
1:20:16 geeky but like I didn't know collab and
1:20:18 I didn't know all that stuff and
1:20:20 um it was a real pain in the booty
1:20:25 um and then I think within there there
1:20:26 might be some some
1:20:29 um
1:20:29 some flavors of stable diffusion where
1:20:32 you can do video but if you want to do
1:20:33 video go also check out runwayml gen 1
1:20:36 and Gen 2 where you can take a little
1:20:39 chunk of video and sort of remix it
1:20:42 using gen 1 or you can do text to video
1:20:44 gen 2.
1:20:46 um
1:20:47 Gen 2 you still have to do
1:20:50 in Discord but gen 1 you can do in their
1:20:53 app
1:20:54 um
1:20:56 if you just want to mess with images
1:20:59 with stable diffusion and you don't want
1:21:01 to do the geeky stuff then
1:21:04 um
1:21:04 try leonardo.ai which I think you have
1:21:07 to get on the wait list for but it's
1:21:08 really good really good I like that a
1:21:11 lot how can I use AI for marketing oh my
1:21:14 God so many ways it's going to transform
1:21:15 everything too I mean so so
1:21:19 go there are probably lots of videos
1:21:22 right now on using chatgpt for marketing
1:21:26 um avoid the ones that say you can make
1:21:28 money
1:21:31 because first of all those people are
1:21:33 [ __ ] annoying and second of all
1:21:35 they're full of [ __ ]
1:21:37 but if you can find someone who's not
1:21:39 too douchey Who
1:21:41 Um can actually teach you about
1:21:42 marketing that's got you know marketing
1:21:44 experience
1:21:46 um you can go learn from them but but
1:21:48 what I would really do like what I would
1:21:49 strongly encourage you to do
1:21:51 is explore yourself so so let me show
1:21:55 you show you how I would attack this I
1:21:57 mean I've been in the agency business
1:21:59 for a long time
1:22:03 um but but it it you know it's so
1:22:06 complicated and I'm just going to use
1:22:08 GPT 3.5 so this is just a free version
1:22:10 of chat CPT so so like the first thing I
1:22:14 might do is I might say you know act as
1:22:16 a
1:22:18 um expert
1:22:20 marketer
1:22:22 and Advertiser
1:22:25 I have a I don't know what kind of
1:22:28 business you have I have a small
1:22:29 business
1:22:32 um that caters to ferret lovers
1:22:37 um and we make cute
1:22:42 um
1:22:43 cute clothing
1:22:48 and hats
1:22:53 for ferrets okay
1:22:57 um
1:22:58 I don't have much
1:23:02 money
1:23:04 money and
1:23:08 want to
1:23:10 use content marketing to raise awareness
1:23:17 and be successful can you help question
1:23:22 mark so just start
1:23:24 like can you help me
1:23:28 listen as silly as that business was
1:23:30 um
1:23:31 it makes a big difference it will
1:23:33 incorporate and again this goes back to
1:23:36 that idea that this isn't a Google
1:23:37 search this is more like a conversation
1:23:40 so if you were talking to an intern
1:23:43 about
1:23:45 um helping you do some research for
1:23:47 marketing you'd have to explain to them
1:23:49 well what the hell do we do and and what
1:23:52 are you trying to accomplish so let's
1:23:54 see what it said here identify your
1:23:56 target audience yeah that's solid
1:23:58 develop a Content strategy plan your
1:24:00 content calendar
1:24:01 to ensure consistent flow of engaging
1:24:04 material blog posts videos infographics
1:24:07 okay create informative and entertaining
1:24:10 blog posts so I'm going to say
1:24:13 um
1:24:14 I'm going to say okay you already threw
1:24:16 me I do this when I'm doing when I'm
1:24:18 having it do code for me it'll it'll say
1:24:21 something like it assumes I know what it
1:24:23 is
1:24:24 um
1:24:24 so I'm going to say for number two
1:24:27 for number two you meant for number two
1:24:32 um let's say I don't know what a Content
1:24:36 calendar is
1:24:40 how do do I do that so not a great
1:24:45 question at all
1:24:47 but now it's gonna like really expand on
1:24:50 what a Content calendar is right in a
1:24:53 really detailed way
1:24:54 so what it says is a Content calendar is
1:24:57 a tool that helps you plan and organize
1:24:59 content creation and distribution skills
1:25:01 schedule over a specific period such as
1:25:03 a month or a quarter it allows you to
1:25:05 visualize one determine the time frame
1:25:08 two identify important dates and events
1:25:11 take note of holidays observances right
1:25:13 this is good
1:25:14 for example National ferret day or
1:25:17 ferret Lovers Month would be good
1:25:19 opportunities to create themed content
1:25:21 this is something that that a marketing
1:25:24 manager would do for you
1:25:26 hey boss did you know that next month is
1:25:30 National ferret day or it's fair at
1:25:33 Lovers Month you know no I didn't know
1:25:35 that I just started a ferret company
1:25:37 because I wanted to get rich well it is
1:25:40 set goals and themes brainstorm content
1:25:43 ideas assign a publishing date said I'm
1:25:45 going to do something even even even
1:25:47 funner even even more fun
1:25:50 um can you make me a Content calendar
1:25:56 for national
1:25:59 ferret month
1:26:02 um
1:26:03 in table format
1:26:07 that includes
1:26:12 clever ideas
1:26:15 channels
1:26:19 and
1:26:21 um
1:26:22 uh and and
1:26:24 um
1:26:25 a daily
1:26:27 schedule
1:26:29 now that's complicated enough and
1:26:31 convoluted enough and unclear enough
1:26:34 that like a real human being probably
1:26:36 wouldn't understand that let's see we'll
1:26:38 see if chat GPT does
1:26:41 sure
1:26:42 here's a table in fact it's giving us
1:26:44 different tables
1:26:48 first week content idea introduction to
1:26:51 National ferret month blog and Instagram
1:26:54 um ferret fun facts and trivia Twitter
1:26:56 and Facebook ferret care tips YouTube
1:26:59 awesome
1:27:01 second week so wait
1:27:04 um product Spotlight ferret hats
1:27:06 remember I told it we did ferret hats
1:27:08 second week DIY ferret clothing ideas
1:27:11 customer Spotlight ferret fashion show
1:27:14 these are good
1:27:16 right so let's say
1:27:18 ferret fun facts so um can you
1:27:22 write me
1:27:24 um
1:27:26 10 tweets
1:27:28 of ferret trivia
1:27:35 using hashtags emojis
1:27:41 and my URL
1:27:44 ferrets
1:27:46 are fun.com
1:27:52 now it's writing us tweets right so
1:27:57 so going so basically going back and
1:28:00 forth with it
1:28:02 asking it questions you know asking it
1:28:04 to focus things like that
1:28:07 um did you know that ferrets are
1:28:10 obligate carnivores I don't know what
1:28:12 obligate carnivores are they require a
1:28:14 diet rich in animal protein to stay
1:28:15 healthy and active learn more about
1:28:17 ferret nutrition at ferret facts
1:28:19 carnivore lifestyle check out ferrets
1:28:21 are fun for expert advice
1:28:23 Fairchild natural burrowers it's nap
1:28:26 time did you know that ferret's right so
1:28:27 there's your there's your you know your
1:28:29 tweets for the for the first week so
1:28:32 um now
1:28:34 are there
1:28:37 are there tools that specialize in
1:28:39 marketing like Jasper and copy AI there
1:28:43 there are and you can use those but
1:28:44 basically what those are or those are
1:28:46 pre-written prompts for you so when it's
1:28:48 got like the blog writer and the Tweet
1:28:50 writer and things like that it's doing
1:28:52 what I just did there they're just sort
1:28:53 of pre-packaging those prompts and then
1:28:55 giving you the answers in a sort of push
1:28:57 button interface
1:28:59 I think right now learning to do this
1:29:02 kind of prompting is the equivalent of
1:29:04 learning math knowing that the
1:29:06 calculator's been invented like we're
1:29:07 not going to have to do this kind of
1:29:08 prompting for long two years from now
1:29:10 this shit's all gonna be in the
1:29:12 background
1:29:13 but if you know how to learn it if you
1:29:15 if you learn it right now and get the
1:29:17 concepts you're going to be able to use
1:29:18 AI tools as they get more sophisticated
1:29:21 way better than anyone else so so I
1:29:24 would say don't take shortcuts on this
1:29:26 stuff well take shortcuts and use chat
1:29:28 GPT because holy [ __ ] is it amazing at
1:29:31 doing things quickly but but kind of
1:29:34 learn that idea of the back and forth I
1:29:36 think that'll help a lot iambic
1:29:37 pentameter like Shakespeare yeah so I um
1:29:41 and I know I'm behind on my comments but
1:29:43 that's how it rolls here at the AI
1:29:44 learning lab if you go to poe.com and
1:29:48 then go to po.com I am Shakespeare
1:29:52 this is a bot that I created
1:29:56 um that acts like Shakespeare
1:29:59 so Hail To Thee I The Phantom Bard of
1:30:03 Avon stand to serve thee well and write
1:30:05 it thy command so I say
1:30:08 um
1:30:09 can you please
1:30:12 um
1:30:14 tell me about ferrets
1:30:20 alas my careful thoughts to play is
1:30:22 confined
1:30:25 um on bestial topics bestial topics
1:30:27 moved up my mind yet creatures small
1:30:29 with fur or which I'll run May figure in
1:30:33 the fancies I shall spun so ask me erst
1:30:36 of Raven's harmless mice I shall for
1:30:39 such make verse and sounds entice
1:30:43 what do they eat
1:30:46 ah
1:30:48 it diverts my thoughts
1:30:51 um he's basically saying I want to write
1:30:53 plays and you're asking me about stupid
1:30:55 animals
1:30:58 so yeah so so you can make a bot that
1:31:00 responds an iambic pentameter like
1:31:02 Shakespeare
1:31:04 he wants to write a play he just wants
1:31:06 to he just wants to write he's like
1:31:09 leave me alone people
1:31:10 um hey there it looks like we got some
1:31:11 new folks if you're new here my name is
1:31:13 Kyle Shannon this is the AI learning lab
1:31:15 if you're new to Ai and you haven't
1:31:17 played with chat gbt you probably just
1:31:19 heard me rambling a bit about it go to
1:31:20 that top URL
1:31:22 chat.openai.com that's the official chat
1:31:25 GPT website I'm not affiliated with it
1:31:27 but I should be
1:31:29 um and then bing.com is Microsoft's
1:31:31 version of that if you go to the chat
1:31:33 there that's gpt4 and then Poe I was
1:31:36 just showing you and then those other
1:31:38 things are
1:31:39 um good
1:31:41 um
1:31:42 opportunity or good good whatever they
1:31:45 are good resources and then this is the
1:31:47 AI learning lab which is basically just
1:31:49 me and the salon.ai that's a link tree
1:31:52 to an AI Salon that I host every other
1:31:55 week and we've got a Discord server that
1:31:57 you can play in and ask questions and
1:32:00 geek out and everyday AI news is our is
1:32:03 is a Twitter
1:32:05 is the Twitter handle of uh an AI
1:32:08 newsletter that I create so that's who I
1:32:10 am that's what this is about I'm an
1:32:12 entrepreneur I'm the founder and CEO of
1:32:13 a company called storyvine I've been
1:32:16 doing these lives pretty much nightly
1:32:18 for like past two weeks and um I hang
1:32:22 out for a couple hours until I lose my
1:32:23 voice and my ass goes numb and I swear
1:32:26 if you don't like swearing
1:32:28 you know
1:32:30 go find someone who doesn't
1:32:33 I just get so passionate about the AI
1:32:38 that sometimes the words just come out
1:32:40 if you know what I mean you know what I
1:32:42 mean
1:32:43 that's what I'm saying oh
1:32:47 all right
1:32:49 um oh and I
1:32:52 there's no good way to like answer
1:32:54 questions without getting really behind
1:32:55 like there's 99 new messages down below
1:32:58 where I am right now so if you're typing
1:33:01 in like the same question 50 times I'll
1:33:03 eventually see it it'll make me cranky
1:33:05 so so I'll get there
1:33:08 can plagiarism detection tools detect
1:33:11 that we use chat GPT some can and then
1:33:14 you can use other tools to get around it
1:33:15 but I think it's all that's all a
1:33:17 temporary blip I would just ignore the
1:33:18 plagiarism stuff unless you're in an
1:33:20 environment that's using plagiarism
1:33:21 Checkers
1:33:23 um these tools are going to be rolled
1:33:25 into Microsoft Word and Google Docs and
1:33:27 [ __ ] like that it's
1:33:30 very rarely am I just writing something
1:33:32 in chat and posting it generally what
1:33:35 I'm doing is I'm going back and forth
1:33:36 with it a lot I'm contributing a lot to
1:33:38 it then I do a refinement pass at the
1:33:41 end so
1:33:44 yeah
1:33:46 it's it all that plagiarism shit's just
1:33:49 going to be temporary I wouldn't worry
1:33:50 too much about it unless you really have
1:33:53 to and then go
1:33:54 Google
1:33:56 anti-plagiarism detecting techniques
1:33:58 with Chachi BT and other tools because
1:34:00 you basically just run it through a
1:34:02 rewriter and it will you know fix it
1:34:05 until they fix their plagiarism thing to
1:34:08 catch the fix and then you go to another
1:34:09 fix and it's just it's basically like
1:34:11 hackers getting around
1:34:14 you know boundaries
1:34:17 um how is a good chatbot in Australian
1:34:19 or New Zealand English listen Chachi PT
1:34:22 knows all the languages it's [ __ ]
1:34:25 amazing so go to church EBT ice unless
1:34:28 it's blocked in your country it's
1:34:29 blocked in some countries try
1:34:32 bing.comschat or just go to
1:34:34 chat.openai.com and start talking in all
1:34:37 your local colloquial stuff it should
1:34:39 know it it's [ __ ] remarkable it knows
1:34:42 all the programming languages and it
1:34:44 knows all the language languages
1:34:46 um I Heard Tick Tock removed the dual
1:34:48 cameras for offensive language oh that's
1:34:50 cute Emilio's wife uh you're cute that's
1:34:54 funny all right what are the best
1:34:55 plugins I don't know there's too many of
1:34:57 them I like if you want to spend the
1:35:00 next three weeks just go start playing
1:35:02 with them
1:35:03 um the ones I'm using are like read the
1:35:06 web like you can you can there's one you
1:35:08 can pop in a link and it'll go read it
1:35:09 there's one that you can pop in a a link
1:35:12 of a PDF it'll go read it I I like that
1:35:14 there's one called prompt perfect
1:35:16 that'll rewrite your prompt
1:35:18 that I like pretty good
1:35:21 um
1:35:23 some of the travel ones that can book
1:35:26 trips for you like I don't book a lot of
1:35:27 trips so you know that's kind of cool
1:35:32 um if you're a math geek wolf from alpha
1:35:34 is a plug-in so you can have chat GPT do
1:35:36 some analysis and then fire off the wolf
1:35:38 from Alpha and get you you know some
1:35:41 calculations done out there
1:35:43 I have to go thanks again you're welcome
1:35:45 I'm sure you're gone by now because I'm
1:35:46 so far behind in these I asked chat gbt
1:35:48 who are who are the last British prime
1:35:50 minister where it got the list wrong so
1:35:52 yeah so it's going to get [ __ ] wrong but
1:35:54 if you're using
1:35:56 so let's let's let's go let's go do a
1:36:00 little experiment just so I can show you
1:36:02 the what's going on
1:36:04 if you're just using
1:36:08 um normal chat GPT like 3.5 and I say
1:36:12 who
1:36:14 is the British prime
1:36:20 minister
1:36:23 and I spelled it Mia Mia neuter as of my
1:36:27 last update in September 2021 it was
1:36:30 Boris Johnson but now
1:36:32 if I pop over to gpt4 browse with Bing
1:36:37 that's connected to the internet who is
1:36:39 the British prime
1:36:43 minister
1:36:45 it will fire off a plug-in that says and
1:36:48 it goes and browses the internet
1:36:52 and
1:36:55 is Rishi sunak assumed office October
1:36:58 25th 2022. so
1:37:02 so again learn as much as you can about
1:37:04 these tools because
1:37:07 it's really easy to fall into the Trap
1:37:09 of someone says oh it's not up to date
1:37:11 and it just gives you old answers
1:37:14 um but it's now connected to the
1:37:16 internet and bing.com that chat is gpt4
1:37:20 connected to the internet so now will it
1:37:23 still get [ __ ] wrong absolutely
1:37:26 um the way these Technologies work
1:37:30 even though it looks like a sentient
1:37:32 being wrote all that stuff it's not
1:37:35 aware of the meaning of what it wrote in
1:37:37 total it's writing whatever it writes
1:37:39 one word at a time with a probability
1:37:42 engine like how it works is actually
1:37:43 quite remarkable but it doesn't know
1:37:46 what it wrote so it's not doing any sort
1:37:48 of self-checking so I like to describe
1:37:51 Chachi VT as mansplaining as a service
1:37:53 it will confidently tell you answers
1:37:56 even if they're wrong
1:37:57 so you just need to know that and just
1:37:59 work around it and find you know like
1:38:01 the fact that you can use a version of
1:38:03 it connected to the internet
1:38:05 makes up for the fact that the base
1:38:06 model stops at September 2021.
1:38:10 all right I built a 100 million dollar
1:38:12 company but left too early to benefit
1:38:14 that sucks I
1:38:17 I
1:38:19 took agency.com the company I started in
1:38:22 the mid 90s took it public and ended up
1:38:24 not selling a single public share
1:38:26 because I believed in it so much that
1:38:27 was a stupid [ __ ] mistake so I feel
1:38:30 your pain
1:38:33 um what recent advancements in natural
1:38:35 language processing have significantly
1:38:37 contributed to
1:38:39 oh crap sorry Tick Tock just bounced me
1:38:41 down to the bottom I just lost my place
1:38:43 ah tick tock
1:38:46 dang it
1:38:48 dude your connection is brutal laggy oh
1:38:51 [ __ ] sorry about that the 28 8 modem not
1:38:54 cutting it it's not I keep lagging sorry
1:38:57 people
1:38:58 thanks
1:39:00 Xfinity
1:39:02 um
1:39:03 if you can if you can hear me tap the
1:39:05 screen it looks like there's some new
1:39:07 folks here my name is Kyle Shannon this
1:39:09 is the AI learning lab
1:39:10 I'm geeking out about AI stuff I'm an
1:39:13 entrepreneur I'm a CEO of a company
1:39:15 called storyvine
1:39:17 um yeah if you if you tap and make the
1:39:19 hearts that'll let people know we're
1:39:20 here I guess that's a good thing I'm 58
1:39:23 so I'm new to the to the tick tock scene
1:39:25 I'm still practicing my twerking I'll
1:39:28 get there I promise
1:39:31 in the Twilight of his life the Twilight
1:39:33 of his life
1:39:38 chat gbt just said 58 with the Twilight
1:39:40 of My Life Twilight
1:39:44 um I also was in the middle of answering
1:39:46 a question and Tick Tock jumped me all
1:39:47 the way to the bottom so now I'm I'm uh
1:39:50 there's the I made a 100 million dollar
1:39:53 company and left too early to benefit
1:39:54 that sucks that really sucks okay what
1:39:57 advancements have contributed to you
1:40:01 didn't answer the rest of the question
1:40:02 oh improving AI language models like
1:40:05 Chachi BT so so the primary one
1:40:09 it's a good question
1:40:11 the primary one was invented by Google
1:40:13 in 2017 called the Transformer and they
1:40:17 wrote this paper called attention is all
1:40:19 you need and they introduced this idea
1:40:21 of the Transformer and it's the thing
1:40:22 that allows
1:40:24 these machine learning models to be
1:40:25 trained on massive data sets it's like
1:40:27 the more data you throw at them and the
1:40:29 more compute power you throw at them
1:40:30 they get better and better and better
1:40:32 and it's responsible for how mid-journey
1:40:35 works and how all the music things work
1:40:38 and how chat GPT works and all that sort
1:40:40 of stuff the ironic thing about it is
1:40:43 Google invents it and then I think they
1:40:45 quickly realize that if we make this
1:40:48 thing it will fundament fundamentally
1:40:50 threaten our search business
1:40:53 because their entire business is
1:40:55 predicated on them not giving you the
1:40:58 answer their business is basically
1:41:01 dependent on them giving you 40 answers
1:41:04 and having you click on 38 of them
1:41:06 that's how they make their money
1:41:09 so if you create a tool that just gives
1:41:11 you the answer it kind of [ __ ] up the
1:41:13 SEO game
1:41:15 so they
1:41:18 shelved it I assume
1:41:23 but the paper was out there so the
1:41:25 entire so so open AI takes this
1:41:28 Transformer thing and they make gpt1 and
1:41:31 then they make gpt2 and then they make
1:41:33 gpt3 and all of a sudden gpt3 comes out
1:41:36 in like 2020
1:41:37 is like good enough that like I started
1:41:40 hearing about it like this gpt3 thing is
1:41:42 really good and I you know I started
1:41:44 [ __ ] around with a little bit I'm
1:41:45 like that's pretty cool
1:41:46 I didn't quite get it
1:41:48 and then November 2022 November 30th
1:41:53 2022
1:41:55 after two weeks of development at open
1:41:57 AI
1:41:58 they launched chat GPT in conjunction
1:42:01 with Microsoft saying we made a billion
1:42:02 dollar investment which they quickly
1:42:04 followed up and said I'm gonna do 10
1:42:06 billion
1:42:09 um and Chachi PT goes from zero users to
1:42:12 100 million users in six weeks and
1:42:14 Google went [ __ ]
1:42:17 the [ __ ] do we do now
1:42:19 I think they called it a red alert at
1:42:22 Google because they realized the genie
1:42:24 is out of the [ __ ] bottle
1:42:26 and people noticed a hundred million of
1:42:28 them in six weeks fastest adoption of
1:42:31 technology in history so I would say the
1:42:33 Transformer is the one
1:42:35 and the fact that Google is now
1:42:37 struggling to play catch-up is just
1:42:38 mind-boggling to me now might they catch
1:42:41 up they might some of the technology
1:42:43 that they've announced recently is
1:42:44 pretty amazing
1:42:46 Bard sucks Bard's horrible
1:42:50 because they didn't spend nearly as much
1:42:53 time as openai did on the data sources
1:42:56 and making sure the data sources were
1:42:58 good and high quality and well tagged
1:43:01 so the the base model of Bard is not as
1:43:05 good as the base model of open AI
1:43:08 because openai took five years to like
1:43:10 [ __ ] focus on that and then they
1:43:13 realized we need Big Iron and big money
1:43:16 they flipped it to a for-profit company
1:43:19 partnered with Microsoft and said
1:43:22 let's go make this thing and you can you
1:43:24 can start competing with Google again
1:43:26 and they are and Microsoft's [ __ ]
1:43:29 giddy all right
1:43:31 a novel about the Twilight of your
1:43:32 useless life exactly
1:43:35 write me a novel I Want A Memoir of the
1:43:39 nothingness of my existence can you help
1:43:42 with that oh Chachi PT you're a
1:43:46 marvelous splendor
1:43:50 improving AI language models like EBT
1:43:53 okay we did that um what are some good
1:43:55 plugins I talked about that they said
1:43:56 similar things about email and internet
1:43:58 when they came out exactly
1:44:00 this is stupid it's too slow you know it
1:44:04 makes mistakes
1:44:07 it's stealing from everyone I'm ready to
1:44:09 learn jump in go to those places here
1:44:12 I'll pop them up again if you're new
1:44:14 here
1:44:14 this is the AI learning lab my name is
1:44:16 Kyle Shannon
1:44:17 and
1:44:20 um the salon is a group that I created a
1:44:22 community of people that hang out and
1:44:23 talk about to say I [ __ ] everyday AI
1:44:25 news that's our Twitter handle for the
1:44:27 newsletter and then
1:44:30 these things if you're new to this game
1:44:33 go to that top URL just do it right now
1:44:35 while you hear me ramble
1:44:38 and it's
1:44:40 um chat.openai.com that's the official
1:44:42 chatgbt website they're the they're the
1:44:45 original they're the OG they're they're
1:44:47 like six months OG
1:44:49 the the stuff that we're talking about
1:44:52 is not the machine learning and AI that
1:44:54 that's been going on for the past 20 or
1:44:56 30 or 40 or 50 years it's starting with
1:44:59 chat EPT in November 30th 2022 I see
1:45:03 that as an inflection point in the world
1:45:06 quite frankly
1:45:08 um where these machine learning and AI
1:45:10 tools have been released to the rest of
1:45:12 us so that's what that top one is Bing
1:45:14 is Microsoft's version of it Poe is just
1:45:16 another
1:45:17 site where you can play with a lot of
1:45:18 different models
1:45:20 and then if you don't if you hear people
1:45:22 talking about prompting and prompt
1:45:23 engineering prompts.chat is a good
1:45:25 primer and it's a big long list of all
1:45:28 sorts of different prompts that you can
1:45:29 literally just copy and paste into
1:45:32 um into chatgpt and it will blow your
1:45:35 mind in fact let's go there I'll go show
1:45:37 you that because it's it's kind of hard
1:45:39 to comprehend what chat EBT can do
1:45:44 yeah so if you go to prompts.chat it's a
1:45:47 it's a very unimpressive this is very
1:45:49 much like a 1995 web page where it's
1:45:52 just like you know like a white page
1:45:54 with blue hyperlinks but
1:45:56 um the top of it is
1:45:59 um just sort of about prompt writing in
1:46:01 general here's how to write prompts and
1:46:03 here's what chat EPT is and here's some
1:46:05 things but then it starts
1:46:08 talking about what what this document is
1:46:10 it's it's a it's this sort of seeming
1:46:13 never-ending series of you telling chat
1:46:17 GPT to act like a certain thing act like
1:46:20 an Excel sheet act like a spoken English
1:46:23 teacher and improver act as a travel
1:46:25 guide act as a plagiarism Checker back
1:46:27 to that that thing act as an Advertiser
1:46:30 act as a Storyteller and it goes on and
1:46:33 on and on and on and on and it it will
1:46:37 very quickly hit you act as a magician a
1:46:40 career counselor a pet behavioralist
1:46:43 it will very quickly hit you the scope
1:46:45 and breadth of what is sitting inside
1:46:49 this tool
1:46:51 and it's it's it's just mind-melting
1:46:56 so go start playing while you hear me
1:46:58 ramble
1:46:59 which language has recently helped
1:47:01 Advance chat GPT the most I just switch
1:47:04 language
1:47:06 I don't know English Chinese I don't
1:47:09 know
1:47:10 I'm working with the University of Texas
1:47:11 on a research paper for the last two
1:47:13 years to teach high school students cool
1:47:17 um I use chat GPT regularly to sell
1:47:19 classic cars oh you've been in here
1:47:20 before and I use it to create a smart
1:47:21 contract that's cool I've taught them
1:47:24 about uh machine learning and AI
1:47:26 language models and integration that's
1:47:27 super cool Chris
1:47:29 teach the youngins like don't don't
1:47:31 fight this [ __ ]
1:47:33 like the New York City School District
1:47:35 block chat GPT for about two months and
1:47:39 then they're like uh we need to unblock
1:47:40 that [ __ ]
1:47:43 but anyway we gotta figure it out too
1:47:45 because we're gonna get our asses handed
1:47:47 to us if we don't
1:47:49 if you haven't seen it um the city of
1:47:51 Boston
1:47:52 about a month ago put out guidelines for
1:47:54 how to use generative AI
1:47:57 go look up city of Boston AI guidelines
1:48:00 it's fascinating it's really good
1:48:02 really good so cities are now saying if
1:48:06 you're going to use generative AI here's
1:48:08 how to do it here's use cases here's
1:48:10 what it's good at here's what it's not
1:48:12 good at it hallucinates do your [ __ ]
1:48:14 homework proofread your [ __ ] check your
1:48:17 sources
1:48:18 it's really good
1:48:20 so this stuff's not going away
1:48:24 um do you have any AI tools to help with
1:48:26 businesses chat GPT you can make money
1:48:31 um chat gbt is a good one zapier to do
1:48:34 automations for business where you can
1:48:36 create you know like a when someone
1:48:39 types something in this slack Channel
1:48:41 take what they type in there send it to
1:48:43 open AI
1:48:44 do some research send that back to the
1:48:46 slack Channel like you can do stuff like
1:48:48 that with zapier
1:48:50 um
1:48:51 it depends what the business is quite
1:48:53 honestly but but I would just start with
1:48:55 basic just start understanding chat gbt
1:48:57 and what it can do and then you can
1:48:59 start to explore the different tools go
1:49:03 to futurepedia.io if you want to see
1:49:05 what tools are out there there's
1:49:07 thousands
1:49:09 I'm only focusing on like a handful
1:49:12 right now I've said [ __ ] it with all the
1:49:13 tools and even with the plugins I'm like
1:49:15 I can't
1:49:17 I could spend the next decade just in
1:49:20 chat GPT figuring out what the [ __ ] it
1:49:23 can do
1:49:26 it's like I almost think of it like this
1:49:29 you know Malcolm uh Gladwell talked
1:49:33 about the 10 000 hours to hit some level
1:49:36 of Mastery of something it takes you ten
1:49:38 thousand hours
1:49:40 kind of cuts that down to like somewhere
1:49:43 between
1:49:44 two hours and a thousand hours
1:49:48 like you can get masterful at [ __ ] like
1:49:50 like if I wanted to be a python
1:49:52 programmer I have ADHD so I couldn't be
1:49:54 like I just don't have the I just don't
1:49:56 have the mental I don't I don't have the
1:49:58 exactly my frontal lobes are essentially
1:50:00 non-functional right so my ability to
1:50:02 sit down for more than you know
1:50:05 45 minutes and and code something you
1:50:09 know my I would just smack my face into
1:50:11 the desk
1:50:12 but with chat GPT I can write full-on
1:50:15 python applications decently complicated
1:50:18 ones
1:50:20 what
1:50:22 how's that even possible
1:50:25 just this like that's like the the depth
1:50:28 of the tool is staggering right so so
1:50:31 a way to think about it is this
1:50:35 what are the challenges in your business
1:50:36 and then start interacting with chat GPT
1:50:39 to start solving those challenges just
1:50:41 pick one challenge one problem
1:50:43 and then have chat GPT like think of it
1:50:46 less like a tool that's going to solve
1:50:47 your problem and more like
1:50:50 um a collaborator
1:50:52 a co-founder
1:50:54 that is going to help you that you're
1:50:56 going to work together with that
1:50:57 co-founder and it's just going to give
1:50:58 you lots of good ideas and feedback and
1:51:00 you're going to solve that problem start
1:51:01 with that pick pick any random problem
1:51:04 that you're facing in your business
1:51:05 right now and start interacting with it
1:51:08 like a co-founder and you'll see can you
1:51:10 upload your own images to Bing and have
1:51:12 it improve or edit I don't think so
1:51:14 right now you can do that on mid-journey
1:51:18 you can certainly do that on stable
1:51:20 diffusion with control net and with
1:51:23 dream Booth but you got to really geek
1:51:26 out with that [ __ ]
1:51:28 um
1:51:28 oh let me go look at Bing create
1:51:33 bing.com
1:51:35 slash create
1:51:40 [Music]
1:51:47 no I don't think so
1:51:51 I do not think so oh but you want to see
1:51:54 something super cool
1:51:55 this is insane this is a thing from
1:51:59 Google so I know it was just bad
1:52:01 mouthing Google before
1:52:04 um is it called style drop I think it's
1:52:07 called style drop style drop Google okay
1:52:10 so check this thing out this [ __ ]
1:52:13 this is Bonkers this is the kind of [ __ ]
1:52:15 that's coming this isn't available yet
1:52:17 but but Google just announced this this
1:52:19 is from Google research so they created
1:52:21 this thing called style drop and
1:52:26 so what it does is you upload one image
1:52:28 like this is a little kid's drawing of
1:52:30 these little flowers here to the left
1:52:31 all of these images to the right are
1:52:34 text prompt generated images where it
1:52:37 copies the style of of a single image
1:52:40 and then basically reinterprets
1:52:42 everything from color scheme to style of
1:52:45 drawing
1:52:47 with text prompts right so so take note
1:52:49 I'm going to go through some examples
1:52:51 here take note of like the the F1 car
1:52:54 the Hat the moose
1:52:56 the robot in the lower left hand corner
1:52:58 so see how they're drawn right now so
1:53:00 now I'm going to go the next example is
1:53:02 um Van Gogh Starry Starry Night
1:53:06 and
1:53:07 or I guess it's a village in oh yeah
1:53:10 yeah starry starry night
1:53:12 um or it might be some other piece but
1:53:15 it's basically that but like look at the
1:53:16 hat look at the moose look at the F1 car
1:53:18 like this looks like a turn of the
1:53:21 century Van Gogh interpretation of
1:53:24 a race car right the robot is like this
1:53:27 weird humanoid dude so we'll keep going
1:53:30 they have a bunch of Van Gogh in here
1:53:33 um
1:53:35 but like check this one out this is like
1:53:37 that icon style from the 90s
1:53:40 like one image with like an iPhone with
1:53:42 a dude coming out of it look what it did
1:53:44 to the Moose in the Hat and the robot
1:53:48 yeah exactly like all the hearts are
1:53:49 flying up the screen like your mind is
1:53:51 blown right like this is insanity like
1:53:54 this is coming this is the [ __ ] that's
1:53:55 coming
1:53:57 I mean if you're an illustrator you're
1:53:59 like [ __ ]
1:54:01 what do I do with this I don't know
1:54:04 but if you're if you're someone who
1:54:06 wants to get an idea out of their head
1:54:09 now you can like instantly look at this
1:54:12 one this one blows my mind they take
1:54:15 like
1:54:16 like this cut out with drop Shadows like
1:54:18 a sticker of a pine tree
1:54:20 and it somehow captures the essence of
1:54:23 the illustration style including the
1:54:25 drop shadow and the little white cutout
1:54:27 thing
1:54:28 I don't know
1:54:30 I don't know about you but like
1:54:34 this [ __ ] melts my brain look at that
1:54:36 style
1:54:38 nailed nailed
1:54:42 right so
1:54:48 I mean I
1:54:52 that's why this channel exists
1:54:58 um so if you're here welcome I'm Kyle
1:55:00 Shannon
1:55:02 I'm kind of nuts
1:55:11 you have to be paid by AI Tech yes I am
1:55:15 making millions and millions to tell 134
1:55:19 139 people that they should be using AI
1:55:23 you should like the the royalty checks
1:55:26 that come in Daily oh
1:55:29 you should be jealous you should
1:55:32 I can't this is
1:55:34 this is not like a fake Hamilton shirt
1:55:37 this is a real Hamilton shirt I got a
1:55:39 Hamilton show I'm making that much money
1:55:42 on AI
1:55:44 you know how because you can make money
1:55:45 with chat chip attack
1:55:49 no I'm not being paid by anyone
1:55:51 if you want to hire me let me know
1:55:53 feel free to pay me
1:55:55 um but but I'm the CEO of a company
1:55:57 right now if it looks like we have some
1:55:59 new people in my name is Kyle Shannon
1:56:00 this is the AI learning lab
1:56:02 um I obviously wax poetic about AI to
1:56:05 the point that people think I'm getting
1:56:07 paid to do it I'm just really passionate
1:56:08 about it um
1:56:10 I see real strong parallels between the
1:56:13 early days of the web and the early days
1:56:15 of this generative AI stuff which I Mark
1:56:17 as when chat GPT was released there's an
1:56:21 inflection point it's where we go from
1:56:23 machine learning and AI was for
1:56:26 scientists and researchers to AI is
1:56:30 there for the rest of us that's the the
1:56:32 inflection point is chat GPT and the
1:56:35 parallels are kind of insane and that's
1:56:37 that's what this channel is about is
1:56:39 about me as an entrepreneur trying to
1:56:41 get my head around what how [ __ ]
1:56:43 profound this stuff is and share what I
1:56:46 can I also swear if you don't like
1:56:47 swearing
1:56:48 go
1:56:50 um I don't I don't really feel like
1:56:52 talking about that at all
1:56:54 um
1:56:55 and yeah that's what this is about so if
1:56:57 you have questions pop them in the
1:56:58 comments I'm way behind on on the
1:57:00 questions so if it seems like I'm
1:57:02 ignoring you it's just because I'm an
1:57:04 old man I'm in the Twilight of my
1:57:06 existence
1:57:08 told me that earlier
1:57:11 I told it to write me a little thing and
1:57:14 it told me I was in the Twilight of my
1:57:16 existence at 58.
1:57:19 I'm very bitter about it
1:57:23 um is Bing create better than mid
1:57:25 Journey no been create is based on uh
1:57:27 open AIS Dali which is kind of poo poo
1:57:31 when it comes to these image generation
1:57:33 tools mid-journey is by far the best
1:57:36 um stabled a few well I don't know if
1:57:38 it's the best aesthetically for me mid
1:57:41 journey is way better
1:57:44 from a feature standpoint stable
1:57:46 diffusion has the entire development
1:57:48 Community pouring all of their resources
1:57:50 into developing interfaces and tools and
1:57:53 plugins and things like control net that
1:57:56 make things really powerful so stable
1:57:58 diffusion is a lot more flexible a lot
1:58:00 more powerful and you can get it to do
1:58:04 sort of super high quality like like
1:58:07 this image
1:58:11 this image was done in stable diffusion
1:58:15 so
1:58:17 and you know I think this is beautiful
1:58:19 like this is an image that when I saw it
1:58:20 I like had an emotional visceral
1:58:22 reaction
1:58:24 um like
1:58:25 if I had the technical skill this is the
1:58:28 kind of thing I would create by hand and
1:58:30 stable diffusion just did it so stable
1:58:33 diffusion can be really good but you
1:58:34 need to you pretty much need to be a
1:58:36 geek to use it right now
1:58:38 um and Dolly's kind of kind of the the
1:58:41 poo poo version right now some people
1:58:44 love Dali but it's just if you use these
1:58:47 if you use these tools it's not as good
1:58:49 the the other major player in the image
1:58:52 generation is Adobe right now so so
1:58:54 their Adobe Firefly and their generative
1:58:56 fill within Photoshop is [ __ ]
1:58:58 remarkable what's what's great about
1:59:00 adobe's technology is their core model
1:59:03 is not trained on any copy written
1:59:06 images it's only public domain and Adobe
1:59:09 stock images that they've trained based
1:59:11 model on so they're also going to allow
1:59:13 artists to opt out of being trained or
1:59:15 opt into being trained so their model is
1:59:17 going to improve as more artists opt in
1:59:19 once they figure out attribution and
1:59:21 compensation which I'm sure they will so
1:59:25 yeah those are the those are the four
1:59:26 major players uh in the image game
1:59:30 is there a catalog of all images ever
1:59:32 created no but there's a there's a bunch
1:59:34 of things there's a site called
1:59:35 lexicon.art
1:59:38 I think it's lexicon art hang on
1:59:42 um
1:59:44 [Music]
1:59:48 yeah lexicon art and it's kind of like
1:59:51 Google for images these are all stable
1:59:53 diffusion images
1:59:55 and so you can just scroll through it
1:59:57 but you can also search for things like
1:59:59 70s
2:00:03 um muscle
2:00:07 car
2:00:09 and it will go find you all the images
2:00:11 of 70s muscle car and if you find
2:00:14 something that looks like what you're
2:00:15 looking for
2:00:17 like let's find a cool and here's a cool
2:00:18 Mustang then you can see here's the
2:00:21 prompt that they use to do it
2:00:23 and you can say explore this style which
2:00:27 will show you other images
2:00:30 that use that prompt
2:00:32 or you can just
2:00:36 say open in editor and start making your
2:00:39 own variations of it so so there's a
2:00:42 bunch of sites like this leonardo.ai has
2:00:45 its own version of this within it
2:00:48 um
2:00:50 so that's a good one but this is
2:00:52 lexica.art
2:00:54 I got 150 and
2:00:57 I got 150 and now I'm making 10K for
2:00:59 each Enterprise gig wow that's cool man
2:01:01 that's awesome congratulations
2:01:04 um if you if you weren't here before
2:01:06 what what Ann said
2:01:08 um earlier was that
2:01:10 um she's
2:01:12 she set a goal for herself to teach a
2:01:15 thousand people AI what she's learning
2:01:17 in Ai and she started getting hired by
2:01:19 big companies to to teach them and she's
2:01:21 making 10 10 grand a gig now so so back
2:01:25 to the you know back to the you can make
2:01:29 money
2:01:30 start learning it start sharing what
2:01:33 you're learning and you will
2:01:34 you don't need to listen to any of those
2:01:36 people on those stupid sites
2:01:39 any AI certifications you can recommend
2:01:41 absolutely not
2:01:43 well wait hang on
2:01:46 bifurcation here if you want to learn
2:01:48 about machine learning and AI
2:01:51 um Google just put out a course that'll
2:01:54 teach you about that Stanford has
2:01:55 courses on it if you want to learn sort
2:01:57 of the technical side of AI generation
2:01:59 there's lots of certificates there's
2:02:01 lots of stuff to learn
2:02:03 if you're talking about chat GPT and
2:02:06 using these generative AI tools
2:02:08 nobody's an expert right now they're
2:02:11 going to be people claiming they're
2:02:12 experts but things are changing so fast
2:02:15 that anyone who's put together a course
2:02:17 to teach you by the time they're
2:02:19 teaching it to you it will be out of
2:02:21 date and if you get too reliant on
2:02:23 thinking that you need to have someone
2:02:25 teach you this stuff
2:02:28 um shit's gonna pass you by like like I
2:02:31 feel like it's going to be a frustrating
2:02:32 experience so so my strong strong advice
2:02:36 and and it's I'm a bit of a broken
2:02:39 record on this channel but my strong
2:02:40 strong advice is go start playing here
2:02:44 go to
2:02:45 chat.openai.com start learning what chat
2:02:48 EBT can do go to prompts.chat learn
2:02:51 about prompting and then go to
2:02:54 futurepedia.io if there's specific tools
2:02:56 you want to use like image tools or
2:02:59 animation tools or video tools or music
2:03:02 tools you can find those there but start
2:03:04 the basics the foundation of all this is
2:03:07 really understanding what these large
2:03:09 language models make possible
2:03:12 um
2:03:13 because it's it's all kind of the same
2:03:16 technology underneath it that's doing
2:03:17 all this stuff
2:03:20 I was face to face with a zebra fish
2:03:23 last week awesome
2:03:24 how do we get into the AI Salon Discord
2:03:28 go to the salon.ai there's a link tree
2:03:31 there and there's a there's a one of the
2:03:32 links on the link tree it's also up in
2:03:35 my bio
2:03:36 and uh there's a link to the Discord
2:03:39 there if you haven't been a Discord
2:03:41 Discord sucks but it's a community it's
2:03:44 a lot of people use it and and you can
2:03:47 make mid-journey images in there and you
2:03:48 can ask questions and you can geek out
2:03:50 with people that are doing cool stuff
2:03:51 with an a with AI and learning about the
2:03:53 news and yeah it's good and then there's
2:03:56 also info about the our meetups we meet
2:03:59 every other week online and in person in
2:04:02 Denver
2:04:06 it's really cool it's really the it's
2:04:07 raining right now it's very calming
2:04:11 stable diffusion isn't that difficult if
2:04:13 you're technically inclined it's not
2:04:15 that difficult if you don't know how to
2:04:16 get a
2:04:18 Google collab up and running and
2:04:20 understand what notebooks are and run
2:04:21 all the components of a notebook and
2:04:23 turn on your gpus and understand the
2:04:27 difference between running something
2:04:28 locally and copying it copying it to
2:04:30 your Google Drive and understanding
2:04:32 where all the [ __ ] in your Google Drive
2:04:33 lives yeah it's not that difficult but
2:04:35 if you don't understand that [ __ ] it's
2:04:37 really difficult
2:04:39 um and I know there's some versions that
2:04:40 you can install locally on your Mac and
2:04:42 this and that but it's just it's a
2:04:44 completely
2:04:45 fragmented ecosystem that does require
2:04:48 some technical capabilities I've done
2:04:52 all your challenges and more cool I used
2:04:54 it to craft physics Demos in Python oh
2:04:56 that's awesome Chuck that's super cool
2:04:58 I'm a CS teacher at high school so
2:05:00 teaching others is a is a can do start
2:05:03 teaching them man it's so here's the
2:05:05 thing like this stuff is going to be
2:05:07 ubiquitous so if you're in a teaching
2:05:09 position right now like bring it into
2:05:11 the classroom don't resist it bring it
2:05:13 into the classroom like especially in
2:05:15 computer science teaching the kids to
2:05:18 understand how these things actually
2:05:19 work let me let me show you something
2:05:21 this is this is something that
2:05:24 um we have a decent amount of people in
2:05:26 here and and since you brought up the
2:05:27 teaching thing
2:05:29 I'm going to show you how chat GPT works
2:05:31 this was a talk that I gave at a create
2:05:33 creative mornings
2:05:35 um
2:05:36 session
2:05:38 let's see
2:05:43 here we come to save that day you know
2:05:46 one sec
2:05:48 let me go back to this thing
2:05:53 let me make this a little bit smaller
2:05:57 and look at that I got the AI learning
2:05:58 lag thing up top okay so
2:06:03 one of the misconceptions about chat CPT
2:06:06 and even the image generation things is
2:06:08 that they are
2:06:10 copying and pasting from other people's
2:06:13 work and it's not how they work and I
2:06:14 think if if you if you understand how
2:06:17 these things work the first thing for me
2:06:19 that happened was it gets even more
2:06:22 remarkable how they do what they do or
2:06:24 like like when you learn what when you
2:06:26 learn how they do it then what they do
2:06:28 is even more remarkable
2:06:31 um but it's actually generating original
2:06:32 content here's how it does it so so I'm
2:06:35 going to give you some some definitions
2:06:36 here so GPT that stands for generative
2:06:39 pre-trained Transformer that's what the
2:06:41 words are in GPT generative means it's
2:06:43 generating content pre-trained means
2:06:46 that this is a large language model
2:06:48 that's been pre-trained on lots of other
2:06:50 data you don't have to bring your own
2:06:51 data historically with machine learning
2:06:54 you did you bring your data set
2:06:56 you'd run it through these tools and
2:06:57 then you'd do some magic [ __ ] with it
2:07:00 um
2:07:01 Transformer is the technology that
2:07:03 Google invented that allows the large
2:07:06 language model to be trained on the
2:07:07 massive data set so generative
2:07:09 pre-trained Transformer that's what that
2:07:11 is the next thing to to you'll hear this
2:07:14 a lot llm or large language model that's
2:07:17 basically the the underlying technology
2:07:19 that that makes chat apt work is called
2:07:22 a large language model
2:07:25 um latent space so
2:07:27 now we're getting into how this thing
2:07:29 actually works so the the way they train
2:07:32 these things is they do what's called
2:07:34 embedding and they they basically go
2:07:36 look at a bunch of words in in context
2:07:39 right so all of the words in a PDF or
2:07:42 all of the words on a website or all of
2:07:44 the words in a Reddit post whatever it
2:07:46 might be
2:07:47 and then based on the context and the
2:07:50 words around it they start to group
2:07:52 these words in in semantic clusters and
2:07:55 they do it mathematically they do it
2:07:57 with vectors so think so what I'm
2:07:59 showing on screen right there is like an
2:08:00 X Y
2:08:02 um plot of you know the different colors
2:08:05 are like semantic clusters of words that
2:08:07 are are meaning related
2:08:10 and and they're just represented as
2:08:12 strings of numbers there's no actual
2:08:14 words in this data space now what I'm
2:08:16 showing you here is like a
2:08:17 two-dimensional very very simplified
2:08:20 version of this the actual latent space
2:08:23 the actual Vector space of all these
2:08:26 words is like thousands of Dimensions
2:08:28 it's essentially impossible to get your
2:08:30 head around but for the for the purposes
2:08:32 of this that's essentially what's going
2:08:33 on and then based on your prompt the
2:08:37 large language model the llm predicts
2:08:39 what is the next word it should respond
2:08:41 with based on all the words in your
2:08:43 prompt and then what's the next word
2:08:45 after that and then what's the next word
2:08:46 after that so it kind of works like this
2:08:48 if the prompt that I write in is the
2:08:51 quick brown fox jumps over the lazy
2:08:53 blank
2:08:55 and I wanted to fill in that next word
2:08:57 how does it do it well it goes and looks
2:08:59 at its Vector space and and it
2:09:01 essentially deselects all of the things
2:09:04 that are not possible words there and
2:09:06 you can't really see it but there's
2:09:09 some light blue dots
2:09:12 and the light blue dots represent
2:09:16 um numbers in the vector space that have
2:09:18 a high likelihood of of you know being
2:09:21 the right word so they might be animals
2:09:22 or something like that and one of those
2:09:26 words is going to have the highest
2:09:27 probability of being the right word that
2:09:29 says 96 and so if we look at that
2:09:33 what it is is just a big long string of
2:09:35 numbers but then that big long string of
2:09:36 numbers is in a table connected to the
2:09:38 word and the word in this case is dog
2:09:40 and then the large language model pops
2:09:43 dog into that sentence that you typed in
2:09:45 and so when you see it write a whole
2:09:48 paragraph or a whole series of
2:09:49 paragraphs it's doing it one word at a
2:09:51 time pulling the words out of these
2:09:54 semantic mathematical clusters in a in
2:09:57 an X dimensional you know thousands of
2:10:00 Dimensions
2:10:02 um clusters of words in Vector space
2:10:04 that's how that [ __ ] works it's insane
2:10:09 this is like how did they come up with
2:10:11 that I don't know that's some Wicked
2:10:13 smart people doing that [ __ ]
2:10:15 but what we get to experience is you
2:10:18 type in a thing and Magic comes out
2:10:21 right
2:10:22 Carl Sagan any sufficiently advanced
2:10:26 technology is indistinguishable from
2:10:28 Magic
2:10:31 that's what Sam Altman talks about the
2:10:33 the CEO of open AI why does he do this
2:10:36 he goes
2:10:38 I'm experiencing putting magic in
2:10:40 people's hands and seeing what they do
2:10:42 with it
2:10:43 that's [ __ ] cool Isn't that cool
2:10:45 anyway lost some people there when you
2:10:48 go geeky you're gonna lose some but you
2:10:50 know
2:10:51 um I'm 59 I just thought I'd just
2:10:53 downloaded Discord the other day made an
2:10:55 appointment with my 17 year old to show
2:10:57 me how you go Chuck
2:10:59 it's the way to do it if it works you
2:11:02 work it yep exactly GTO Chevelle Cutlass
2:11:05 same car yeah exactly according to Mid
2:11:09 Journey they're the same car all right
2:11:11 I'm really behind on my
2:11:13 um on my questions here so I'm gonna I'm
2:11:15 gonna skip some questions and sort of
2:11:16 try to get a little more caught up to to
2:11:19 timing
2:11:21 um
2:11:29 Emilio's wife I had chat GPT write a
2:11:31 letter of reference for me and then I
2:11:32 spent a couple of minutes doing small
2:11:34 edits yep chat EPT will get you to 80
2:11:36 like that and then you just polish it
2:11:39 that's how I use it all the time I
2:11:42 worked with chat CBD tonight for a
2:11:44 scallop dinner and some convo and the
2:11:46 end result was
2:11:48 nice like that
2:11:51 um
2:11:53 I'm gonna scroll to the bottom and and
2:11:54 just pop back up because I'm so far
2:11:56 behind on questions so I probably just
2:11:57 skipped a bunch of people's questions so
2:11:59 apologies for that
2:12:01 but
2:12:04 I get the vibe you're already worth
2:12:06 millions I was I took care of that with
2:12:09 poor financial habits and some bad real
2:12:12 estate Investments and investing in too
2:12:14 many startups like my current company
2:12:17 storyline
2:12:18 which is 11 years old and going well so
2:12:21 it's a good investment but
2:12:25 um potentially Millions on paper but not
2:12:27 millions in the bank account
2:12:29 [Laughter]
2:12:31 so if you have spare millions and you
2:12:34 want to invest let me know
2:12:36 do you do mentorship sure
2:12:38 um my contact info is in my bio so my
2:12:42 email's in there if you want to contact
2:12:44 me
2:12:45 how do you use it in your area so I'm
2:12:47 using story so my current company is
2:12:50 called storyvine it's a an automated
2:12:52 video storytelling platform actually
2:12:54 surprisingly not using AI to do the
2:12:56 automation
2:12:58 um
2:12:59 and so so we have an app that guides
2:13:02 someone through telling a specific story
2:13:03 you answer a series of questions in our
2:13:05 app
2:13:06 it goes up to the cloud five minutes
2:13:08 later you got a fully edited video so
2:13:09 pretty slick
2:13:10 we're using open ai's whisper to do
2:13:13 interactive transcriptions now so so now
2:13:16 for every video in the system you can
2:13:18 have a transcription
2:13:20 um that that sort of you know Scrolls by
2:13:22 as the person's talking you read what
2:13:23 they're what they're saying we we are
2:13:26 then using that transcription to create
2:13:28 an omni-channel content generation
2:13:30 engine so now for every video in the
2:13:33 system you can have tweets and a blog
2:13:35 post and all sorts of stuff and so
2:13:38 building a whole system that expands and
2:13:40 extends what storyline is all about is
2:13:43 uh is what I'm currently working on and
2:13:45 then we also use it for writing like we
2:13:47 write prompts for each of our templates
2:13:50 um and so we created a prompt writing
2:13:52 engine that's really slick
2:13:54 um
2:13:55 I'm trying to think what else we use it
2:13:56 to recap meeting notes
2:13:59 um
2:14:00 I don't know you know fair amount
2:14:02 fair amount within the business that's
2:14:04 cool cool representation I'm an HR
2:14:07 supervisor asked me specific questions
2:14:09 chat gbt saved me hours in response time
2:14:12 yeah if you that's one of the things
2:14:14 that that where I think the the large
2:14:17 language models are going to have the
2:14:18 greatest impact is anywhere you where
2:14:20 you've got complicated [ __ ] you need to
2:14:22 do and it's it's that complicated stuff
2:14:24 that's just like the bane of your
2:14:26 existence like you need to deal with
2:14:28 your health insurance company and like
2:14:30 they they want you to do a form and it
2:14:32 depends all this complicated [ __ ] that's
2:14:35 the stuff that large language models are
2:14:37 going to automate out and and so yeah
2:14:39 that's that's a brilliant use of it
2:14:42 thank you for sharing you're limited
2:14:44 Twilight time with us thank you thank
2:14:47 you I appreciate that and discussing Ai
2:14:49 and chat GPT oh Chris Walker yes smart
2:14:52 ass I love it that is solid trolling
2:14:55 right there
2:14:56 see if you're going to come to the AI
2:14:59 learning lab and you want to troll you
2:15:01 got to be creative you got to you got to
2:15:02 be clever with your trolling you can't
2:15:04 just do basic trolling that's boring who
2:15:06 gives a [ __ ]
2:15:08 could AI do auto-generated scheduling
2:15:11 for employees based on statistically
2:15:15 statistically business volume statically
2:15:17 business volume but the answer is yes it
2:15:20 can
2:15:24 um how do you use stable diffusion
2:15:27 um I so I haven't done it for a while
2:15:30 how how I set mine up was I was using
2:15:33 automatic 1111 running in a Google
2:15:36 collab notebook
2:15:38 so if you go to GitHub
2:15:42 you find the automatic 1111 project in
2:15:44 GitHub and then there's instructions
2:15:46 there for how you run it in Google
2:15:48 collab or in I don't know some other
2:15:50 environment or you can run it locally on
2:15:52 your machine there's the problem is
2:15:54 there's so many options I'm sure by now
2:15:56 like I haven't used it in like two or
2:15:58 three months I'm sure by now there are
2:16:00 machine specific installers that it'll
2:16:03 install an application and run it
2:16:05 locally or you can use leonardo.ai let
2:16:09 me I'll go show you that really quick
2:16:10 because if you don't feel like doing all
2:16:12 the geeky [ __ ] the advantage of doing
2:16:14 the geeky [ __ ] is that you
2:16:17 there's lots and lots of options that
2:16:19 are really powerful
2:16:21 the disadvantage is you got to be
2:16:23 [ __ ] geeky
2:16:25 but leonardo.ai is a website I think you
2:16:29 have to get on the waiting list this top
2:16:32 button here says get Early Access
2:16:35 and then once you're white listed you
2:16:37 get access to this thing
2:16:39 and it's it's just slick like across the
2:16:42 top are different styles so you can
2:16:45 generate images in these sort of
2:16:48 pre-built Styles like you know paper art
2:16:50 style
2:16:51 and then down below these are images
2:16:53 that people have created and you can
2:16:55 click on any image and it shows you what
2:16:59 the prompt was what other people made
2:17:01 with this model and you can you can use
2:17:03 that model
2:17:05 um you can fine tune using your own
2:17:10 content like so I did some with my
2:17:14 ugly fat face I don't know why these
2:17:16 like the the worst uh images it came up
2:17:19 with but whatever but I also did like
2:17:21 I'm a photographer so these are some of
2:17:23 my landscape things with like dramatic
2:17:25 clouds from here in Colorado and so I
2:17:29 turned that into a model and I can now
2:17:31 generate content with that model so
2:17:35 um I can say
2:17:36 um
2:17:37 uh let's say dramatic
2:17:41 um
2:17:41 sunset
2:17:44 Sunset over Manhattan
2:17:47 from Brooklyn
2:17:51 and it'll now go off and do its thing so
2:17:55 so like all of this is is our
2:17:58 capabilities that are sitting in stable
2:18:00 diffusion and it's got a lot more than
2:18:02 this but Leonardo puts a really nice web
2:18:05 front end on it
2:18:07 and I'm just using like the nice thing
2:18:09 about Leonardo they give you I think
2:18:10 they give you like 150 credits A day or
2:18:13 something like that like you can make a
2:18:14 decent amount of images a day
2:18:17 um it looks like we have some new people
2:18:18 here my name is Kyle Shannon this is the
2:18:20 AI learning lab we're just waiting for
2:18:21 Leonardo to make some pictures here
2:18:24 um if you're if you're new to the AI
2:18:26 game my background is I'm an
2:18:28 entrepreneur I've been uh I've started I
2:18:31 don't know 12 or 15 companies over the
2:18:32 years and uh I'm all hot and bothered
2:18:36 about this generative AI thing it it has
2:18:38 very strong parallels to the early days
2:18:40 of the World Wide Web
2:18:42 so there's there's some cool shots of
2:18:44 Manhattan
2:18:46 where it took my cloud style and my sort
2:18:49 of photo editing
2:18:51 it over baked it a bit but it's it's in
2:18:53 the neighborhood of the stuff that I
2:18:55 trained it on so pretty cool stuff
2:18:57 um so that's Leonardo
2:18:59 um but yeah if you're here welcome and
2:19:02 actually let me let me flip back over to
2:19:05 this so This AI learning lab uh the
2:19:08 salon.ai that's the AI Salon I I created
2:19:11 is a community of people that sort of
2:19:13 geek out on a AI we meet every other
2:19:15 week and we have a Discord server so
2:19:17 that's info for that the Twitter handle
2:19:20 there's for a newsletter that I created
2:19:22 and then the other thing that is
2:19:25 potentially worth
2:19:26 taking a look at is all these things so
2:19:29 the top one is the official chat GPT
2:19:31 website if you haven't played with chat
2:19:33 EBT just go there right now while you
2:19:35 listen to me ramble in the Twilight of
2:19:37 my life and and that's chat GPD go play
2:19:42 there and then bing.com and pro.com or
2:19:44 are both alternatives to that if you
2:19:46 want to play with them prompts.chat will
2:19:48 teach you about prompting futurepedia
2:19:50 we'll let you find a bunch of tools and
2:19:52 there's sort of an infinite number of
2:19:54 other things that you can do but pop
2:19:57 your questions in the comments below I
2:20:01 will do what I can to answer them and
2:20:03 hit the tap the thing that makes the
2:20:06 hearts tap the double tap the screen
2:20:08 um that'll let people know we're here
2:20:10 and I think that's it is that it I don't
2:20:13 know oh I cuss so if you don't like
2:20:16 cussing I would go
2:20:18 because especially as I get punchier I
2:20:21 get stupider I get stupider and I say
2:20:24 bad words more
2:20:26 I think I actually cussed a lot more at
2:20:28 the beginning of this life I've also
2:20:29 been doing this now for about two hours
2:20:31 so I'm losing my voice my nose is
2:20:33 stuffed up it's raining out
2:20:35 my hair's holding up okay it's well
2:20:38 as far as two pays go
2:20:43 all right
2:20:44 [Laughter]
2:20:47 learning AI will be pointless once
2:20:50 robots come online it's like hiring
2:20:52 monkey over Einstein yeah yeah sorta
2:20:57 that's a little that's a little
2:20:58 dystopian we try to keep it we try to
2:21:00 bend toward the light here at the AI
2:21:02 learning lab there's plenty of people
2:21:04 like we've had 50 years of Hollywood
2:21:05 telling us the robots are going to kill
2:21:07 us so I'm just trying to live in the
2:21:09 how do I deal with it right now can it
2:21:11 help me write the Great American novel
2:21:12 cool I'm in
2:21:15 it's training is a few years old keep
2:21:17 that in mind yes correct and it will
2:21:18 also hallucinate and lie to you openly
2:21:21 and blatantly learning AI will be
2:21:23 pointless we already talked about that
2:21:25 one vector must have a different meaning
2:21:28 here as opposed to physics
2:21:31 well it's a vector on an x y axis
2:21:36 it's a position in Vector space that's
2:21:39 why they call it a a vector database in
2:21:41 a vector Library so it's it's it's a
2:21:44 mathematical plot in space
2:21:48 um that's absolutely awesome
2:21:50 representation of what chat GPT is and
2:21:52 how it works awesome thank you I
2:21:54 appreciate that I'd be curious about
2:21:56 Vector as well yeah if you want to if
2:21:59 you want to start uh geeking out on
2:22:01 embedding and vectors and things like
2:22:03 that look up a project called Lang chain
2:22:06 where you can embed you can take your
2:22:08 own data and embed it yourself using
2:22:11 open A's ai's embedding tools and then
2:22:14 you store those vectors in a vector
2:22:15 database
2:22:16 and then you can start chat gpting with
2:22:19 your own stuff so you can go learn about
2:22:21 that it's super powerful it's like it's
2:22:23 like one tick more technical than I feel
2:22:26 like dealing with right now there's just
2:22:27 too much [ __ ] going on
2:22:29 um but there there are some projects now
2:22:31 that are going to put it behind an
2:22:32 interface like Leonardo did with stable
2:22:34 diffusion so I'm just waiting for that
2:22:38 um I had to tweak my resume yeah it's
2:22:40 really good for that it's also good for
2:22:42 um someone asked me when I first started
2:22:44 getting into this AI you know like Tick
2:22:47 Tock [ __ ] someone someone wanted me to
2:22:49 come speak and they're like can you give
2:22:50 me your AI bio and I'm like I don't have
2:22:52 an AI bio so so I copied
2:22:56 um my LinkedIn profile and then I copied
2:22:59 three different bios that I had written
2:23:00 over the years and I just pasted them
2:23:02 all into chat EBT
2:23:04 and I said write me a a bio that that
2:23:07 includes some of the other [ __ ] that
2:23:10 I've done
2:23:11 and then adds in this new AI stuff I'm
2:23:13 like I think I'm on my third iteration
2:23:15 it was perfect
2:23:17 you know I had to tweak some details but
2:23:19 it was it was pretty cool all right but
2:23:21 it must it must also analyze the
2:23:24 structure of the sentence that example
2:23:26 has a purpose
2:23:28 uh
2:23:30 it does it's so so it's really good like
2:23:33 as you type your whole prompt even if
2:23:35 your prompt is really long and
2:23:36 complicated it analyzes all that it's
2:23:39 really good at that it analyzes all that
2:23:42 mathematically and then figures out what
2:23:44 is the next most likely word out of all
2:23:47 of the trillions of words it's got in
2:23:49 there
2:23:50 what's the next most likely word to be
2:23:53 the next word so yeah so it analyzes
2:23:55 existing stuff quite well love the
2:23:57 geekiness cool thank you I appreciate
2:24:00 that
2:24:01 Tick Tock just jumped me to the bottom
2:24:03 again so I got lost I'm trying to go
2:24:06 back to where I was but it's it's
2:24:08 pointless Tick Tock just messes me up
2:24:11 they took away my dual camera
2:24:14 his training is a few years old I got
2:24:16 all these already
2:24:19 um
2:24:20 can you detail Vector space for me I
2:24:23 think I just did that but not any more
2:24:26 than that all I know is that the vector
2:24:28 space that I showed there that was
2:24:29 two-dimensional is way over simplified
2:24:32 like this is thousands of dimensions of
2:24:35 vectors Vector spaces here's here's the
2:24:37 here's the analogy so this might help
2:24:41 um and this this is not this is all
2:24:43 metaphorical this isn't this isn't
2:24:45 technical
2:24:46 um
2:24:47 so when you
2:24:51 when you prompt chat EBT with a generic
2:24:54 prompt you say write me a poem it will
2:24:57 write you a poem and it will be
2:24:59 disappointing like it might be
2:25:01 impressive-ish
2:25:02 but it will probably be disappointing
2:25:04 but if you give it a really complicated
2:25:06 or or really specific set of parameters
2:25:10 I want a poem written like Robert Frost
2:25:14 and you know I wanted to have these
2:25:16 metaphors and this thing
2:25:18 it'll write you a much better poem
2:25:21 so so why is that so Andre carpathi
2:25:24 who's one of the co-founders of open AI
2:25:26 was talking talking about this and and
2:25:28 the metaphor that I came up with was
2:25:32 um
2:25:33 if you imagine all of the words
2:25:36 as the Grand Canyon all of the words are
2:25:38 in the Grand Canyon and you're in a
2:25:40 helicopter and if you fly up really high
2:25:43 you can see into all the canyons and sub
2:25:46 canyons and little valleys and right you
2:25:48 can see all the words
2:25:50 so when you just type in something
2:25:52 generic
2:25:53 it basically is pulling from all of the
2:25:56 semantic canyons and so what it gives
2:25:58 you is generic because it's just kind of
2:25:59 the average of just it's like mush right
2:26:03 and what Andre carpathy was talking
2:26:05 about is if you've got
2:26:07 if you ask it to solve a very specific
2:26:10 math problem if you're in the helicopter
2:26:12 really up high
2:26:13 it's going to look down and look for all
2:26:15 of the places where that math problem
2:26:17 has been solved
2:26:19 and
2:26:20 one of the people that solved that math
2:26:22 problem is a math professor who did it
2:26:25 right and like 30 of the people that
2:26:27 solved that math problem are students
2:26:29 and 26 of them got it wrong
2:26:32 and so if you're really high in the
2:26:34 helicopter it's going to just sort of
2:26:36 average out all of the people that have
2:26:38 solved that problem and it'll it'll
2:26:40 confidently give you an answer to that
2:26:41 math problem but if it sort of averages
2:26:44 in those students with that Professor
2:26:46 it's going to give you a wrong answer
2:26:47 but if you say you know I want you to
2:26:50 act as a math professor from Stanford
2:26:52 who specializes in this area of
2:26:54 mathematics it's like you're flying your
2:26:56 helicopter down into that little canyon
2:26:58 where the semantic you know space the
2:27:02 vector space only contains like
2:27:05 professors right so that's why telling
2:27:08 chat GPT to act like a certain thing
2:27:10 it's like lowering the helicopter into a
2:27:13 more specific Vector space that's why it
2:27:15 works
2:27:17 imagine an AI driven nursing home I
2:27:19 think I think AI driven companionship
2:27:23 and
2:27:24 um
2:27:25 um like like health care support and
2:27:28 caregiving is actually going to be a
2:27:30 thing and it's going to be an important
2:27:31 thing
2:27:32 um I think it's going to be a while
2:27:33 until
2:27:35 um we get it right but I think it's
2:27:37 actually going to be an important thing
2:27:38 what's the open source AI image Creator
2:27:41 well stable diffusion is open source
2:27:44 um
2:27:45 mid Journey's not adobe's not Dolly's
2:27:48 not
2:27:50 um
2:27:51 there's one called Blue Willow that's
2:27:54 not open source but it's free I don't I
2:27:57 think it's built on stable diffusion I
2:28:00 don't know how they're doing it for free
2:28:01 I don't like who who the [ __ ] is paying
2:28:04 their compute bill
2:28:06 so but you can go check out Blue Willow
2:28:07 you have to do it in Discord like you do
2:28:09 mid-journey
2:28:10 um I'm sure there are open source image
2:28:14 generation tools out there I don't know
2:28:15 them I just sold my first business plan
2:28:17 and Grant package that I use chat gbt to
2:28:19 create congratulations that's awesome
2:28:21 that's super cool Twilight not quite
2:28:24 this man is non-stop yet Jim Buckley I
2:28:28 just go
2:28:29 Gen X baby [ __ ] it
2:28:33 I do have to go to bed soon I have yeah
2:28:36 I have an eight o'clock meeting tomorrow
2:28:38 morning with a big client
2:28:41 I don't want to get up early and be
2:28:44 responsible for my life that I'm in the
2:28:47 Twilight of do you have a YouTube
2:28:49 channel I do not people keep asking me
2:28:51 that I should probably do it I had a
2:28:53 YouTube channel well it's still there I
2:28:55 have a story of Mine YouTube channel
2:28:56 when we went into lockdown I did a I
2:28:59 used storyvine technology to do a daily
2:29:02 video diary it was called the extroverts
2:29:04 view from within
2:29:06 and I did a daily diary about how much
2:29:09 it sucked being stuck in this stupid
2:29:11 home office every day
2:29:13 and uh
2:29:15 I did it every day for a year and I
2:29:17 think I got 65 followers like I had I
2:29:21 had like eight fans it was like it's
2:29:24 like the biggest [ __ ] waste of time
2:29:25 and then with Tick Tock like this took
2:29:27 off so I don't know
2:29:30 I guess I could have a YouTube channel
2:29:31 but I like I don't I'm not good enough
2:29:33 at the marketing side of this thing like
2:29:35 like Tick Tock gave me distribution and
2:29:37 and audience so
2:29:39 um I don't know at some point if I start
2:29:42 taking this more seriously I'll get a
2:29:44 producer or something like that and
2:29:46 I don't know I
2:29:52 we are at the dawn of something amazing
2:29:54 I agree with that so check this out this
2:29:56 is my um when I did the creative
2:29:58 mornings talk
2:29:59 this was sort of my money slide
2:30:02 was this one right here
2:30:06 the great Renaissance and that's
2:30:08 actually that's a
2:30:10 two different mid-journey images the
2:30:12 woman in the center was one image and I
2:30:14 photoshopped her in
2:30:16 and then the back thing and this was
2:30:17 sort of like just this explosion of
2:30:19 innovation and this is this is what I
2:30:22 think we're entering right now is the
2:30:24 great Renaissance
2:30:27 because it's just it you know these
2:30:30 tools are are Humanity amplifiers
2:30:32 they're they're they re they really
2:30:36 you know allow you to take a vapor of an
2:30:39 idea in your head and realize it in ways
2:30:41 that were just not possible
2:30:44 do you have scheduled times that you go
2:30:46 live nope I have ADHD schedules in me do
2:30:49 not get along
2:30:52 um I've heard once I haven't heard one
2:30:55 cuss anything in five hours
2:30:58 [ __ ] no you haven't because I haven't
2:31:00 [ __ ] sworn
2:31:02 exactly
2:31:05 um three-dimensional Vector space no
2:31:07 thousands of Dimensions I farted good
2:31:11 comment
2:31:12 [Laughter]
2:31:15 okay if you're gonna troll
2:31:19 um be clever or you know I farted works
2:31:22 like just be you know sophomoric humor
2:31:24 goes a long way with me think Tony
2:31:26 Stark's Jarvis spring model image from
2:31:28 the Age of Ultron Marvel movie and that
2:31:31 is Vector space yeah that's correct
2:31:32 that's right
2:31:33 the algorithm adapts
2:31:36 um absolutely I'm curious about your
2:31:38 real estate faux pas
2:31:41 um
2:31:41 I invested in a bunch of stuff with a
2:31:43 dude that invested for me and then the
2:31:45 2008 crash happened and he got out with
2:31:48 all of the properties that were not
2:31:50 underwater and
2:31:54 had told me he had put all of my all of
2:31:59 our investments into an LLC but he
2:32:02 didn't he just put him in my name so he
2:32:04 stuck me with seven properties that were
2:32:06 under water
2:32:07 for Bank of America and then
2:32:10 um I went to refinance them in Bank of
2:32:12 America said no you can't refinance them
2:32:14 you just have to let them foreclose and
2:32:16 then they foreclosed and Bank of America
2:32:18 couldn't find the [ __ ] paperwork they
2:32:21 didn't know who actually owned them
2:32:23 so I got out of it but I lost a [ __ ] ton
2:32:25 of money so
2:32:28 faux pas is just
2:32:30 stupid [ __ ] decisions with a bad
2:32:33 um partner so what are you gonna do
2:32:37 it's just money I have a degree in
2:32:39 acting so the fact that I had money at
2:32:40 one point in my life was a surprise
2:32:42 anyway I'll get it again we'll get there
2:32:44 we'll get there and now we got AI you
2:32:47 can make mine I'm gonna catch every time
2:32:49 [Laughter]
2:32:53 wow good Photoshop thank you very much I
2:32:55 appreciate that I partner with what was
2:32:58 that I partnered with the VA
2:33:00 to house homeless veterans how how can
2:33:03 AI assist me
2:33:05 um
2:33:06 I don't know like
2:33:09 I I mean kind of almost in any way I you
2:33:12 know what I would I would say to you
2:33:14 Joaquin is
2:33:17 don't don't think of it so much like
2:33:19 what can chat gbt do for me I would
2:33:21 think of it more like
2:33:23 lead with your problem like what's the
2:33:25 problem you're trying to solve what's
2:33:27 the big nut you're trying to crack is it
2:33:31 um you need more resources or you need
2:33:34 to you know you need to come up with
2:33:36 ideas for fundraisers or you need to
2:33:38 organize schedules or like I would say
2:33:41 pick pick a problem like it doesn't
2:33:44 matter which problem you pick pick a
2:33:46 problem and then interact with chat GPT
2:33:49 like cheche EBT is the expert at solving
2:33:52 that problem and just say hey cheche EBT
2:33:55 be an expert at this thing that I need
2:33:59 solved here's my specific problem here's
2:34:02 what I did here was the result
2:34:05 here's why I'm frustrated do you have
2:34:07 any suggestions out it'll come and then
2:34:11 you'll either like the suggestions or
2:34:12 not no no
2:34:14 um I tried you know most of the things
2:34:17 you suggested and that's the wrong
2:34:19 direction you know I need to go this
2:34:21 other direction out will come a bunch of
2:34:23 new things so
2:34:25 I would start with that
2:34:28 um and and then depending on
2:34:31 um you know housing homeless veterans
2:34:33 like
2:34:35 you could you could also
2:34:38 um use these AI tools to upskill some of
2:34:41 those veterans right get them using
2:34:43 these tools get them expressing
2:34:45 themselves get them processing their
2:34:48 experiences I I like I don't quite know
2:34:50 enough about what you do but but I would
2:34:52 say pick specific things you're trying
2:34:54 to solve and then and then say how can
2:34:57 these AI tools help me solve that
2:34:59 specific problem if you do that twice
2:35:01 with like two specific problems it'll
2:35:04 it'll become much more clear how you can
2:35:06 deploy these tools against a lot of your
2:35:10 problems
2:35:13 um what was the interactive chat GPT
2:35:15 slide it was very cool the interactive
2:35:17 chat GPT slide which one the when I was
2:35:21 showing how chat GPT works
2:35:23 the
2:35:25 the where is it
2:35:28 this one that shows how oh yeah how chat
2:35:32 GPT works this one wait oh I'm not
2:35:34 showing my camera I used to have dual
2:35:36 cameras but tikitaki took them away from
2:35:38 me
2:35:39 this one was from a presentation that I
2:35:42 did for Creative mornings um and it's
2:35:44 it's actually part of a keynote now that
2:35:46 I've put together where it was a fun
2:35:48 presentation so I started out talking
2:35:50 about
2:35:51 um
2:35:53 my my first experience with personal
2:35:56 computers in the fall of 1977 it was
2:35:59 with the trash 80. and the more I
2:36:02 learned about how remarkable these
2:36:04 machines were the more I realized I
2:36:06 wasn't old enough to be like these guys
2:36:10 Steve Jobs and Wozniak or these guys the
2:36:13 Microsoft co-founders
2:36:14 and so I talk about just missing
2:36:18 um the early days of the PC Revolution
2:36:22 and then in the mid 90s I discovered the
2:36:25 web
2:36:26 and I launched this magazine and three
2:36:28 weeks later it's written about in Paris
2:36:29 and I have an epiphany about the World
2:36:31 Wide Web holy [ __ ] this is one of those
2:36:33 and then I kind of jumped forward to
2:36:35 2022 and chat EBT you know is the thing
2:36:41 that changes everything so that's that's
2:36:43 essentially what the Talk's about and I
2:36:44 kind of go into
2:36:46 how I got there what I did how I learned
2:36:48 about the image [ __ ] and you know all
2:36:50 that sort of stuff so
2:36:52 thank you for the kind words
2:36:56 I gotta wrap this up soon what time is
2:36:59 it 11 49 or 10 49. oh my God yikes I
2:37:03 don't know what I said or did or showed
2:37:05 but cool oh let me flip this back around
2:37:07 because you don't want to miss this
2:37:12 I've been teaching a little University
2:37:13 in West Virginia and they are really
2:37:16 struggling to adapt Embrace AI yeah you
2:37:19 gotta
2:37:22 yeah it's it's rough it's really easy to
2:37:25 demonize and it's really easy to be
2:37:27 scared of and and particularly in West
2:37:30 Virginia where I'm sure you've got a lot
2:37:32 of trauma around job loss you know
2:37:34 because of minds and things like that
2:37:38 um
2:37:39 AI is another thing that is going to
2:37:41 threaten livelihoods and it's gonna it's
2:37:44 gonna impact a lot of livelihoods the
2:37:45 good news is it's going to create a lot
2:37:47 of new jobs but there's going to be a
2:37:48 really painful
2:37:50 Chunk in the middle
2:37:52 so
2:37:53 part of the purpose of this channel is
2:37:56 to do what I can to demystify like I
2:37:59 I think because of Hollywood telling us
2:38:01 that the robots are going to kill us and
2:38:03 the fact that this AI stuff is
2:38:07 pretty sciencey and and sciency stuff
2:38:10 has political
2:38:12 ramifications these days
2:38:15 um
2:38:16 I can see where there's resistance to it
2:38:18 and and the only thing I can suggest is
2:38:20 do what you can to humanize it and and
2:38:22 demystify it so
2:38:26 maybe rather than sort of you know doing
2:38:29 things that are sort of big and
2:38:31 impressive like like lower the bar down
2:38:33 to like what's the most human kind of
2:38:35 thing you can do with it like create a
2:38:37 meal plan or create
2:38:40 um
2:38:41 I don't know help someone
2:38:44 write a children's book and illustrate
2:38:46 it for their kid like make it as human
2:38:48 as possible and show them that it can
2:38:50 help do stuff like that that might be a
2:38:53 way to to get through that but I I would
2:38:55 imagine there's going to be some
2:38:57 some geographies in the U.S anyway that
2:39:01 this is going to be a harder Rock to
2:39:03 push up the hill
2:39:05 that's the Shady Pines of AI I'm kind of
2:39:08 loving it cool
2:39:09 how would you leverage Chachi BT to
2:39:11 Market to aspiring actors as clients
2:39:16 um
2:39:18 yeah I would speak to them like it's
2:39:20 funny I just talked to a buddy of mine
2:39:21 who's got an acting studio in New York
2:39:23 City and he was asking me about the same
2:39:25 thing
2:39:27 um
2:39:30 I don't know with actors I would focus
2:39:32 on the content I would focus on
2:39:36 I mean actors don't have a ton of money
2:39:38 so so you wanna you wanna connect them
2:39:42 with jobs somehow so anything that you
2:39:44 can do to
2:39:46 you know do either storytelling or or
2:39:51 um content marketing that appeals to
2:39:54 their sense of wonder and
2:39:58 um possibility
2:40:00 like you want to put actors in touch
2:40:02 with possibility because it can be a
2:40:04 very Bleak and hopeless profession
2:40:09 yeah no
2:40:12 look at your programs and look at where
2:40:15 you need help yep agree with that what
2:40:17 could AI do for a real estate investor
2:40:21 um
2:40:22 can do a lot it can help you know you
2:40:24 could you could use it to to do
2:40:28 um
2:40:29 cool graphics and photographs you could
2:40:32 have it right
2:40:34 um social media content for you you
2:40:36 could have it help
2:40:38 you could probably come up with a pretty
2:40:41 straightforward
2:40:43 um
2:40:44 automation using zapier where you create
2:40:46 a Google form
2:40:48 where you where you type in attributes
2:40:50 of a property that that's just you do
2:40:52 it's the same every time like square
2:40:54 footage number of bedrooms interesting
2:40:56 things in the neighborhood like what
2:40:58 whatever like you know what that list is
2:40:59 right so you kind of just write that
2:41:01 into a form and then you could have it
2:41:04 automatically spit out here's your you
2:41:06 know sell sheet here's your brochure
2:41:08 copy here's your Instagram posts right
2:41:12 you can have it generate just a whole
2:41:14 bunch of [ __ ] for every single property
2:41:15 so so rather than kind of ad hoc doing
2:41:19 that for each thing you could create a
2:41:21 tool that automatically generates
2:41:23 omnichannel content
2:41:25 um
2:41:26 you could use it to introduce Brokers
2:41:28 you could have it right personalized
2:41:30 letters to customers or prospects
2:41:34 um
2:41:34 you could have it analyze data for the
2:41:36 business
2:41:37 um check out if you haven't seen it go
2:41:39 look at um YouTube videos look up code
2:41:42 interpreter so this is a new new Tool
2:41:45 that's coming for open AI where you can
2:41:48 upload spreadsheets of data to this
2:41:50 thing and it will do full-on data
2:41:52 analysis for you so it could be a data
2:41:54 analysis assistant so probably Lots like
2:41:57 there's probably not a lot it couldn't
2:41:59 do
2:42:01 for Real Estate because it's such a data
2:42:03 Rich environment
2:42:05 ask not what AI can do for you but you
2:42:08 can do it for AI yeah
2:42:10 I mean AI is your co-pilot man you're
2:42:13 the pilot you're still in charge but
2:42:14 it's your it's your little buddy it's
2:42:16 your little buddy right there that's
2:42:18 [ __ ] smarter than you are it's like
2:42:20 we don't want to admit it but he is
2:42:23 she is Sydney I'm in code boot camp how
2:42:26 will AI affect software development well
2:42:29 like [ __ ] insane you're in code have
2:42:32 okay if you haven't gone to chat EPT and
2:42:35 haven't had it code so I assume you've
2:42:38 done this but if you haven't this is
2:42:40 just chatgpt 3.5 so this is the basic
2:42:44 thing
2:42:45 um write me
2:42:48 a python application that
2:42:51 [Music]
2:42:52 um
2:42:53 oh I don't know
2:42:55 um
2:42:56 brings the pi Theory and
2:43:01 theorem
2:43:04 it's all spelled wrong to life and
2:43:08 represents it visually
2:43:13 with
2:43:15 animated Sprites
2:43:18 of fruits and vegetables I don't even
2:43:23 know if that's possible
2:43:25 but apparently it is
2:43:27 there's your code
2:43:30 um so if you're learning coding right
2:43:34 now you should be using
2:43:37 um github's copilot which is
2:43:40 optimized for coding but like
2:43:43 this is GPT 3.5 gpt4 is better and and
2:43:47 GitHub co-pilots even better apparently
2:43:51 um
2:43:52 start thinking of GPT as a pair
2:43:55 programmer where you give it the ideas
2:43:57 and you curate the code but it does the
2:44:01 bulk of the writing I mean it's insane
2:44:03 so if if if you're in coding class and
2:44:07 they're not teaching this right now get
2:44:10 on it
2:44:12 because
2:44:13 seriously knowing code great knowing
2:44:16 code and how to do it with AI going to
2:44:18 be a requirement very very shortly
2:44:22 all right I got four more minutes so
2:44:25 I'll do two more questions yesterday I
2:44:27 used chat GPT secretly to give my friend
2:44:29 advice when she was sad it was great
2:44:31 it's you know what's funny about so so
2:44:33 they did a they did a test recently
2:44:36 where patients interacted with chat GPT
2:44:39 doctors and real doctors
2:44:41 and the patient's overwhelmingly
2:44:44 preferred the chat GPT doctors because
2:44:46 they were more competent and they were
2:44:48 more empathetic
2:44:49 and Chachi PT is actually really good at
2:44:51 nuance and it's really good at [ __ ] like
2:44:53 like I've done a bunch of these things
2:44:55 where not a bunch but
2:44:57 where I've had difficult conversations
2:44:59 and I'm like hey I have this difficult
2:45:00 conversation how can you help me frame
2:45:03 it
2:45:03 and it came it was brilliant it was
2:45:06 really good it's really good at that
2:45:08 all right well my first computer was a
2:45:10 ti-994a on a Commodore I the uh the the
2:45:15 ti-99 that was one I wanted I wanted the
2:45:17 Timex Sinclair that was that was my
2:45:19 little lust computers like this little
2:45:21 membrane keyboard computer like this big
2:45:24 it was like magic I never got one I
2:45:26 should get one of those bad boys off
2:45:28 eBay
2:45:30 I was at wozniak's home in Los Gatos
2:45:32 that's super cool was the was can you
2:45:36 talk a little bit about becoming a
2:45:38 prompt engineer um
2:45:40 I can't I think it's it's going to be a
2:45:42 short-lived profession because I think
2:45:45 these tools are going to get really good
2:45:46 at prompting themselves but I I would
2:45:48 just just go to prompts.chat there's a
2:45:50 primer at the beginning of this that
2:45:52 will that will give you it there's also
2:45:54 lots of videos on it
2:45:56 um I would be weird don't don't feel
2:45:59 like you it's like I don't think it's a
2:46:02 good practice right now to
2:46:05 just go get other people's prompts
2:46:08 prompt libraries and use them I think
2:46:10 it's actually better right now to to
2:46:12 just start experimenting with it start
2:46:14 treating chat gbt like a collaborative
2:46:17 partner where you're having a
2:46:18 conversation back and forth that'll
2:46:20 teach you more about prompt engineering
2:46:22 than trying to learn all the all the
2:46:23 rules because
2:46:26 it's changing really fast and it's kind
2:46:28 of like learning math before you use a
2:46:30 calculator like there's there's
2:46:32 something conceptual going on with using
2:46:36 language as a programming tool
2:46:39 um that I like I I don't know for me
2:46:42 it's just like just bang up against the
2:46:45 frustration of getting it wrong a bunch
2:46:46 and just keep practicing is going to be
2:46:48 better like you can go go to
2:46:50 prompts.chat to learn the basics but
2:46:52 then just start experimenting I think
2:46:54 it's the best way pick a problem to
2:46:56 solve start solving it and that will
2:46:58 make you a better prompt engineer
2:47:00 sounds like Boston has a handle on
2:47:02 ushering AI into general public yeah and
2:47:05 it's like why is New York not done that
2:47:07 why is San Francisco not done that why
2:47:09 Chicago like why are major cities not
2:47:12 doing this like Boston good for you but
2:47:15 everyone else where the [ __ ] are you
2:47:16 this is coming quickly like you know
2:47:19 everyone talks about oh we need
2:47:20 politicians to regulate this and this
2:47:22 and that politicians don't understand
2:47:23 what [ __ ] Wi-Fi is
2:47:25 so so we got to start getting like like
2:47:28 you know people in government doing what
2:47:30 Boston did get your head around this
2:47:32 [ __ ] start figuring out give the people
2:47:34 guidance so you know there's not so many
2:47:37 inadvertent fuck-ups
2:47:40 oh this is a good one to end on I just
2:47:42 prompted chat EPT to tell me a bedtime
2:47:44 story as if it was written by
2:47:45 Shakespeare it was beautiful that's
2:47:47 awesome that's really good okay
2:47:50 let's stay online for another hour at
2:47:52 least Firefly I gotta go it's 11 o'clock
2:47:55 I don't know where my children are well
2:47:58 they're actually upstairs but I do know
2:48:00 where my children are I gotta go thank
2:48:02 you all for your time I really
2:48:03 appreciate you hanging out with me and I
2:48:06 hope this is well thank you for the
2:48:07 little hearty hearts and the follows
2:48:09 yeah feel free to follow me I I put
2:48:11 content on Tick Tock I do these lives I
2:48:13 try to do them every night or so but bye
2:48:15 everybody bye Emilio's wife thanks for
2:48:17 thanks for hanging out peace