AI Learning Lab

aiLL 6/10/23 2 - Unlocking AI's Creative Potential

bG-xTu-Is_U
Video2023-09-033:06:485 views

Description

I explore how to tap into AI's remarkable creative capabilities. We look at generating images, music, stories, code, and more with tools like ChatGPT, Midjourney, and Stable Diffusion. Key themes include democratizing access, responsible development, and maintaining our humanity. The pace of change is swift, but together we can craft an inspiring future. AI Learning Lab address: https://tiktok.com/@aiLearningLab #ai #creativity #future #chatgpt 0:27 - The transformative power of AI creativity 0:50 - Leveraging ChatGPT for creative problem solving 1:45 - Democratizing access to AI through Microsoft and OpenAI 2:15 - Surfing the AI tsunami versus waiting on the beach 3:14 - Living and creating at a higher level with AI assistance 5:14 - Rewriting LinkedIn profiles with ChatGPT 15:21 - AI as a humanity amplifier versus robot overlord 18:17 - AI-generated music, images, stories, and code 23:02 - Narrowing AI into a semantic canyon 32:22 - AI ending soul-crushing bureaucracy 35:09 - Entering the age of the liberal arts major 43:23 - Democratizing AI through open source models 52:17 - The future all-in-one AI personal assistant 1:34:43 - AI progressing too swiftly to control 1:56:46 - Choosing the AI tool that scares you the most 2:25:00 - Preparing students for an AI-enabled world

Chapters

Transcript

0:03 um
0:22 hello hello hello it's Kyle Shannon with
0:25 the AI learning lab welcome
0:27 just getting started here on a Saturday
0:30 night I got an iced coffee because I was
0:32 a little sleepy and uh there was coffee
0:35 in the pot so
0:37 a refreshing summer beverage
0:39 but I drink it in the winter too
0:43 honest AI learning lab if you have
0:45 questions about AI Luby oh newbie to AI
0:48 love your content thank you
0:51 um peaceful peaceful Peach
0:53 yeah welcome if you're new to AI you've
0:56 probably heard this if you've been on
0:57 the channel because I I do this a lot go
0:59 there go to chat.openai.com that is the
1:02 official chatgpt website and just start
1:05 playing while I ramble on
1:08 um you know follow examples that I show
1:10 but also just you know pop in things
1:11 that you're curious about interested in
1:13 see if it can do stuff like challenge it
1:16 ask it to do things you think it's not
1:18 possible for it to do and then when it
1:21 does it just just pick your jaw back up
1:23 off the floor and uh and keep going
1:27 um so yeah so those things those URLs
1:29 right there if you're new here are I put
1:32 these up here just as kind of a public
1:33 service announcement
1:35 all three of those are places where you
1:37 can play with chat gbt
1:39 oh your favorite Channel thank you so
1:42 much
1:43 that's silly I'm just like an old dude
1:45 sitting here chatting uh chat about AI
1:48 but I appreciate the kind words I hope I
1:50 hope this makes a difference
1:53 um
1:53 about those top one is chat EPT second
1:56 one is Bing that's Microsoft's search
1:59 engine but it also has a chat button and
2:00 if you go to that chat button that's
2:02 gpt4 for free free well they stick some
2:06 ads in there and then there's Poe and
2:08 Poe gives you access to Poe's version of
2:11 chat GPT anthropics Claude models and
2:15 chat gbt three and four you have to pay
2:18 for some of those things but you get to
2:20 compare the different models and it's
2:22 pretty good for newbies also the speed
2:24 of chat gbt does things it is alarmingly
2:27 fast it is alarmingly fast
2:30 um
2:31 did that
2:33 here's what here's here's here's what
2:36 shifts for me with chat gbt you know the
2:39 like let's say you've got a thing you
2:41 gotta write whatever the thing is you
2:42 got to write a letter you gotta write a
2:44 bio you gotta write a social media post
2:46 you know like the hour or two that you
2:51 normally take to do the first draft well
2:54 that happens in 30 seconds
2:58 so you get to like 80 percent of
3:00 completion in like 30 seconds and then
3:03 you have that two hours rather than
3:05 taking you know two hours or an hour or
3:07 whatever it is half an hour to get to
3:09 the first draft you're kind of instantly
3:10 at the first draft and what it's done
3:13 for me is it loud it allows me to live
3:15 at a higher creative level
3:18 um just more consistently like I'm I'm
3:20 not getting stuck down in the weeds of
3:22 that that that kind of lower level
3:24 execution just like blocking out the
3:26 words blocking out the outline whatever
3:27 it is made a three-course meal with GPT
3:30 tonight nice guacamole biryan tacos and
3:34 strawberry nice that's awesome I'm gonna
3:38 use it to revamp my LinkedIn page that's
3:39 actually a really good idea
3:41 oh that's really interesting let's go do
3:44 that let's go do something live with
3:45 that I think that's a really cool idea
3:47 let's go
3:49 let's let's flip on over
3:54 yeah so okay hang on so I'm gonna go to
3:57 my LinkedIn page
4:02 here we come to save the day
4:06 um I'm gonna go to
4:08 have Shannon
4:10 hi Kyle what are you doing Kyle
4:13 oh you know I'm just on Tick Tock are
4:16 you an influencer Kyle I don't have a
4:18 pink stripe in my hair do I well your
4:21 hair does look kind of funny all right
4:23 so I'm going to copy all of
4:30 all of my LinkedIn profile
4:36 do I do it no
4:38 why not because you're old and you don't
4:40 know how to use a computer you old man
4:44 God damn it
4:46 apparently I don't know how to use a
4:48 computer am I hitting the wrong shift
4:50 key or not a shift key
4:52 there we go okay so I'm going to copy my
4:55 LinkedIn page
4:56 and then we're going to pop over here to
4:58 chatgpt you can make money which hatch
5:01 every day
5:03 by the way people that say that on Tick
5:05 Tock or YouTube do not listen to them
5:07 they are full of [ __ ] okay they're
5:09 trying to sell you something or they're
5:11 annoying and why do they talk like that
5:12 the hell is that
5:15 um okay so I'm gonna go to gpt4
5:19 do I want to do browse with Bing I don't
5:22 think so I'm just going to do regular
5:23 gpt4 why I don't know sometimes you just
5:26 gotta play
5:28 um and then what I'm going to do is I'm
5:29 going to say based on this LinkedIn
5:33 profile
5:37 um and then I'm oh damn it it's gonna
5:39 tell me I'm sorry for the confusion but
5:40 it appears that there's no LinkedIn
5:42 profile here I know it's at gbt thank
5:45 you for correcting me based on this
5:49 LinkedIn profile oh by the way a bunch
5:52 of new people just showed up hi my name
5:54 is Kyle Shannon this is the AI learning
5:56 lab
5:57 um if you've got questions about AI pop
5:58 them down below I try to answer them all
6:00 uh I will get annoyingly behind in the
6:03 questions and it will seem like I'm not
6:04 paying attention to you but I love you I
6:07 love you all you're the best uh if you
6:09 would go ahead tap the screen make it
6:11 seem like you're excited to be here and
6:13 that'll get other people here share this
6:15 live I don't know how you do that but
6:17 apparently that'll get more people here
6:19 and then if you have questions pop them
6:21 in um someone said they were going to
6:22 use chat gbt to to rewrite thanks for
6:25 all the hearts that's cool uh they were
6:27 going to use chat EBT to rewrite their
6:29 LinkedIn profile so I'm going to try it
6:31 now live live on the internet well on
6:34 the tick tock on the internet is this
6:36 the tick tock or is this the internet I
6:38 don't know all right based on this
6:41 LinkedIn profile
6:44 and then I'm going to hit shift return
6:45 to make some line breaks I'm going to
6:47 paste in my LinkedIn profile
6:50 um let's see can you
6:53 recommend
6:57 um
6:59 things I can do to improve it
7:07 as well as some ideas for
7:13 what
7:15 um
7:18 I don't know let's just let's just let's
7:21 just keep it simple
7:22 can you recommend things I can do to
7:24 improve it so based on those LinkedIn
7:26 profile paste
7:28 give me some ideas let's see what it
7:30 comes up with so there just read my
7:31 LinkedIn profile like that
7:33 Kyle Shannon's LinkedIn profile seems
7:36 comprehensive and detailed showcasing
7:39 this wealth of experience and skills
7:41 apparently chat GPT is an ego boost as
7:44 well
7:45 Kyle is an awesome human being no it
7:47 doesn't say that
7:49 uh what does it say still here are a few
7:51 recommendations improve the profile
7:53 summary your profile summary is a great
7:56 place to Showcase your personality
7:57 passion and unique value proposition
8:00 hey how's it going what's your value
8:03 proposition
8:05 um well the current summary does a
8:06 decent job of this consider making it
8:08 more concise powerful focusing on your
8:11 key achievements
8:13 um goals and skills okay before I read
8:15 the other ones I'm going to say
8:17 um awesome oh by the way saying things
8:20 like awesome and please and thank you to
8:22 chatgpt just starts to happen after you
8:25 start talking with it because it's like
8:26 it's like working with chat gbt once you
8:29 get into it is kind of like working with
8:32 a remote co-worker that's [ __ ]
8:33 smarter than anyone you've ever worked
8:35 with that's kind of kind of what the
8:36 experience is awesome GPT
8:39 um
8:40 can you wait Can
8:43 you please
8:47 um show me my current
8:51 profile
8:53 and
8:55 then write me a new one based on your
9:00 Reco
9:03 mendations question mark
9:07 and it should do that sure I can provide
9:09 a revised version of your summary so so
9:11 this is one of the things chat EBT is
9:14 really good at is like analyzing other
9:16 things you could put another thing in
9:17 there and do that or you can say here's
9:19 a blog post I wrote now write me a new
9:21 post on the I don't know Mesopotamian
9:24 War following that blog you know profile
9:27 all right
9:30 current profile
9:32 my background and skills are eclectic
9:34 but a Common Thread that has emerged is
9:36 storytelling in one form or another
9:38 that's true I wrote that okay there's my
9:41 experience proposed profile as an
9:45 accomplished entrepreneur I merge
9:47 storytelling technology and creativity
9:50 to invent and Lead Innovative projects
9:53 that's pretty good from
9:54 cofoundingagency.com one of the largest
9:56 interactive marketing firms to launching
9:58 storyvine a platform and enabling
10:00 large-scale high quality video
10:02 storytelling my career is characterized
10:05 by pioneering initiatives again it's an
10:07 ego boost but that's pretty good like
10:09 it's well written my passion lies in
10:11 leveraging my writing public speaking
10:13 and creative strategy skills to ignite
10:15 power wait to ignite impactful
10:18 narratives across diverse sectors now as
10:20 the chief generative officer at content
10:22 Evolution I'm eager to unlock the
10:24 potentials yeah that's pretty solid
10:26 that's pretty flipping solid so all
10:30 right there you have it people gpt4
10:33 gpt4 oh someone said weak I don't know
10:36 what they're referring to but that's all
10:38 right
10:38 um pop your questions down below it
10:41 looks like a lot of you put have put
10:42 questions in here if you're new here
10:46 my name is Kyle Shannon this is the AI
10:49 learning lab I'm a long time
10:50 entrepreneur
10:51 um as just evidence by popping my
10:53 LinkedIn profile in there sorry I don't
10:56 normally do that but I thought that was
10:57 an interesting exercise
10:59 um these URLs the top one is the
11:01 official chat GPT website I'm not
11:03 connected with these in any way this is
11:05 just
11:06 um me trying to make it less confusing
11:09 um the top one is the official chatgpt
11:11 website the next one is Bing that's
11:13 Microsoft's search engine but also has a
11:16 chat button if you click that chat
11:17 button that's chat gpt4
11:20 and then pro.com lets you play with a
11:22 bunch of different
11:23 um chat chat large language models and
11:25 compare them um so so those three are
11:27 good and then prompts.chat teaches you
11:29 how to prompts and future PDF is just a
11:31 list of tools if you want to geek out
11:34 excuse me I just had a big uh I went out
11:37 with
11:38 one of my employees is in town and
11:41 invited me out to dinner so I was just
11:43 out and had a large Italian dinner and
11:45 I'm I'm really um full
11:48 so so I'm having coffee you gotta wake
11:51 up I'm having like a post lunch slump
11:54 but it was dinner
11:58 okay
12:01 all right does prompts.chat have a
12:03 prompt for LinkedIn I don't know but
12:07 just go load it and then just do a find
12:09 on the page and see if there's a
12:10 LinkedIn one there but you can just like
12:12 you can see what I just did there you
12:13 can just copy and paste your LinkedIn
12:15 profile you can say
12:17 um you know
12:19 act as a marketing consultant or act as
12:23 a consultant that specializes in
12:27 um
12:27 LinkedIn marketing and positioning or
12:30 something like that just tell you you
12:32 don't need to do super you don't need to
12:34 be super specific with that stuff if you
12:36 just kind of put it in the neighborhood
12:37 it'll be pretty good but because it's
12:39 reading your existing LinkedIn profile
12:41 it's going to be able to pull a lot of
12:42 stuff out of that the trick with these
12:44 GPT models is to provide them with a lot
12:47 of context like more than you think
12:48 Google for years has trained us just put
12:51 like the minimum amount
12:53 that that makes sense into the chat box
12:55 otherwise it'll screw things up and
12:57 you'll get way too many things check EBT
12:59 is kind of the opposite it's like be a
13:01 bit more flowery in your language and
13:03 and you know treat it like a person and
13:06 say okay you know I've got an idea for
13:08 the Great American novel and here's some
13:11 ideas and I want you to help me organize
13:13 the outline and that that you know ask
13:16 me some questions about it and you give
13:17 it lots of context and then it does
13:19 pretty well
13:20 oh snap big bing plug-in now too really
13:24 did that happen
13:26 if that happened that would be cool
13:29 that's Poe that's not big that's Twitter
13:32 that's not Bing here's Bing chat
13:38 I don't send a plugins
13:42 why do you say the plugins I did not see
13:45 the plugins
13:47 um Microsoft has said they will support
13:49 plugins wait let me reload this page
13:53 let me go to chat
13:55 let me go to reset new topic
13:58 I've did not see the plugins
14:02 feedback
14:04 am I dumb dumb
14:06 am I dumb dumb I'm a bit dumb dumb if
14:09 I'm a dumb dumb anyway uh Microsoft just
14:13 said they are going to support chat
14:14 gpt's plug-in architecture I don't see
14:17 it here though so if you see it if I'm
14:18 just missing it wait do I need to do
14:20 what what you need to do
14:22 do I need to set it in my settings shut
14:25 airings
14:30 I don't see him anywhere all right if
14:33 they're in there and I can't find them
14:34 tell me and I'll show everybody all
14:36 right could you give it
14:38 the link to your profile yes you could
14:40 okay so so let me show you a cool thing
14:42 um with uh Chachi VT connected to if
14:46 you're new here my name is Kyle Shannon
14:47 this is the AI learning lab uh I wax
14:49 poetic about AI crap so if you want to
14:52 talk about it pop things down below I
14:55 try to answer everything so I am already
14:57 behind I'm 20 messages behind so if it
15:00 seems like I'm behind that's because I
15:03 am
15:04 um okay so if you get chat GPT and you
15:07 pay for GPT plus that's the little
15:10 yellow flag there if you want to turn on
15:13 um the internet and the plugins you have
15:15 to go to your settings
15:16 so at the bottom of the screen here
15:20 I click on my name I click on settings
15:22 and then you go to Beta features
15:26 and you see where it says browse with
15:28 Bing and plugins you have to turn those
15:30 on because they're beta features I guess
15:32 they
15:33 maybe they don't let too many people
15:35 using it's funny Sam Altman said the
15:36 other day that he doesn't feel that
15:38 there's a product Market fit for plugins
15:40 but he's made it really [ __ ] hard to
15:41 find them so you got to do that and then
15:44 now I've got three different modes of of
15:47 gpt4 I've got default I've got browse
15:49 with Bing and I've got plugins so I just
15:53 click uh browse with Bing and now if I
15:55 go to my LinkedIn profile I copy it
16:00 and then I go back to chat JP Tay make
16:04 money with chat back hey
16:07 um based on this
16:10 LinkedIn
16:12 profile
16:14 colon
16:15 URL right wait sorry
16:19 URL
16:22 um let's see
16:25 um
16:26 what
16:30 um
16:31 what are the most relevant
16:34 experiences
16:39 um
16:40 if I want to be a tick tock influencer
16:46 [Laughter]
16:48 even though I don't know how to twerk
16:54 [Laughter]
16:58 there will be no twerking on this
16:59 channel I promise
17:02 it will be bad for all involved
17:04 I'm sorry but I'm unable to access the
17:07 LinkedIn profile directly due to
17:09 linkedin's website oh I guess it can't
17:11 log in because it's you've gotta you've
17:14 got to sign in
17:15 so I can't do that but let's okay
17:17 so there's more than one one mile
17:21 there's more than one way to skin the
17:23 cat which is that's probably triggering
17:25 to someone because like you shouldn't be
17:26 violent to animals I'm I'm old in Gen X
17:30 and that's just the way I talk so sorry
17:32 all right I mean no harm to cats let's
17:35 go to plugins so so here's the other
17:38 thing when you have the way plugins work
17:41 you you have these three modes you go to
17:43 the plugins mode
17:45 and then you click on this
17:48 pull down menu and you see a long list
17:50 of plugins that I've got installed and
17:53 then for each session you can activate
17:55 up to three so I've got three activated
17:57 right now that are related to travel
18:00 because I showed that as an example last
18:03 time but let's go to the plug-in store
18:08 so the plug-in store has it takes it a
18:12 while to load because they're probably
18:13 using some shitty database here because
18:15 because chat GPT was tossed together
18:17 over a weekend basically
18:21 and that's not too far from the truth
18:23 come on man load load all right
18:28 we're gonna go all let's see wow there
18:31 are now
18:32 98 pages of four plugins each so like
18:36 three days ago there were 250 now
18:38 there's almost 400. holy [ __ ] okay so
18:42 we're gonna go to search plugins we're
18:44 gonna see if there's a LinkedIn plugin
18:46 LinkedIn no plugins for LinkedIn all
18:48 right there you go go read the
18:51 documentation for writing plugins
18:52 they're really easy make a LinkedIn
18:54 plugin you'll be rich yeah
19:00 oh I don't know if you knew this if you
19:01 haven't been here before I'm an idiot
19:04 um
19:06 [Music]
19:08 all right all right it looks like it's
19:09 just it's just a a handful of us in here
19:12 we lost uh we lost The Interlopers this
19:16 is this is this is the community right
19:18 here okay
19:22 so the answer to can I just use my
19:25 LinkedIn profile is no because you have
19:27 to log in and there's not a plug-in for
19:28 it I just asked chat gbt to make a chat
19:31 EBT like site using hugging face models
19:34 awesome did it work that's super cool
19:36 crazy ass guy and I'm not calling him a
19:39 crazy ass guy because I'm being
19:41 derogatory that's his self-proclaimed
19:44 name
19:46 crazy ass guy you're crazy ass guy
19:52 all right no religion take the religion
19:55 out
19:57 please I used AI to summarize a macro
20:00 what
20:02 revonomic
20:04 I found it weak AI confused prediction
20:08 and projection
20:10 um okay I don't know what that is if you
20:13 if you have really deep knowledge in
20:16 something so so I don't
20:19 I have no idea how how you're using it
20:23 if you're prompting catch EBT in kind of
20:26 a generic way it may give you generic
20:29 answers there may be areas where where
20:31 it doesn't have things if you haven't
20:33 tried it tell it to act as a specialist
20:36 in oh I I assume that maybe that means
20:38 macro economic
20:40 macro revonomic I don't know what
20:42 revonomic is but if it meant if you
20:44 meant economic
20:45 but basically tell it to be a specialist
20:47 in that area the the way the large
20:49 language models work if you don't narrow
20:52 it into kind of a semantic Canyon where
20:55 where you're so so here's a way to
20:57 explain it Andre carpathy from openai
21:00 was explaining how these things were and
21:02 and
21:03 um oh [ __ ] what was the example he used
21:06 um oh
21:07 okay he was saying so if you ask chat
21:09 GPT to solve a complicated math problem
21:12 for you it'll sort of look at all of the
21:15 data it knows which is essentially all
21:18 of the internet and it and it will say
21:20 okay this person wants to solve the math
21:22 problem and it will look at all of its
21:24 data and it'll find all the places where
21:26 people solve that math problem now
21:28 within all those people that solve that
21:30 math problem there are math professors
21:32 who know how to solve it and then there
21:35 are lots and lots of students who don't
21:36 know how to solve it chat apt doesn't
21:38 know the difference it's just looking at
21:40 all of them and sort of amalgamating
21:43 answers and that's why it hallucinates
21:45 right because it's it's sort of pulling
21:47 things from everywhere it doesn't
21:48 actually know it's not copying and
21:49 pasting specific references it's kind of
21:52 averaging the [ __ ] together so if you
21:55 say I want you to act as an advanced
21:57 mathematics professor and I want you to
21:59 answer this specific kind of problem it
22:00 narrows it into a semantic Canyon that's
22:03 going to be more likely to have correct
22:05 answers in it so if you didn't do that
22:06 that's something worth trying and there
22:08 there are also things that chatgpt is
22:10 not going to be good at right now and
22:13 also four will be better than 3.5
22:16 um could you have instead entered it
22:18 into chatgpt first and then typed a
22:21 second prompt referencing it
22:25 um
22:27 that's kind of what I did I mean on the
22:30 on the second prompt
22:32 um but you can't there's not a way so
22:35 chat TBT is a very very thin application
22:38 right now it doesn't have a lot of
22:40 features it doesn't have memory like
22:41 it's got It's got sort of this 3000 word
22:45 sliding context window that sort of
22:48 member memory in a session and you can
22:50 go back to a chat
22:51 um but but it doesn't quite have memory
22:53 you can't like dump it in and say I want
22:54 to keep referencing that thing I think
22:57 that's something that will come like
22:58 that's a really good feature that they
23:00 should do
23:01 all right chat GPT can make a person
23:03 feel like Hugh Hefner uh okay
23:06 um I was just telling you how to write
23:08 one it didn't tell you anything about
23:10 your paste past
23:13 it's just telling you how to write one
23:15 it didn't tell you anything about your
23:17 past yeah it did
23:19 it did it told me I found it agency.com
23:21 it told me I founded storyline it told
23:23 me I was the chief generative officer at
23:25 Clinton evolution
23:27 now like I was just doing it quick on
23:30 the Fly Like if I were actually looking
23:32 at that I probably dig a bit deeper
23:34 um you should like the fact that you're
23:37 critiquing it is good like that's that's
23:39 a thing you should do when you work with
23:41 these large language models is critique
23:42 them oh by the way it looks like there's
23:44 new people here here if you're wondering
23:46 who the hell I am
23:47 I'm not Rosie O'Donnell and I'm not fat
23:50 Alec Baldwin but I've been accused of
23:52 that in the past week a couple others
23:54 too
23:55 um these URLs right here if you've not
23:57 played with with chat GPT it is my
24:00 strong strong wish desire that you just
24:02 go there and do it the top one is the
24:04 official chat GPT website the next one
24:06 down is Bing that's Microsoft's version
24:08 of chat GPT it's gpt4 for free and po
24:11 lets you play with a bunch of them
24:13 prompts.chat will teach you about
24:15 prompting and give you a bunch of
24:16 examples to go play with and like show
24:19 you how [ __ ] capable these models are
24:21 and then future PD is just like an
24:23 infinite list of all the AI tools like
24:25 you know every tool a rabbit hole so if
24:29 you've got ADHD this is your time in
24:31 history because these tools are [ __ ]
24:34 insane have you seen Jen Jen uh
24:37 Gen 2 by Runway check this out
24:41 80d moment of the day uh
24:45 let's see Twitter
24:48 like old woo the Twitter
25:00 why is this
25:03 there we go to my profile
25:08 where do I put it
25:15 now maybe I just did it on LinkedIn hold
25:17 please
25:24 yeah so so I did this thing on LinkedIn
25:26 so check out this video so
25:29 this was this video was this was my
25:31 first attempt at making a text to video
25:34 using Runway ml their Gen 2 text to
25:38 video tool so I typed in woman walking
25:40 through field of flowers with the breeze
25:44 blowing cinematic tracking shot it only
25:46 does four second Snippets and they don't
25:48 have any sound but but this was
25:50 literally my first attempt and it made
25:53 that in like 30 seconds
25:55 so just from a text prompt so this is
25:58 this is where we're heading and it's
25:59 it's uh runwayml they have an app the
26:02 app
26:03 right now it seems like they're rolling
26:05 this feature out to to I don't have it
26:08 yet in the in the iPhone app but if you
26:10 go to their Discord server
26:13 you basically go into
26:17 so I'm in the if you don't know what
26:19 Discord is this this is Discord these
26:21 are different servers so I'm in the
26:23 runway server and then these are
26:25 different channels and see these things
26:27 that say generate one two three so
26:29 that's where I can generate something so
26:31 how you generate something in the runway
26:33 server
26:34 is you say at
26:37 gen
26:39 Dash 2
26:41 and then you type in your prompt so
26:43 we'll do
26:44 um
26:45 let's see
26:47 1901 race car
26:51 um
26:55 um
26:55 in Indian
26:59 appleis 500.
27:03 and then so it goes off and generates
27:06 something and the other thing it does
27:08 which is weird so I'm in this Gen 2
27:11 Channel and it says it it sent my thing
27:14 off and then what it does it it actually
27:16 spawns a thread
27:18 so so this sort of Gen 2 thread
27:21 where it's going to actually show the
27:24 thing but it says text prompt 1901 race
27:27 car
27:28 CFG scale 10 interpolate false upscale
27:31 false there's the seed number so the
27:34 seeds cool with these generative AI
27:36 things like if I want to do other
27:37 Generations that are similar to this one
27:39 once it comes out I can use that seed
27:41 and then green screen is false so so
27:44 those are some of the parameters that
27:45 are available available to you obviously
27:47 and then that's going to go off and do
27:49 its thing I don't know how long it's
27:50 going to take but I'll come back and
27:51 check that out anyway that's gen 2.
27:54 that's pretty slick stuff
27:57 all right week I don't know what was
27:59 weak but probably me I am weak I'm weak
28:02 I'm an old man I'm old I'm tired the
28:05 muscle mass is drifting away it's
28:08 falling off the bones I'm getting
28:10 brittle
28:12 can you make money with Chachi BT you
28:14 can make money
28:16 here's the deal Chachi PT is just a tool
28:19 you can make money with anything
28:21 you can make money with a bread maker
28:23 you can make money with a shovel you can
28:25 make money with Photoshop you can make
28:27 money with chat gbt
28:30 the first step is to understand what the
28:32 tool is and what it's good at the the
28:35 amazing thing about chat EBT is it's
28:37 good at a lot
28:39 how you make money is you solve
28:41 someone's problem in an elegant enough
28:43 way that they're willing to pay you
28:44 money for it like that's business that's
28:46 that that's how you make money with chat
28:48 GPT
28:50 so it could be you want to make t-shirts
28:51 okay great chatgpt makes it really easy
28:54 to make t-shirts that's great you know
28:56 what's the bad part it makes it really
28:58 easy for other people to make t-shirts
29:00 right so so the comp the competition for
29:03 t-shirts just went up right because it's
29:05 it's going to be really easy for people
29:06 that historically weren't good at
29:08 copywriting now they're going to be
29:09 great at copywriting
29:11 um so so what I would say is it's why I
29:14 keep pointing people to these um these
29:17 URLs if you haven't played with these
29:19 tools Step One is go play with the tools
29:21 and what I would say is
29:23 take the thing that you're really good
29:24 at and start digging deeper on the thing
29:27 that you're really good at see what they
29:28 can do and then start thinking about huh
29:30 have I ever had an idea for a business
29:32 or is there anything where I could you
29:34 know solve someone's problem or
29:35 entertain them right whatever is is
29:38 within kind of Who You Are
29:40 and then use chechi PT to help you
29:43 figure that out and that's how you make
29:45 money with chat gbt all right
29:48 um that was sort of a random question in
29:50 the middle I'm trying to go back to
29:51 where I was before because I have some
29:53 sort of weird OCD thing where I gotta
29:55 answer all the questions I'm like must
29:57 answer all the questions oh I missed one
30:00 what am I gonna do
30:02 and that's how I normally talk
30:05 by the way
30:08 all right I'm back to where I was
30:13 I gotta find the one where the where the
30:15 Nadia called me weak
30:18 Nadia you've been here before I don't
30:19 think you were calling me weak I think
30:21 you were calling something it did we
30:23 definitely hi Jen
30:25 uh he's gonna need one terabyte of space
30:28 that's why we have cloud storage because
30:30 because between Google Drive and iCloud
30:34 like I don't worry about my hard drive
30:37 size anymore just [ __ ] goes up to the
30:39 cloud and then I eventually
30:42 I don't ever throw anything away I'm a
30:44 digital hoarder by the way anyone else
30:46 digital hoarders Gen X or digital
30:49 hoarders it's like I'll go back to that
30:51 40 megabyte hard drive
30:53 that I that I filled with [ __ ] in 1994.
30:57 here's a
30:58 for the Gen xers out there here's the
31:00 side quest drive from Urban desires my
31:02 uh online magazine that I started in
31:04 1994.
31:06 why do I have that I don't know because
31:08 I might need it someday
31:11 bro should try the Falcon 70b model I
31:15 know that sounds amazing the the problem
31:16 crazy ass guy about trying all these
31:19 these um these uh open source models is
31:22 like a new one comes out literally like
31:24 every other day there's a new [ __ ]
31:26 model it's like it's like I'm exhausted
31:29 it's like
31:32 like at some point when someone does
31:34 kind of the the automatic 1111 install
31:36 where where where you can just put it on
31:40 your local drive it runs on an M1 Mac
31:42 which is what I've got
31:44 um
31:45 I can just have a pull down menu of all
31:47 the language models and like it defaults
31:49 to the latest one or some [ __ ] like that
31:51 and then it makes it as good as anything
31:53 like like at that point I'll start using
31:55 it but right now like I don't have time
31:57 I don't have time to dig into these
31:59 things
32:01 um I found it weak confused it
32:02 projection
32:04 and prediction in macro Evo or Eco
32:09 economy Evo I don't know what you're
32:11 talking about Nadia
32:14 um keep pushing it harder
32:16 like the thing that I found about it is
32:18 is I find that that the initial passes
32:22 that it does on stuff suck
32:25 and then if you get better at prompting
32:27 they suck less but like I have yet to
32:30 have it
32:31 I've gotten it to like
32:34 90 95 I've never had it generate
32:37 anything that I didn't need to edit
32:39 anything in and most of the time it's
32:41 between like 60 and 80 percent
32:44 and then I won't take the 60 and start
32:46 working with it I'll keep [ __ ] with
32:48 the prompts until I get it to like
32:49 between 80 and 90 and that's the the
32:52 percentage I'm talking about is just
32:53 like subjective like that writing is
32:56 good enough that I can go from here
32:59 um so it it does need a lot and if what
33:01 you're talking about is kind of deep
33:03 research or deep analysis it might not
33:05 be good at that [ __ ] the thing I'd keep
33:07 your eye out for though is code
33:08 interpreter which if what you're talking
33:10 about is doing more analytical stuff on
33:13 data code interpreter might be might be
33:16 the thing I don't I haven't had a chance
33:17 to play with it yet that for for those
33:19 of you that don't know code interpreters
33:21 a new kind of mode of chat GPT where you
33:24 can upload a document to it like a giant
33:26 spreadsheet and it will just read the
33:29 spreadsheet and then you can say what
33:30 are some interesting Trends in here
33:32 it'll it'll generate a whole report a
33:34 data analysis of what's there you can
33:36 say what um what were anomalies in the
33:40 data it'll find anomalies in the data
33:41 you can tell it to create a chart it'll
33:43 write the python code generate the chart
33:45 and hand it to you like so that's coming
33:47 so again I wish I knew more about what
33:49 you were talking about but I don't sorry
33:52 I did agree okay you should
33:55 plug your blog it's great do you mean
33:57 everyday AI so I've got a
34:00 um
34:02 newsletter it's like a Blog
34:05 I think that's what you're talking about
34:06 uh but thank you thank you for the kind
34:09 words oh by the way if you're new here
34:10 my name is Kyle Shannon this is the AI
34:11 learning lab I Babylon incoherently
34:14 about generative AI if you've got
34:16 questions about it pop them
34:19 bug pop them in the question below oh
34:21 look at the hair ah we're gonna have a
34:23 good hair night tonight I'm gonna hang
34:25 on I'm gonna put some coffee in it
34:27 because this is this is not going to
34:28 calm down all right
34:30 um
34:32 I talk about Ai and and do self quaffing
34:35 that's what this channel is all about
34:38 um I do self-deprecating humor I swear a
34:40 lot and then I'll try to I'll show
34:42 people what to do so if you're new
34:44 um these URLs right here the top one is
34:47 the official chat GPT website the next
34:49 one down is bing.com that's Microsoft
34:51 search engine but there's a chat button
34:53 there that's gpt4 for free and then
34:56 poe.com lets you try a bunch of
34:58 different chat engines and compare them
35:00 so those are just three places to go
35:02 play so while I Babble on if you haven't
35:04 used chat GPT go do it now go do it now
35:08 you won't regret it
35:10 um okay I should plug my blog my
35:12 newsletter so my newsletter if you go to
35:14 the Twitter
35:17 and you search for every day AI
35:23 let's see if it comes up with just that
35:25 yeah it should
35:28 so the the handle is at everyday AI news
35:31 the Twitter Channel we put out uh
35:34 Twitter threads on new stuff that's
35:36 coming out and then our website is at
35:38 beehive and the newsletters at beehive
35:41 so
35:44 this is everyday AI so we're up to issue
35:48 20. and each issue has a lovely
35:52 mid-journey generated cover
35:54 or or thumbnail and then we you know
35:58 each of them let's see what was a cool
35:59 one
36:01 um ai's Cambrian explosion this was
36:03 talking about code interpreter that I
36:05 just talked about before so we talk
36:07 about you know
36:08 big things that are happening and all of
36:11 these are links and then Bing chat and
36:14 mind reading and all sorts of stuff
36:17 this this was one that talked about lots
36:19 of different open source models like MPT
36:22 7B which if you haven't heard about MPT
36:26 7B shame on you where have you been so
36:29 so some of the shits in here is really
36:32 geeky but it's it's good but anyway
36:33 thank you for that I appreciate you
36:35 putting that out or uh
36:37 reminding me to plug that that's a good
36:40 idea all right
36:43 um has open AI scheduled the next big
36:45 ingestion of tag content to move Beyond
36:47 20 it's 2021 actually September 2021
36:51 I think they're working on training gpt5
36:54 right now they they may have slow rolled
36:56 it or they might slow roll the release
36:58 because like they haven't really fully
37:00 rolled out gpt4 yet it's supposed to be
37:02 multimodal meaning you're supposed to be
37:05 able to also upload images to gpt4 and
37:08 it will understand the images and you
37:10 can ask questions about it that's not in
37:12 there yet it's supposed to have a 32 000
37:15 token window that's not in there yet
37:18 um
37:19 but it is connected to the internet now
37:22 so that limitation is effectively gone
37:24 right so if I go to chat GPT stop making
37:27 that stupid [ __ ] joke probably never
37:32 okay so if you pay for GPT plus and
37:36 you've enabled the beta features go to
37:38 your settings and enable the beta
37:39 features you now have this option called
37:41 browse with Bing so I can say
37:47 um is it supposed to rain tomorrow in
37:53 Denver
37:54 that is clearly past September 2021
37:58 that's actually in the future
38:01 so it says it went out to Bing and it
38:04 says Denver weather forecast you could
38:06 you so you can open this this uh plug-in
38:08 sort of thing here and it'll tell you
38:10 what it did it pulled the hourly
38:11 forecast it says yes it's supposed to
38:14 rain tomorrow in Denver Colorado
38:15 according to the hourly forecast
38:18 and then it gives you a citation so the
38:22 current gpt4 is is not limited the core
38:25 base model does still stop at September
38:27 2021 but now that you've got it
38:30 connected to the internets then it's
38:32 much less of an issue than it used to be
38:36 all right oh I thought you were using a
38:39 bing plug-in in oh in gpt4 like you can
38:44 Wolfram Alpha ah okay so Jesse yes this
38:47 is very confusing so
38:50 open Ai and Microsoft are doing what's
38:52 in the in the marketing world called you
38:55 know cross promotion or co-marketing and
38:58 it's good for them as Brands and it's
39:00 confusing as [ __ ] for us as people so
39:04 all right bing.com is Microsoft search
39:08 engine
39:09 and it
39:17 it looks like in the top left
39:22 then it gives you this thing that is
39:26 that is ostensibly chat GPT
39:29 but it's connected to Bing chat which
39:32 this already is like so this is chat GPT
39:34 integrated inside Bing's chat so so and
39:39 it's and it's gpt4 and it's free okay so
39:44 that's that then within the official
39:46 chatgpt website
39:48 this is this is totally confusing like
39:51 even trying to explain it I'm like wait
39:52 what how the [ __ ] is this working okay
39:54 within the chatgpt website you've got
39:58 gpt4 which is like the default is just
40:01 normal gpt4 but then this says browse
40:03 with Bing that's basically just gbt4
40:07 connected to the internet they're using
40:10 probably Bing search technology to
40:12 search the internet but what that
40:14 effectively is is just a bing plug-in is
40:17 what it looks like and then you've got
40:19 plugins and I'm sure there's a plug-in
40:20 in there that's that's Bing as well so
40:23 it's just really confusing right now so
40:25 thank you for pointing that out it
40:28 so here's the other thing back in the in
40:32 the in the mid 90s when I was
40:34 um started doing uh interest well I
40:36 don't know if it was interesting I was
40:37 starting companies using the the early
40:40 World Wide Web Technologies when when
40:42 the World Wide Web was first coming out
40:43 we were figuring out how to use it
40:46 um it was very very similar to what's
40:48 going on right now there were like you
40:51 know a thousand tools everyone claimed
40:54 their tool did the magic thing and all
40:56 these little tools were just these
40:57 little sort of nuggets of utility
41:00 and if you weren't obsessing about it
41:03 daily like the world would kind of the
41:05 technical world would kind of pass you
41:06 by so if you were in it you were kind of
41:08 like obsessive if you came out in it
41:10 from the outside it just seemed like
41:12 this chaotic [ __ ] nightmare which is
41:15 exactly what we're in the middle of
41:16 right now so
41:18 if you don't feel like getting OCD about
41:21 all this stuff then the reason I put
41:23 these URLs up on screen is these are
41:25 just like what where I would start if
41:28 you if you're trying to get your head
41:29 around this generative AI stuff is just
41:31 go to chat.openai.com just start using
41:34 chat gbt and and start figuring out what
41:36 it can do all that other stuff right now
41:38 is noise because what's going to happen
41:40 is
41:41 all of those individual toys are going
41:43 to start to get integrated into this
41:46 thing and chat apt at some point will be
41:48 truly multi-modal where you can just say
41:50 I want a song I want an animation I want
41:53 a movie I want a picture I want a novel
41:55 I want a graphic novel and out it'll
41:57 come right that's kind of three to five
41:59 years that's where we'll be but right
42:02 now the the good news is we as mere
42:05 mortals get to play with these tools
42:07 the the bad news is it's a chaotic
42:10 [ __ ] nightmare
42:12 all right AI any AI bone skeleton type
42:17 animation
42:19 if you mean bone skeleton type animation
42:23 as in like 3D modeling I assume you
42:26 don't mean
42:27 um
42:28 medical bone skeleton oh what the [ __ ]
42:32 is it called it just came out to me Toby
42:35 oh Jesus
42:37 [ __ ] I don't remember the name of it my
42:40 brain my brain is failing me let's see
42:44 text to
42:46 3D character
42:51 Taffy
42:53 check this out
42:56 I think this is what you're asking about
43:03 so it says so there's your dude it says
43:07 made it make him shorter now so this is
43:09 they're just inputting text
43:12 it's giving him uniforms
43:14 what if we made the character a woman
43:17 now it's a woman
43:21 um take out the hair add a buzz cut do
43:23 something else right there's your buzz
43:25 cut on the on the woman
43:30 now they're back to the man uh
43:33 four stages of him turning into
43:36 a monster
43:41 so this is a company that's already
43:42 doing 3D character development that that
43:46 basically is throwing a large language
43:48 model on top of it so this gives you
43:51 fully rigged
43:54 fully customized characters by
43:57 describing them and it is at
44:02 Tom
44:04 where is it Taffy announces and this is
44:07 not it's not available yet so this is a
44:10 press release
44:11 Taffy t-a-f-i
44:15 why these press releases never have
44:17 links
44:20 why why you people why you do this to me
44:24 you make me so confused you made me so
44:27 confused I want to cry
44:29 oh
44:33 all right
44:34 make taffy.com m-a-k-e-t-a-f-i.com
44:41 and then it says Join Beta now I think
44:44 I've joined but now that I'm here let's
44:46 go ahead and join
44:48 name Kyle Shannon
44:51 Kyle is Story vine
44:53 all right my request has been submitted
44:56 so now
44:58 you know as much as I know and that
45:00 that's what this Channel's about I know
45:02 like this much more than anyone else on
45:04 this channel and probably a lot less
45:06 than most of the people on this channel
45:08 um but that's the whole idea like right
45:10 now like like the only my big advice is
45:14 get curious get curious get adventurous
45:17 just start playing
45:19 um because if you think that these
45:22 tools are not going to affect your work
45:25 I think you're wrong
45:29 bro what you need to reinst wait you
45:32 need to reinstall big I don't know what
45:35 that means Bing
45:37 oh wait
45:39 do you install Bing
45:41 huh
45:43 wait there's there's no Bing at oh is it
45:45 on your phone
45:47 Bing
45:50 is there a Bing app
45:55 oh please
45:58 uh
46:01 let me see if it's like Bing
46:04 chat GPT
46:07 plugins it's it's probably the app on my
46:11 phone then
46:12 plugins interesting
46:15 open chat GPT that's telling me how to
46:17 do it there
46:21 will get boat boosted
46:23 anyway okay I'll check it out so so what
46:27 um
46:29 who just said that uh Nadia what Nadia
46:32 is saying is
46:33 um so so Microsoft has said that Bing is
46:36 going to support all the plugins that
46:38 openai has um I can't find it she says
46:41 it's there
46:42 did you get the plugins with the 10 a
46:45 month version or the 20 a month version
46:47 I don't think there is a 10 a month
46:49 version
46:51 um
46:52 if you've gone to the app store and
46:54 bought a chat cheap GPT app that cost
46:56 you money it's probably you're probably
46:58 getting ripped off
47:02 there are only two official places to
47:05 use the actual chat gbt
47:08 chat.openai.com and then the official
47:10 chat GPT app
47:13 um
47:14 so the one that I'm paying for is the
47:16 twenty dollar a month I think there's
47:18 only one level if there's if there's a
47:20 ten dollar level now
47:21 um that's something that's new that I
47:22 don't know about
47:24 all right you used
47:29 you used the Reversed accent that's
47:31 funny I thought just that when you use
47:34 the English fake accent
47:36 yeah I don't the the accents are just
47:38 you know they're a remnant of my acting
47:40 training and the fact that I'm a smart
47:42 ass and a Class Clown so
47:45 I just they pop out like whatever seems
47:48 like the appropriate idiotic thing to do
47:50 that's what I do
47:54 I don't think about it a lot this prompt
47:56 really helped me for a profile I'm
47:58 working on for my executive leadership
48:00 course awesome that's cool Jake
48:04 um and the fact that that well I'm
48:06 curious if your executive leadership
48:08 course is encouraging you to use chat
48:11 GPT or if you're using chat gbt in spite
48:14 of the executive leadership course
48:16 hopefully it's the former it was a hook
48:19 holy F and flip
48:21 is there a Facebook yes it's at
48:24 facebook.com I don't know what you mean
48:26 is there a Facebook plugin probably is
48:28 what that meant let's go look
48:31 I do not know
48:33 I can't believe there's 400 plugins now
48:35 this is insanity
48:38 well let's go see if Facebook's done
48:41 something or if someone's done something
48:42 with Facebook my guess is not because
48:46 Facebook is probably pretty pissy that
48:49 openai has gotten all the attention and
48:51 their little metaverse nobody went to
48:53 their metaverse party and they're cranky
48:56 Mark Zuckerberg was like Apple's goggles
48:59 were they were fine it just validates
49:02 that our goggles are cool too and ours
49:05 are cheaper and they do the same thing
49:06 and everything's fine
49:09 he sounded just like that that's when
49:11 Mark Zuckerberg impression
49:13 let's see face book
49:16 nope no plugins for Facebook
49:19 so there
49:22 we're early people
49:24 all right
49:26 I'm an interloper Eric you are an
49:28 interloper
49:29 you dang traffic cone photographing Eric
49:33 sander by the way is an awesome dude and
49:36 an awesome artist and is bringing his
49:40 big brain and his big creativity to the
49:43 world of AI and I think he's going to do
49:46 some remarkable [ __ ] with it so you
49:48 should look up Eric sander
49:52 e-r-i-k-s-a-n-n-e-r and he's just a nice
49:54 guy too
49:57 the hair is deciding to do things on its
50:00 own
50:02 all right wait
50:04 in your query give it the link to your
50:06 public profile oh wait how do I get hold
50:10 how do I get my public profile so so
50:13 wait if you're if you're new to the
50:15 channel uh about an hour ago we were
50:18 trying to use a a LinkedIn
50:21 hyperlink
50:23 and it wouldn't do it but Eric's telling
50:25 me I've got to go get my public profile
50:28 view full profile
50:31 is this my public profile okay like
50:34 let's see let's go as chat gbt
50:37 because that's all I do anymore I don't
50:39 use Google at all anymore like why would
50:42 you go to Google that's going to give
50:44 you like 40 [ __ ] links you know 38 of
50:46 which are you know spam fake ass SEO
50:49 optimized and injected websites when you
50:52 can just say
50:53 um what is the difference between a
50:58 private and public
51:03 LinkedIn URL
51:07 all right the private URL now I know the
51:09 difference between private and public
51:11 you dumb thing
51:13 okay I don't want an explanation I want
51:19 the syntax
51:24 of the public URL
51:28 for
51:30 LinkedIn if if
51:37 the Syntax for the public one
51:43 all right maybe I just maybe I just used
51:45 the private one all right so this looks
51:48 like it so let's go try this again all
51:50 right so linkedin.com Kyle Shannon okay
51:54 so we're gonna go back to chat EBT
51:57 we're gonna go gpt4 browse with Bing
52:00 so the basic idea here I showed people
52:02 earlier that you can copy and paste all
52:04 of the text of your LinkedIn profile and
52:06 then start having it rewrite it and
52:08 recommend how to make it better but what
52:10 we're going to try now is based on
52:13 based on this
52:17 link
52:21 there's my LinkedIn public link
52:24 um
52:27 what what's
52:29 Kyle been up to in the last five years
52:37 let's see if it
52:40 okay it says browsing the web now it
52:43 apart it appears I'm unable to access
52:46 your profile directly so I don't know
52:48 what I'm doing Eric I might just be
52:50 I might just be a dum-dum
52:53 ah
52:55 very it's very possible that I'm just a
52:58 dum-dum
52:59 all right I missed the travel Centric
53:02 live got a tldr
53:05 wait got a tldr and how to get the most
53:07 out of it so
53:09 now that there's 400 plug-ins so so I'll
53:13 just
53:14 so when you're in when you're in the GPT
53:16 and you go to plugins
53:19 like right now it says there's no
53:21 plugins enabled right here
53:23 so these are the ones I've got installed
53:26 so you can you can activate up to three
53:29 of them so what I did was I went and I
53:32 found
53:33 um
53:35 I found some plugins that were travel
53:37 plugins there was one here called
53:38 experiences there's one called get your
53:40 guide find tours excursions and other
53:43 travel activities
53:45 and then kayak was there but then I used
53:48 trip.com so I so I activated those three
53:52 and so effectively what I've got now is
53:56 a chat GPT that's connected to those
53:58 travel services so I can do specific
54:00 trip planning with them so that's
54:02 basically it and then it's just a matter
54:03 of prompting and having it give you [ __ ]
54:06 and telling it to telling it to put it
54:08 in table format and all that sort of
54:11 stuff so you can just start playing with
54:13 the uh you can just start playing with
54:15 the
54:16 um
54:17 the prompts to to dial in what you do
54:19 macroeconomic yeah Nadia like it's it's
54:23 possible I don't know if you were using
54:25 GPT 3.5 or 4.
54:27 but we're in a global hegemony Nadia
54:30 you're you're
54:31 much smarter than I am so it's possible
54:34 that I'm impressed with chat gbt because
54:37 it's giving me my level of Education
54:40 um but but that said I'm being a little
54:44 bit
54:45 um uh whatever you call it
54:46 um uh
54:48 self-deprecating it was like
54:50 self-aggrandizing was the word that came
54:52 in my head proving my intelligence
54:54 um
54:55 there there's probably a way to get it
54:58 to focus and you might need to um it
55:01 might be the kind of thing where you use
55:04 um Internet connected gpt4 and and point
55:08 it toward
55:09 um
55:12 resources like websites or papers or
55:15 things like that um where where you're
55:17 actually querying on you know some
55:19 deeper knowledge or data
55:23 um what if there's no solution to the
55:25 math problem it will probably tell you
55:27 that it like it's really good the other
55:29 thing that you can tell it to do is
55:31 instead of just telling it to give you
55:32 an answer to a math problem you can tell
55:34 it to show its work and it will walk
55:36 through the steps that that it needs to
55:39 go through
55:40 um to get there and there there's
55:42 there's new research coming out that by
55:44 giving it the by telling it to give you
55:46 the steps it actually gets better
55:49 um from an accuracy accuracy standpoint
55:51 because the other thing about these
55:52 tools is they they do what's called
55:54 hallucinating where chat CPT is not
55:57 actually aware of what it's typing
55:59 because what it's doing is it's it's
56:01 it's creating a probability of each word
56:03 individually like what should be the
56:05 next word you know to answer this prompt
56:09 but it's not aware of the whole of what
56:11 it created so it can do things that look
56:14 right but are just completely false
56:16 that's why it does that
56:18 um
56:19 what book do I recommend for a good
56:21 foundation on Ai No well there are
56:24 probably a lot of books on like machine
56:27 learning if you want to learn how to do
56:29 the machine learning and the training
56:30 and all that sort of stuff
56:32 um I would say just go hang out on
56:34 Reddit subreddits GitHub and hugging
56:38 face and just go start hanging out with
56:40 that crowd and watching YouTube videos
56:42 um if you're talking more about this
56:44 generative AI stuff anyone that writes a
56:46 book on this [ __ ] right now is full of
56:48 [ __ ] no no one knows about this there's
56:50 no experts right now I'm not an expert
56:52 I'm just a dude just barely ahead of
56:56 where anyone else here is with the
56:58 generative AI stuff because it's all
56:59 it's all very new
57:01 uh
57:03 uh use cases for the furniture business
57:05 well
57:06 so so one is and and this is I I think
57:10 we're we're entering oh that's the
57:13 that's the 1901 race car that I made
57:15 earlier on
57:18 oh that's really funny I said a 1901
57:20 race car at the Indianapolis 500
57:24 and it made like a it's it's it's like
57:28 that's like the front wait sorry like
57:30 this is like the front of a race car but
57:33 this is like some weird suspension
57:35 bridge and it's like in front of these
57:36 Greek ruins it's like it's
57:39 and it's kind of he's got a helmet on
57:42 like a sun hat
57:46 oh I don't like my result I'm gonna I'm
57:48 gonna do some training here there we
57:50 just did some training for Gen 2.
57:52 um what was I saying oh the furniture
57:53 business okay so let's pop over to a mid
57:55 Journey Channel
57:57 and Let's do let's go to Mid journey and
58:01 let's do oh wait I'm gonna do a cool
58:04 thing okay
58:06 so one of the one of the things that I
58:08 show you here is poe.com and
58:12 one of the cool things about poe.com
58:18 is that you can make bots so I made a
58:22 Shakespeare bot that when you ask it
58:23 stuff it responds in iambic pentameter
58:26 which is useful to no one but it's kind
58:27 of cool I made a yodobot which is useful
58:30 to everyone because Yoda is the best
58:32 but someone made this mid-journey bot so
58:35 so if you just go to I am or you go to
58:37 po.com slash mid Journey
58:40 it creates a thing that will rewrite
58:44 your shitty prompt for mid-journey into
58:46 a really specific one so I want to do
58:50 um
58:53 mid-century modern chair
58:57 um
58:58 in a Star Wars theme
59:05 um photo
59:08 um of a dis of an award-winning design
59:16 prototype
59:19 so I typed that in and then it rewrites
59:22 it for me
59:24 color photo of an award-winning design
59:26 prototype for a mid-century modern chair
59:28 with a Star Wars theme the chair is
59:30 Sleek and minimalist clean lines the
59:32 Star Wars theme is subtly integrated
59:34 into this design so this bot like takes
59:38 your shitty prompt and like kicks it up
59:41 a notch which is super cool so now we're
59:42 going to go over to Discord this is
59:44 Discord and I'm in the AI Salon Discord
59:47 in the mid-journey play Channel
59:50 and the way you make these mid Journey
59:52 things is you type in slash imagine
59:55 and then it there's an area for the
59:57 prompt and so I'm literally just going
59:58 to paste in the prompt that that bot
1:00:00 wrote and hit return
1:00:02 and if you're thinking to yourself Kyle
1:00:06 this seems like a really complicated
1:00:08 Nightmare and hell of an interface to
1:00:11 make an image you're right
1:00:15 um again oh so by the way everyone hi my
1:00:17 name is Kyle Shannon this is the AI
1:00:19 learning lab uh I talk about all things
1:00:21 AI here I am more than a hundred
1:00:23 messages behind I try to answer all the
1:00:25 questions but I fall really far behind
1:00:27 so if it seems like I'm not paying
1:00:30 attention to you that's because I just
1:00:31 haven't gotten to you yet
1:00:33 um so someone just asked a question
1:00:35 about what's a use case in the furniture
1:00:37 business so while the while this thing
1:00:39 renders um and it's cool as we're
1:00:41 watching this thing render right it's
1:00:42 it's you kind of get to watch the
1:00:44 machine
1:00:46 um sort of denoising this this very
1:00:49 nondescript image and it gets
1:00:50 increasingly more detailed that's that's
1:00:52 the way the stuff works pretty cool
1:00:55 um
1:00:56 you could use it to do things like
1:00:58 marketing planning coming up with you
1:01:00 know social media content writing
1:01:01 scripts for Tick Tock but one of the
1:01:04 things that I think is particularly
1:01:05 exciting these are now at 93 they'll get
1:01:08 one level more detailed and then I'll
1:01:09 sort of zoom in on zoom in on them
1:01:12 so one thing that I think is going to be
1:01:14 a whole category of
1:01:18 um
1:01:19 of innovation is designers concepting in
1:01:24 mid-journey like look at that like
1:01:26 that's a mid-century modern chair that
1:01:29 Star Wars you know inspired I mean
1:01:32 that's [ __ ] cool like go make that
1:01:35 like someone would buy that you know and
1:01:37 and whether or not you get the the Star
1:01:40 Wars license or not I like you know you
1:01:42 don't need to but but like this was this
1:01:45 was one stupid idea and this was
1:01:47 actually inspired by on the mid-journey
1:01:50 channel in um in Facebook the official
1:01:52 mid-journey Channel there's this guy
1:01:54 that did Star Wars themed KitchenAid
1:01:57 mixers and they're they're just so
1:02:00 [ __ ] cool like look at that thing I
1:02:02 mean that's just bitching right a
1:02:05 designer would look at that and go oh
1:02:06 that gives me these six ideas right like
1:02:09 so you don't even have to do the
1:02:11 execution work in mid-journer you can
1:02:13 just use it for concepting so I would
1:02:14 say Chachi BT can help with you know all
1:02:17 of the writing of the business um
1:02:20 could also help write the descriptions
1:02:23 of New pieces or Furniture lines or
1:02:25 things like that
1:02:26 all right if you want chat EBT to have
1:02:29 memory you need to pay for premium yeah
1:02:31 exactly
1:02:33 um they don't have that yet but they
1:02:35 will
1:02:37 how much is chat GPT it is 20 a month
1:02:39 well it's okay it that is a complicated
1:02:43 question because the world is
1:02:45 complicated right now so
1:02:47 chat GPT the official chat EBT site
1:02:50 their 3.5 version is free
1:02:53 and it's actually really good and really
1:02:55 fast so there's no there's no shame in
1:02:57 using 3.5
1:03:00 if you want to use four and you want
1:03:02 plugins at the official chat gbt site
1:03:05 that's 20 bucks a month if you go one
1:03:07 URL down bing.com is gpt4 but it's in
1:03:11 within Microsoft search engine in my
1:03:13 opinion it's a little lobotomized due to
1:03:17 some issues some PR issues they had
1:03:19 Jesus sorry the Italian food is like
1:03:22 hiccuping and burping and convulsing
1:03:25 this is you know if I just topple over
1:03:27 you know call the authorities
1:03:32 um but bing.com is sort of a free
1:03:35 version of it
1:03:36 and then po.com's got some free things
1:03:38 in there and they've got some you pay
1:03:39 for
1:03:41 all right
1:03:43 mm-hmm
1:03:45 will you forever be a
1:03:48 you will forever be Alec Baldwin to me
1:03:50 so the inside joke there is someone
1:03:52 someone my trolls are somewhere between
1:03:55 trolls and compliments so they said I
1:03:57 was scrolling Tick Tock and thought I
1:03:58 saw Alec Baldwin but he got really fat
1:04:06 activating thanks for sharing you're
1:04:07 welcome I missed the travel Centric live
1:04:09 we talked about that it's a rabbit hole
1:04:12 for sure these things are rabbit holes
1:04:14 my ADHD partner uses chat GPT daily it
1:04:17 helps him so much yeah if you've got
1:04:19 ADHD uh generative AIS is your [ __ ] man
1:04:23 it's like like
1:04:25 I find it very hard to get overwhelmed
1:04:28 with like like I can take very small
1:04:30 pieces of data and synthesize them
1:04:33 really quickly so I can normally get my
1:04:35 head around like a technology or a
1:04:37 concept really quick this AI [ __ ] like I
1:04:40 can like it's it's been six months of
1:04:42 just me like being completely
1:04:44 overwhelmed like with the possibilities
1:04:47 of
1:04:48 just just the implications of this [ __ ]
1:04:51 it's it's insane
1:04:52 squirrel yes
1:04:54 um
1:04:56 need some Chisholm just to touch touch
1:04:58 of the Chisholm ah someone's been here
1:05:00 for a while we've got some return
1:05:01 visitors
1:05:03 um
1:05:04 I have a son who has a touch of the
1:05:06 Chisholm I probably have a touch of the
1:05:07 Chisholm we're neurospices as they like
1:05:10 to say on tick tock
1:05:15 but [ __ ] it you know what being
1:05:16 neurotypical be boring wouldn't it
1:05:21 all right
1:05:22 when you advise flowery content you
1:05:25 presume that people
1:05:27 don't know
1:05:28 chat
1:05:30 GPT when I advise flowery content
1:05:35 yeah well I I kind of have to assume
1:05:37 people don't know chat gbt just because
1:05:40 like right now we were down at like 70
1:05:42 now we're back up to 90. so 20 new
1:05:44 people showed up I don't you know they
1:05:47 might have just stopped because I looked
1:05:48 like Rosie O'Donnell and they're like oh
1:05:50 let me see what Rosie's up to why is she
1:05:51 talking about AI what happened to her
1:05:53 voice
1:05:56 so so I couldn't I kind of assume that
1:05:58 that that people don't know what they're
1:06:00 doing I might start doing
1:06:02 um some sort of uh
1:06:04 like a workshop or something like that
1:06:06 where we actually dig in and we can sort
1:06:08 of you know gauge the level of people
1:06:11 so I don't disagree with you I
1:06:12 appreciate the feedback
1:06:15 all right so I disagree with you
1:06:20 it is an insult to our intelligence well
1:06:23 you don't have to stay like if if I'm if
1:06:25 I'm talking down to you you can go you
1:06:27 don't need to say I'm you know here's
1:06:29 the deal like I don't know who's in the
1:06:31 room I don't know who's got what uh what
1:06:34 level of experience so
1:06:36 um I can't I just can't assume that you
1:06:38 know everyone's been watching my life so
1:06:41 so anyway I would go if if you don't
1:06:44 like what I'm doing
1:06:46 I'm just a humble Country Doctor please
1:06:49 people don't attack me I'm weak I'm
1:06:53 frail
1:06:55 not really I don't really give two shits
1:06:58 but I am trying to provide value so
1:07:02 um
1:07:04 [Music]
1:07:06 and and now Tick Tock has scrolled me
1:07:09 all the way to the bottom I'm trying to
1:07:10 scroll back up
1:07:12 um have I experimented with small I've
1:07:14 not
1:07:16 what's your Discord tag Kyle Shannon
1:07:19 I just changed so my Discord tag used to
1:07:21 be Kyle
1:07:24 shannon383808 and they just changed the
1:07:26 the names in Discord and I grabbed Kyle
1:07:29 Shannon his son I am now officially Kyle
1:07:31 Shannon on the Discord
1:07:33 which when I got into blockchain and
1:07:35 nfts everyone's like you know don't dox
1:07:38 yourself man you shouldn't use your real
1:07:39 name I'm like well I've used my real
1:07:41 name for like 30 years on the internet
1:07:44 and social media so I guess I'm screwed
1:07:46 at this point but whatever
1:07:49 I'm not called like wait what's that
1:07:52 what's that one what's the dude that's
1:07:53 in here crazy ass guy like I don't have
1:07:55 I don't have a cool name like crazy ass
1:07:57 guy
1:08:01 um
1:08:02 this prompt really helped me for profile
1:08:05 oh so now now I'm behind hang on I'll
1:08:07 get caught up I promise
1:08:11 okay use cases for the furniture
1:08:12 business memory I'm getting back to
1:08:15 where we were ADHD partner
1:08:18 Nadia yelling at me for being
1:08:20 condescending and treating people like
1:08:23 they're stupid
1:08:26 um
1:08:30 [Music]
1:08:31 I just saw the Discord tag you don't
1:08:34 need to yell
1:08:36 um oh there's a hundred people here
1:08:37 hello if you're new here my name is Kyle
1:08:39 Shannon this is the AI learning lab I
1:08:40 talk about AI stuff here pop your
1:08:42 comments in the
1:08:45 um
1:08:45 what is this called the discussion
1:08:48 and uh I will do my best to answer them
1:08:51 and
1:08:52 um if you haven't played with uh chat
1:08:54 GPT go play with it those are some URLs
1:08:57 right there the top one's the official
1:08:58 chat GPD site
1:09:01 um have I experimented with small I
1:09:03 don't know what small is let's go let's
1:09:05 go learn that together here's these cool
1:09:07 Star Wars inspired chairs I love that
1:09:10 one this one
1:09:13 in the upper right is so good oh that's
1:09:15 the other thing you can do on Discord
1:09:18 um let me go back to Discord for a
1:09:20 second
1:09:21 I have an 80d moment here
1:09:23 um
1:09:24 so so once you generate things in
1:09:27 mid-journey
1:09:28 um you see these there's U one two three
1:09:32 and four and V one two three and four so
1:09:34 the the the
1:09:36 um images are numbered like one two
1:09:39 three four like you know across and then
1:09:41 down and the Visa variations so what I
1:09:44 can do is I can hit V2 and it'll give me
1:09:46 four variations of this this one in the
1:09:50 upper right hand corner
1:09:52 so we'll let it do that and then you can
1:09:55 take
1:09:56 oh those are cool
1:09:58 let's let's let those build and then I
1:10:00 can take the one I like best out of
1:10:02 these and upscale it
1:10:04 and then I can go over to photoshop's
1:10:07 generative fill and put it in some crazy
1:10:10 room
1:10:13 that's 55 they're getting there
1:10:18 this is a chair I would have in my house
1:10:20 that's really cool
1:10:24 um
1:10:25 83 percent we are watching water boil
1:10:30 we're watching water boil it will never
1:10:34 boil okay it boiled all right
1:10:37 um
1:10:38 which one of those is best they're all
1:10:40 kind of the same
1:10:42 not really a lot of variation there I
1:10:44 think
1:10:45 I think the upper right one's the best
1:10:47 so so now what I'll do is I'll hit U2 so
1:10:50 that means upscale too upscale image two
1:10:53 and now this should go pretty quick
1:10:57 it'll generate
1:11:01 um
1:11:07 a higher risk version of that
1:11:09 so that is cool
1:11:13 tap the screen if you'd buy that isn't
1:11:15 that bitching
1:11:17 all right what was it doing have I
1:11:19 experimented with small I was going to
1:11:20 go figure out what small is
1:11:22 um s m o l a i
1:11:29 build entire code bases with a single
1:11:32 prompt oh okay so this is a this is a
1:11:34 text to
1:11:35 text to code thing now I haven't played
1:11:37 with it that looks really interesting
1:11:38 though
1:11:43 um
1:11:47 all right sometimes I wonder if my brain
1:11:49 could work better with a little HD ADHD
1:11:52 I mean ADHD is a blast it's just that
1:11:55 our
1:11:57 prefrontal cortex is essentially
1:11:59 non-existent so like it seems like we're
1:12:02 gonna get [ __ ] done and we're really
1:12:04 capable about getting stuff started but
1:12:06 you know
1:12:07 then
1:12:09 [ __ ] goes south
1:12:11 that's why when I started my business my
1:12:13 current business storyline I I chose a
1:12:16 co-founder that is
1:12:18 um really good at that [ __ ] so figure
1:12:21 out what your weaknesses are and rather
1:12:23 than trying to you know fix the [ __ ]
1:12:25 that's not fixable backfill with people
1:12:28 that are better at that [ __ ] than you
1:12:29 are what resources to learn more
1:12:32 um
1:12:33 well prompts.chat I put that there that
1:12:36 that'll help you with the prompting
1:12:39 um
1:12:40 I don't know that there's any really
1:12:42 good resources right now
1:12:44 um
1:12:45 um Rachel Woods Tick Tock Channel called
1:12:47 the AI Exchange
1:12:49 is good she's got a newsletter and she's
1:12:52 got a lot of resources out there
1:12:54 um
1:12:55 I'd say you know
1:12:58 start following people on Tick Tock that
1:13:00 are that are you know putting stuff out
1:13:02 like if you're if you're an artist start
1:13:04 following artists if you are into coding
1:13:06 start following coders right right so
1:13:09 there's not really good resources right
1:13:11 now here's the problem everything is
1:13:13 changing so fast there's a couple of
1:13:15 things I would I would guard against
1:13:17 anyone saying yeah
1:13:20 ignore them
1:13:22 and anyone that claims to be an expert
1:13:24 like this channel is about you know me
1:13:27 acting like an expert but not an expert
1:13:29 like I'm figuring this [ __ ] out and
1:13:31 trying to share what I'm learning along
1:13:33 the way because it's changing so fast
1:13:35 like things that were remarkable three
1:13:37 months ago are like either gone or have
1:13:40 been completely absorbed into some other
1:13:42 product it's just like that's how it's
1:13:44 going right now no one knows what
1:13:46 they're doing and no one knows what the
1:13:49 actual impact on work is going to be
1:13:51 because right now if you think about
1:13:53 chat EBT it lives kind of at these
1:13:56 external URLs it's external from our
1:13:58 work life
1:13:59 but Microsoft and Google are both
1:14:02 incorporating these GPT models
1:14:05 completely within their office suites so
1:14:08 it's going to be in Microsoft Word it's
1:14:10 going to be in Excel it's going to be in
1:14:12 PowerPoint it's going to be in
1:14:14 everything it's going to be in Windows
1:14:15 11 at an operating system level
1:14:18 um
1:14:19 what does that mean for work I don't
1:14:20 [ __ ] know no one knows right the
1:14:22 implications of all this stuff are like
1:14:24 unfolding in real time
1:14:27 so I think the best resource is find
1:14:30 people that you resonate with and just
1:14:32 start following them and then
1:14:35 get curious like just keep digging in
1:14:37 keep doing projects where you
1:14:40 use these tools use the tools don't feel
1:14:43 like you got to read about the tools
1:14:45 just start using them start becoming the
1:14:47 expert because that will serve you well
1:14:48 in the employment game
1:14:50 because what's going to happen at some
1:14:52 point here is
1:14:54 the c-suite you know bosses at some
1:14:57 point are going to panic and go we need
1:14:59 [ __ ] someone who knows something
1:15:00 about AI
1:15:01 does anybody know anything about AI I
1:15:04 don't know Sally used chat GPT once make
1:15:06 Sally the VP of AI like that's gonna
1:15:09 happen that's gonna happen in the next
1:15:11 six months to a year so why don't you be
1:15:13 Sally
1:15:15 Sally's awesome and she knows chat gbt
1:15:21 all right maybe try a better camera
1:15:23 angle I know it's my neck and you're
1:15:26 seeing up my nose and I didn't trim my
1:15:27 nose hairs and I'm old and Jesus Christ
1:15:30 I so here's the problem with the higher
1:15:33 camera angle I'd have to I I'd have to
1:15:34 put my my computer significantly above
1:15:37 where I'm sitting so I could do that
1:15:39 wait maybe
1:15:41 here uh there you go that's a little
1:15:43 better
1:15:45 I don't know I don't you know what I
1:15:47 don't really give a [ __ ] that's not
1:15:48 comfortable
1:15:50 sorry you see up my nose
1:15:52 like I said my trolls are like it seems
1:15:55 like he meant well
1:15:58 what's your Discord tag Kyle Shannon
1:16:00 stop yelling
1:16:02 how do I use DaVinci Ai and mid-journey
1:16:05 to make KDP products I don't know what
1:16:07 KDP is
1:16:10 people assume I know what the [ __ ] I'm
1:16:12 doing I
1:16:13 KDP ugh
1:16:15 all right hang on let me go back to chat
1:16:18 EBT and ask it what the [ __ ] KDP
1:16:20 products are
1:16:22 what are
1:16:26 KDP products
1:16:33 oh
1:16:38 Kindle Direct publishing oh
1:16:41 ebooks well
1:16:46 you
1:16:48 so I think a thing so so
1:16:52 I'm assuming you've already got the
1:16:53 written part figured out so if you're
1:16:55 asking about DaVinci and I and
1:16:57 mid-journey
1:16:59 um you've probably got the written part
1:17:01 figured out
1:17:03 so let's see
1:17:10 start getting good at mid-journey use
1:17:13 that
1:17:14 um that Pope thepo.com mid-journey bot
1:17:18 that I told you about
1:17:20 um so so like let's say
1:17:22 um
1:17:23 we're gonna
1:17:25 a new chat in the mid Journey bot and
1:17:27 I'm going to say
1:17:29 um children's book
1:17:33 illustration of
1:17:36 a
1:17:39 I don't know what it's of yet okay what
1:17:41 okay this is gonna be fun okay so we're
1:17:43 gonna go to chat gbt
1:17:46 we're going to do this right within
1:17:47 Claude so this is this is chat gbt
1:17:50 within Claude
1:17:52 and so I'm going to say
1:17:54 um
1:17:55 give me the characters of a
1:18:01 children's book
1:18:05 um
1:18:07 about animals
1:18:11 about let's see a children's book that
1:18:16 is a retelling of Star Wars
1:18:24 um but with
1:18:27 characters as Animals by animals on the
1:18:33 Galapagos golos but Galapagos
1:18:38 Islands
1:18:40 in the time of Darwin
1:18:44 where Darwin shows up on
1:18:50 his ship
1:18:53 please
1:18:56 name and describe the characters okay
1:19:04 all right let's see what it came up with
1:19:09 no oh that didn't do it good at all
1:19:11 Princess Leia Chewbacca didn't rename
1:19:13 them
1:19:14 all right I'm going to real chat GPT God
1:19:16 damn every time I every time I use
1:19:20 um something that's not chat gbt I get
1:19:23 disappointed
1:19:27 um
1:19:28 yeah
1:19:32 [Music]
1:19:35 uh let's see let me go back to Poe and
1:19:37 so I don't have to type that again did I
1:19:39 lose my PO where did it go
1:19:42 oh what did I do
1:19:46 oh I see what I did okay
1:19:52 copy
1:19:56 let's see what real chat GPT does with
1:19:59 this
1:20:03 because I've done this this prompt
1:20:05 before and it came up with brilliant
1:20:06 [ __ ] and that just didn't even
1:20:08 understand the prompt
1:20:10 sure it sounds like a delightful premise
1:20:12 from for a children's book here are
1:20:15 characters from Star Wars Galapagos
1:20:17 Edition
1:20:18 Luke skyhopper now we're getting there
1:20:21 a young and adventurous Galapagos
1:20:24 tortoise wait he's skyhopper and he's a
1:20:26 tortoise
1:20:27 that doesn't make any sense all right we
1:20:29 so we should fix that
1:20:32 um
1:20:32 Princess Leia
1:20:35 Albatros a graceful waved Albatross okay
1:20:39 so so now I've got a description of
1:20:45 uh uh
1:20:48 I've got sort of a description of that
1:20:51 thing so now I'm going to go back to
1:20:52 that Poe mid-journey bot
1:20:56 and I'm gonna put in
1:21:00 um
1:21:03 a children's book
1:21:07 illustration
1:21:10 of and then I'm just going to paste in
1:21:12 that description a graceful
1:21:16 this is for a
1:21:20 Star Wars themed
1:21:24 book of animals let's see what it comes
1:21:28 up with well it came up with something I
1:21:30 don't even care what it came up with it
1:21:31 just like rewrote the prompt so I'm
1:21:33 literally just going to copy and paste
1:21:34 this
1:21:37 and I'm going to go over to mid-journey
1:21:39 and we're going to do slash imagine
1:21:44 slash imagined
1:21:50 and it's gonna do its thing so
1:21:54 um
1:21:55 so just start experimenting and and
1:21:58 start learning about prompt crafting
1:22:00 um a resource that I'll send you to
1:22:05 um
1:22:06 that is super valuable is called the
1:22:09 open
1:22:11 openart.ai if you go to open art
1:22:14 dot a I slash prompt book
1:22:19 this is for stable diffusion but it's
1:22:22 it's
1:22:23 good for mid-journey as well because
1:22:26 it's what this is is a 96 page
1:22:30 presentation that describes all of the
1:22:33 nuances of what these image generation
1:22:36 models like there's an example of how to
1:22:39 write a prompt what the what the
1:22:40 components of a prompt are
1:22:42 but then here are modifiers like it
1:22:45 understands things like photography
1:22:47 terms like there's a whole bunch of
1:22:49 Photography terms it understands and
1:22:51 then things like tilt shift and product
1:22:53 shot and long exposure and color splash
1:22:56 right like with the the red phone booth
1:23:00 in a black and white photo so
1:23:03 this
1:23:06 um
1:23:06 this resource like so so you need to get
1:23:10 your head around how like how capable
1:23:12 these tools are
1:23:14 let's go back to Discord and see if it
1:23:16 did anything well it didn't really do
1:23:18 them as
1:23:21 um
1:23:22 oh because it said photographers
1:23:26 so it so it gave me it gave me examples
1:23:29 of like cinematographers and
1:23:30 photographers and things like that so it
1:23:33 made it photo realistic so what I'm
1:23:35 going to do is I'm going to pull out
1:23:38 um
1:23:41 I'm going to pull out those artists and
1:23:43 I'm going to replace it with something
1:23:44 so I'm going to go slash imagine
1:23:52 I'm going to paste it all in I'm going
1:23:53 to pull out all of these photographers
1:23:55 and fashion designers and film directors
1:23:57 and I'm going to go
1:24:00 um
1:24:01 let's just do I'll be lazy Pixar
1:24:05 um Style
1:24:08 of
1:24:09 um illustration
1:24:11 [Music]
1:24:13 and then we'll see what that does so
1:24:15 that should give us something that's
1:24:16 more
1:24:18 that's less like like it it didn't even
1:24:20 take in any of the
1:24:24 um the Star Wars stuff so yeah that was
1:24:26 crappy but anyway just start start
1:24:28 learning the tools start start learning
1:24:30 how to modify the tools
1:24:32 thank you master of my fortune for the
1:24:34 kind words
1:24:36 um will these ever become pay well no I
1:24:39 don't think so because it's it's in
1:24:41 Microsoft's interest to to so here's
1:24:44 something I learned for every one
1:24:46 percent market share that Microsoft
1:24:48 pulls away from Google it's two billion
1:24:51 dollars in Revenue to their bottom line
1:24:53 just for the search
1:24:56 um
1:24:58 and also
1:25:00 um the CEO of Microsoft is taking great
1:25:02 pleasure in
1:25:04 um [ __ ] with Google right now he said
1:25:06 he's willing to lose money to to to
1:25:08 screw with them which I think is just a
1:25:10 baller move
1:25:13 the other thing about Microsoft is that
1:25:16 they are rolling these tools across all
1:25:20 of their products
1:25:22 did it I don't know that didn't work
1:25:25 that was that was a crappy prompt I
1:25:27 didn't take enough time to look at that
1:25:28 prompt so so that was an absolute fail
1:25:30 but basically Let's do let's do this
1:25:33 um slash imagine
1:25:36 um
1:25:38 um
1:25:38 children's watercolor
1:25:45 illustration of a
1:25:48 albatross
1:25:52 dressed as
1:25:54 Princess Leia
1:25:59 I don't know how to spell that
1:26:01 um
1:26:02 on the rocks by the ocean
1:26:09 dash dash
1:26:11 ar169 we'll make this a 16x9 thing and
1:26:14 we'll go Dash Dash
1:26:17 um style I think it's style is it right
1:26:20 dash dash style
1:26:23 um 750
1:26:25 and let's see what that does
1:26:29 um
1:26:30 so back to will it be paywalled so so
1:26:32 it's in Microsoft's interest not to have
1:26:35 these things cost money so they're
1:26:38 they're rolling it into all of office
1:26:39 365. they're rolling it into everything
1:26:42 um
1:26:43 they'll make money on selling
1:26:45 subscriptions to that so they can get
1:26:47 they can use it as a loss leader to get
1:26:49 people to use their [ __ ] Google's going
1:26:52 to make this for free just like they
1:26:54 make everything for free because why is
1:26:56 it free because you're the product right
1:26:58 they're selling your data
1:27:00 um so what better way to gather data
1:27:03 than have people generating and you know
1:27:06 creating images and you know documents
1:27:09 and things like that so so it
1:27:11 dramatically increases their data data
1:27:13 capabilities and then you've got all
1:27:16 these open source models that are out
1:27:18 there
1:27:19 um that those are always going to be
1:27:21 some version of free right there might
1:27:22 be products built out of them but I
1:27:24 think I think these are always going to
1:27:26 be there
1:27:28 oh
1:27:29 hang on
1:27:33 I messed up with my I made a syntax
1:27:36 error
1:27:38 with my
1:27:39 I can't do style 750 with version 5.1
1:27:43 which I'm using so whatever
1:27:45 all right
1:27:47 now it's going
1:27:48 um so I think it'll I think I think it
1:27:50 will be free for for a while
1:27:53 I want to start a cat walking business
1:27:56 I'll use cat GPT Ah that's dad comedy
1:27:59 except that's from a woman big [ __ ]
1:28:01 420. she's stunned so so it's uh is Dad
1:28:07 humor with the with the stones I'm
1:28:09 guessing Mom that's pretty funny all
1:28:11 right
1:28:12 uh
1:28:13 uh why are you qualified to teach about
1:28:15 AI I'm not so
1:28:18 what what I'm qualified to talk about is
1:28:21 how we'll use AI in business
1:28:24 um I was an innovator in the early days
1:28:26 of the web I didn't invent any of the
1:28:28 web [ __ ] but I started one of the first
1:28:30 digital agencies when I discovered what
1:28:32 the web could do
1:28:35 um and
1:28:36 I've been an entrepreneur for 30 years
1:28:38 and I've been kind of at the Forefront
1:28:41 of you know taking technology and
1:28:44 figuring out ways to use that in
1:28:47 business so what my qualification is is
1:28:50 in being just a little bit ahead of like
1:28:53 knowing what tools are out there
1:28:55 I started a newsletter I started an AI
1:28:59 Salon to surround myself with people
1:29:01 that do this and what I'm really
1:29:02 interested in is how do we use these
1:29:04 tools not how do we make these tools so
1:29:06 if you want someone who's expert in how
1:29:08 to make the tools and fine-tune the
1:29:10 tools and do all the data [ __ ] that's
1:29:12 not me if you want to talk to someone
1:29:15 that's been through this Rodeo
1:29:17 about a dozen times and
1:29:22 passionately believes that these tools
1:29:24 are
1:29:26 the most powerful things we've had in
1:29:28 our hands in the in the history of
1:29:30 humanity then that's why you talk to me
1:29:35 um but but I'm not really qualified to
1:29:37 talk about AI
1:29:38 just how you use it
1:29:43 I'm a proud digital holder I have a box
1:29:45 of side quests yeah I know see the the
1:29:48 uh the digital hoarding is uh
1:29:51 I like I I have no shame in it
1:29:54 but I do keep convincing myself that
1:29:56 like if I wanted to get the stuff off my
1:30:00 cyclist drive like I would need to go on
1:30:02 Amazon and buy like a a scuzzy to
1:30:05 something to USBC adapters you'd
1:30:08 probably have to buy like 17 different
1:30:09 adapters plus the old sidequest drive
1:30:12 you'd have to find one that worked plus
1:30:14 who knows if the data is even on that
1:30:16 thing anymore
1:30:19 um
1:30:19 I need a vacation or a purpose Nadia why
1:30:23 why are you attacking me so much I'm
1:30:25 just a simple man I'm just a simple man
1:30:28 sitting in a chair
1:30:30 gbt will have the same problem any
1:30:33 statistical model has inconsistent
1:30:35 results so many models okay uh you put
1:30:39 it on a cloudflare worker
1:30:41 um
1:30:42 what does your schedule look like where
1:30:44 are you finding time to find all the
1:30:46 stuff you are now because I'm obsessing
1:30:48 about this stuff 24 hours a day
1:30:51 [Music]
1:30:53 Peeves in life is when people say they
1:30:55 don't have time for something like you
1:30:57 don't have time for something if you
1:30:58 don't prioritize it and so I'm
1:31:01 prioritizing it so instead of watching
1:31:02 TV instead of doing anything on the
1:31:05 weekend
1:31:07 um I'm learning about [ __ ] I'm working
1:31:09 on the newsletter I'm doing these I'm
1:31:12 making videos I'm re-architecting my
1:31:15 business to be AI Centric
1:31:18 um I'm figuring out ways to sell the bit
1:31:21 like it's just it's all I'm thinking
1:31:22 about
1:31:23 you know this this and my business and
1:31:25 you know an occasional iced coffee
1:31:31 anyway there's some Princess Leia
1:31:33 Galapagos pictures they're still kind of
1:31:36 shitty but
1:31:39 hello I just accidentally
1:31:42 quit the app can you tap the screen if
1:31:44 I'm if you can still hear me because I
1:31:45 just accidentally quit the app
1:31:47 hahaha
1:31:49 [Applause]
1:31:51 when do you usually go live too much
1:31:53 apparently
1:31:54 according to Nadia this says said I says
1:31:57 I need a vacation all right uh it looks
1:31:59 like we've got some some new folks here
1:32:01 from before
1:32:03 so if you're new here my name is Kyle
1:32:05 Shannon this is the AI learning lab I am
1:32:07 doing what I can to
1:32:10 um help people figure this stuff out if
1:32:12 you've got questions about AI pop them
1:32:13 down below try to keep the trolling to a
1:32:15 minimum minimum I know I look like Rosie
1:32:18 O'Donnell but I can't help it
1:32:20 um if you haven't used chat GPT before
1:32:22 that top URL that's the official chat
1:32:25 gbt website the one below it bing.com is
1:32:28 Microsoft's version of chat GPT click on
1:32:31 the chat button that's gpt4 connected to
1:32:34 the internet for free
1:32:35 and then pro.com has a bunch of
1:32:37 different things you can play with and
1:32:38 then these two things below prompts.chat
1:32:41 will teach you how to prompt if you've
1:32:42 never done it before
1:32:44 um but you can also just go play Go Tell
1:32:46 it to uh
1:32:47 describe your significant other or a pet
1:32:49 and tell it to write a country song
1:32:51 about that person and just sit back and
1:32:53 watch what it does and then go like it
1:32:55 can do that yeah and then tell it to
1:32:58 write you some computer code and then
1:33:00 you'd be like it can do that yeah and
1:33:03 then have it write you a table comparing
1:33:04 two concepts and then you go you can do
1:33:07 that and then at that point you'll be
1:33:09 hooked and it'll be like crack and
1:33:10 you'll be like me and kind of sweaty on
1:33:12 a tick tock live when you're 58 years
1:33:14 old
1:33:15 all right on a personal level AI can
1:33:17 change a person's entire life for the
1:33:19 better it can like I I look at I look at
1:33:21 chat GPT and these tools is like
1:33:24 Humanity amplifiers like you can take
1:33:29 what you want to do you can take a
1:33:31 whisper of an idea that historically
1:33:33 would have taken you a week to get into
1:33:35 any sort of format that you could even
1:33:36 [ __ ] talk about it
1:33:38 and it can go from that Whisper of an
1:33:40 idea to a fully articulated like story
1:33:43 outline or or first draft of a paper or
1:33:47 first draft of a poem or whatever the
1:33:49 [ __ ] it is in like
1:33:51 five ten minutes tops
1:33:54 and then all of that time that you would
1:33:57 have taken blocking out the first draft
1:33:59 you can now do to execute
1:34:01 at an even higher level so I I think it
1:34:05 absolutely does become a tool to amplify
1:34:07 our Humanity originality is gone with AI
1:34:10 I one thousand percent disagree
1:34:14 here's the thing
1:34:16 this is not the robots doing the work
1:34:18 Microsoft has named their AI tools
1:34:21 co-pilot which feels like the right name
1:34:23 to me
1:34:25 humans are the pilot the tools are tools
1:34:28 that humans are using so the human is
1:34:30 the one with the intent right so it's
1:34:33 you putting in the prompts that get the
1:34:36 the image to do the thing you want it to
1:34:38 do or the chat to do the words you want
1:34:40 it to do you're then the one looking at
1:34:43 that and going is this a piece of [ __ ]
1:34:45 or not if it's a piece of [ __ ] keep
1:34:47 going if it's not then refine it make it
1:34:50 better right so the originality the
1:34:54 originality comes from the human and
1:34:56 their creative use of the tool if you
1:34:58 see true creatives who've really adopted
1:35:00 AI tools
1:35:02 they're operating at a totally [ __ ]
1:35:04 different level and it is as it like
1:35:07 it's more creative than anything I've
1:35:08 ever seen and I'm I'm living at a higher
1:35:11 level of creative execution
1:35:13 than I that I've probably ever been in
1:35:16 my life and I'm more creatively inspired
1:35:18 that I've been in decades so sorry 100
1:35:21 disagree on that one that that sounds
1:35:24 like you haven't used these tools a lot
1:35:27 I would encourage you go there and start
1:35:29 playing if you haven't
1:35:31 all right does anyone feel like he's not
1:35:33 reading our comments yes everyone says
1:35:35 that on every one of my lives because I
1:35:36 tend to stay live for like two hours
1:35:38 because as Nadia pointed out I need a
1:35:41 vacation in life
1:35:43 um
1:35:43 I try to read all the comments so as I'm
1:35:47 reading comments you're all commenting
1:35:49 down here and I'm way up there like it
1:35:52 says 99 plus new messages so I'm
1:35:55 probably 300 messages behind so yes I'm
1:35:58 reading your comments Jake
1:36:00 jakeatron
1:36:03 why is AI starting to make up lies false
1:36:06 data it's not starting to make up lies
1:36:09 and false data it doesn't know what it's
1:36:11 typing so when it types something that's
1:36:13 false it actually doesn't know it it's
1:36:16 the nature of how it works
1:36:18 the large language models are a
1:36:20 probability engine so when you type in
1:36:22 your prompt it's turning that into a
1:36:25 mathematical formula that then sort of
1:36:28 points it to this giant mass called the
1:36:31 latent space where there's all these
1:36:33 sort of semantic Canyons where words
1:36:35 live and within this giant mathematical
1:36:38 space it's saying based on this prompt
1:36:41 what is the next word I should respond
1:36:43 with and it types that in and then based
1:36:45 on all of that plus that word it said
1:36:47 what's the next word I should respond to
1:36:49 and it types that in so it's literally
1:36:51 typing word by word your responses
1:36:54 but it's not aware of the totality of
1:36:57 what it wrote so if you prompt it
1:37:02 um
1:37:03 if if you prompt it at too high high a
1:37:06 level it sort of pulls from all the data
1:37:08 and it will just amalgamate all the [ __ ]
1:37:11 that it knows and it will just make [ __ ]
1:37:13 up
1:37:14 if you get better at prompting and get
1:37:16 it to sort of dig deeper into these
1:37:18 semantic Canyons
1:37:20 um then it you can get it to hallucinate
1:37:22 less but it still will hallucinate
1:37:25 um there's a pretty famous uh this past
1:37:28 week a lawyer submitted a legal brief to
1:37:31 a judge written by chat gbt and he
1:37:34 didn't check the citations and they were
1:37:36 all made up and wrong
1:37:38 that's not Chachi PT's fault that's his
1:37:41 fault for not proofreading the stuff
1:37:42 he's a lawyer
1:37:45 um AI makes stuff up it does I have a
1:37:48 problem
1:37:49 with
1:37:50 consensus creation
1:37:53 I don't know what consensus creation is
1:37:55 I don't think
1:37:56 I don't think chat CPT is consensus
1:37:58 creation but it may be
1:38:01 what is your Discord tag I've told you
1:38:03 many times hello again been doing some
1:38:07 research in cheap gpus for training what
1:38:10 would you recommend I don't know
1:38:12 I'm a Mac guy so
1:38:15 um I'm kind of [ __ ] out of luck I think
1:38:18 the new M2 chips are gonna have enough
1:38:21 GPU horsepower to do some interesting
1:38:24 stuff but you know a like a uh
1:38:28 Nvidia 3090 or higher right and they're
1:38:31 probably impossible to get why because
1:38:32 everyone's training their [ __ ] on
1:38:34 Nvidia uh gpus us lawyers sorry that
1:38:39 sorry after
1:38:41 wait U.S lawyer sorry after chat GPT
1:38:44 creates bogus cases he should be sorry
1:38:46 he didn't [ __ ] check his work like of
1:38:50 all the professions to let
1:38:53 that [ __ ] mistake happen a lawyer
1:38:56 should be the last in line
1:38:58 when are your lives are they scheduled
1:39:00 no I have ADD I'll do I'll do two a day
1:39:04 for the next week or two and then I
1:39:06 don't know disappear for a while if I
1:39:07 get bored I don't it's just the nature
1:39:10 of like I I don't know I try to do them
1:39:13 once a day General generally in the
1:39:15 evenings
1:39:17 but weekends who knows it depends on how
1:39:20 big the honeydew list is
1:39:22 please don't bring religion in here
1:39:24 thank you when will Alexa incorporate AI
1:39:26 well Alexa's already Incorporated AI I
1:39:29 assume you're asking when will it
1:39:30 incorporate AI that doesn't suck
1:39:34 um I don't know I don't know I it would
1:39:38 be so
1:39:39 Amazon has announced that they've got
1:39:41 their own large language models and
1:39:43 they're going to support all these other
1:39:44 large language models and
1:39:46 throughout AWS you're going to be used
1:39:48 able to use all these tools but they
1:39:50 haven't released anything yet so who
1:39:52 knows I don't know like the the the
1:39:55 challenge with all for all of these
1:39:57 companies that are not open Ai and
1:39:59 Microsoft
1:40:00 is that no one including Microsoft or
1:40:03 open AI expected chat GPT to be the Mega
1:40:08 hit that it is
1:40:10 and so
1:40:12 so like it got to 100 million users in
1:40:15 six weeks the fastest adoption of
1:40:17 technology in history
1:40:19 um
1:40:21 because of that
1:40:24 all of the companies from meta to Google
1:40:26 to Amazon to just everyone
1:40:30 is going we've got AI too we've been
1:40:32 doing AI we've got models they're all
1:40:35 panicking but they're way behind because
1:40:38 um open AI did it really well and
1:40:42 Microsoft was smart enough to recognize
1:40:44 it and said [ __ ] it we'll put in 10
1:40:45 billion dollars and roll it into every
1:40:48 product that we've got and make this our
1:40:50 big bet for the next 20 years
1:40:52 so
1:40:54 um Alexa will incorporate it when they
1:40:56 probably get it to the point that it
1:40:58 doesn't suck
1:41:00 um and and it's decently tested
1:41:07 it
1:41:10 creates lies period even apologizes for
1:41:13 them yes it does
1:41:15 so so there's something in the nature of
1:41:17 how you're saying that so what he's
1:41:20 saying here is chat gbt lies period it
1:41:23 even apologizes for them yes it lies it
1:41:26 confidently lies I I call it
1:41:27 mansplaining as a service it confidently
1:41:30 lies because it doesn't actually know
1:41:32 what it's typing it's not conscious of
1:41:33 what it wrote
1:41:35 um
1:41:37 I would caution you not to stand on a
1:41:41 soapbox and say that thing's not perfect
1:41:43 and dismiss it
1:41:45 because people are working on all sorts
1:41:49 of systems that are going to fix that
1:41:51 particular the hallucination piece
1:41:54 um
1:41:56 chat EPT launched in November of 2022.
1:41:59 this is six month old technology the the
1:42:01 stuff that we're playing with right now
1:42:03 it's very very new and very very raw
1:42:06 it's also the most powerful tool
1:42:09 that we've ever had our hands on right
1:42:11 so don't collapse the fact that it isn't
1:42:15 perfect right now for how [ __ ]
1:42:17 remarkable it is because it the impact
1:42:20 that this is going to have on our lives
1:42:22 and our businesses is going to be quite
1:42:23 profound and I don't think people are
1:42:24 prepared for how fast it's coming
1:42:27 and and and statements like that
1:42:31 are frustrating because
1:42:34 it's coming don't ignore it because it's
1:42:36 not perfect
1:42:37 what is chat GPT plus why is it better
1:42:39 okay so chat GPT plus is
1:42:44 basically you get access to
1:42:47 well you get a little yellow dollars but
1:42:49 what it gives you is it gives you access
1:42:52 to gpt4 and then if you enable the beta
1:42:55 features it gives you gpt4 connected to
1:42:59 the internet via Bing and it also gives
1:43:02 you access to plugins which I just went
1:43:04 and looked
1:43:05 three days ago there were 250 plug-ins
1:43:07 today there's 400. so there's 400
1:43:10 plugins that amplify gpt4 even more
1:43:15 um so that's that's what it gives you it
1:43:17 also is going to give you access to new
1:43:18 tools as open AI rolls them out
1:43:23 um yeah that's about it
1:43:25 serious question how is chat GPT AI it's
1:43:29 not really it's it's machine learning it
1:43:30 just find finds patterns and repeats
1:43:32 them that's correct so so listen here
1:43:36 you know again
1:43:39 it's kind of irrelevant what you call it
1:43:41 like we're not gonna AI is easier to say
1:43:45 than machine learning and machine
1:43:48 learning sounds like oh the machines are
1:43:51 gonna learn I gotta learn machines I'm
1:43:53 not gonna do that but AI is easy to like
1:43:56 just from a marketing standpoint AI is
1:43:57 easier so they're going to call it AI is
1:44:00 it technically Ai No it's not
1:44:01 technically AI
1:44:03 um hey buddy
1:44:05 hey buddy my little doggy just came in
1:44:07 champ hey champ you're gonna bark for
1:44:10 everybody
1:44:12 um
1:44:14 and yes it is just a prediction engine
1:44:16 but it is a prediction engine that is
1:44:21 [ __ ] insanely remarkably powerful so
1:44:25 from a technical perspective you're
1:44:27 right but again and this gets back to
1:44:30 the question of you know whoever asked
1:44:31 the question about what do you know
1:44:32 about AI what's your expertise what
1:44:34 makes you qualified
1:44:36 um I'm just like an entrepreneur
1:44:38 entrepreneur like business guy trying to
1:44:41 figure out where the power of these
1:44:43 tools is where it's not and how it might
1:44:46 be used in my life and in my business
1:44:48 and
1:44:50 the this [ __ ] is remarkable it's
1:44:54 remarkable when you watch it
1:44:57 do what it does and and you're trying to
1:45:00 solve a problem and it does in 30
1:45:02 seconds what would have taken you three
1:45:05 days
1:45:06 it's just
1:45:08 we've never had tools like this in our
1:45:10 hands so that that's like
1:45:14 I would I would find a way to separate
1:45:18 um
1:45:21 the technical distinctions that that
1:45:23 you're trying to make from the fact that
1:45:26 this stuff is happening and it's
1:45:27 happening fast those are two distinct
1:45:29 things
1:45:30 we can come up with definitions of what
1:45:32 it is and things like that for more
1:45:33 accuracy but in the end it's kind of
1:45:35 irrelevant because these tools are
1:45:37 making a difference less like Alec
1:45:40 Baldwin more like Jack Nicholson I'll
1:45:41 take either one
1:45:42 that's cool who is your money on
1:45:46 who wins the personal assistant race
1:45:49 that Bill Gates talked about
1:45:51 um Apple
1:45:53 I think Apple wins the autonomous agent
1:45:56 race
1:45:57 for a lot of reasons they've got privacy
1:46:00 and authenticity and accuracy are going
1:46:04 to become really important after the
1:46:06 2024 election I think we're about to see
1:46:09 a level of disinformation and
1:46:10 misinformation that that's going to make
1:46:12 2016 and 2020 seem quaint
1:46:15 and so I think there's going to be a
1:46:17 hypersensitivity to privacy and
1:46:20 authenticity and accuracy and I think
1:46:22 Apple's the best position for that
1:46:25 they're machine learning chips like they
1:46:28 can do a lot of their machine learning
1:46:29 stuff locally on devices without having
1:46:32 it to send it out which from from a
1:46:34 privacy and security standpoint is
1:46:36 better
1:46:38 um
1:46:39 they're just they're really good with
1:46:40 interface and the human factors piece of
1:46:43 this they're good at mixing
1:46:45 you know humanity and Technology
1:46:48 um
1:46:50 you know I'd be really surprised if they
1:46:52 weren't
1:46:53 you know kind of the preeminent player
1:46:55 in the autonomous agent game but we'll
1:46:57 see
1:46:58 I don't know that's you know that's
1:47:01 again something
1:47:03 anything for front-end Dev ux yeah have
1:47:06 you seen have you like check this out so
1:47:09 I'll go do one of these on I forget how
1:47:11 to do it let me let me go find a hold
1:47:13 please
1:47:14 let me go find a prompt
1:47:19 so so here here's so I'm gonna go I'm
1:47:22 just gonna go to like
1:47:23 um a Google search or Bing search in
1:47:25 this case
1:47:27 and I'm gonna go
1:47:29 um
1:47:30 uh prompt
1:47:33 for Mid
1:47:37 Journey ux
1:47:41 40 plus prompts to create outstanding ux
1:47:45 design all right perfect let's go grab
1:47:47 those
1:47:51 oh this might be better than me showing
1:47:54 it live so so ux designers are using mid
1:47:57 journey to do all sorts of crazy ux
1:48:01 um ux like there's a site there's like a
1:48:05 an iPad kind of app
1:48:08 what's that there's a wine
1:48:11 like an e-commerce store there's a
1:48:12 motorcycle store so what site is this
1:48:15 this is boot camp slash what is this
1:48:18 sorry
1:48:19 oh it's a medium article 40 plus
1:48:21 mid-journey
1:48:23 mid-journey prompts to create
1:48:25 outstanding ux design so I just I just
1:48:27 searched for
1:48:30 um what did I search for
1:48:32 prompt for mid-journey ux just go search
1:48:34 for that um it like
1:48:37 astounding astounding
1:48:39 um
1:48:40 really Innovative designs now they're
1:48:43 not
1:48:44 um
1:48:45 they're just images right now so so
1:48:48 you'd have to split them out and kind of
1:48:49 recreate them so what ux designers is
1:48:51 they're using mid-journey for ux
1:48:53 concepting and then they're popping it
1:48:55 into their you know figmas and their
1:48:57 other design tools to break it apart
1:48:59 um there are probably the reason I put
1:49:01 future pedia up here there are probably
1:49:04 ux tools now that are actually text to
1:49:07 figma designs that I'm I'm
1:49:10 really confident that exists that you
1:49:13 know and and that ability to do text go
1:49:16 make me an app that does this and looks
1:49:18 like this that's coming so there's
1:49:19 probably a bunch of tools at
1:49:21 futurepedia.io that can do that but I
1:49:23 would dig in dig in and figure that [ __ ]
1:49:27 out because that's going to be powerful
1:49:29 I think we're going to just see an
1:49:31 absolute Renaissance of really
1:49:32 interesting design from physical world
1:49:35 stuff to web design and everything in
1:49:38 between I think I got mixed up with
1:49:39 mid-journey oh okay mid Journey has a 10
1:49:43 or 20 and a sixty dollar subscription I
1:49:46 paid the ten dollar one
1:49:48 um
1:49:49 I don't know that I think the 20 one
1:49:51 just gives you more hours so if you find
1:49:53 yourself like running out of credits
1:49:56 then I think that's why you'd go 20 and
1:49:58 the reason to go 60 is you can do
1:50:00 private prompting so if you're doing
1:50:01 stuff where you don't want anyone to see
1:50:04 your you know you're doing kind of
1:50:05 proprietary [ __ ] you don't want anyone
1:50:06 to see or be able to discover I think
1:50:08 that's the 60 one
1:50:11 um I'm not sure I just paid the ten
1:50:12 dollar one I'm not doing enough on it to
1:50:14 really push the boundaries of anything
1:50:17 mm-hmm
1:50:20 love your content greetings from Cape
1:50:22 Town South Africa thank you very much
1:50:25 do you have a purpose
1:50:27 Jesus Nadia
1:50:31 yeah I'm trying to demystify generative
1:50:34 AI for people who are trying to figure
1:50:36 out what the [ __ ] is going on in the
1:50:38 world that's my purpose right now I also
1:50:40 have a purpose to democratize video
1:50:42 storytelling I also have a purpose to
1:50:44 try to do cool [ __ ] in my life and do
1:50:47 stuff that I'm passionate about
1:50:48 and create community and I don't know
1:50:51 yeah I don't what
1:50:53 does it seem like I don't have a purpose
1:50:55 obviously all right what will AI look
1:50:57 like in 10 years I have no [ __ ] clue
1:50:59 what what I do know is this I don't know
1:51:02 what AI will look like but I have a
1:51:03 sense of what
1:51:07 foreign
1:51:08 what will be the
1:51:13 one of the big change that's that's
1:51:15 precipitated is that
1:51:17 I think that AI is going to make all of
1:51:20 the [ __ ] that's really complicated in
1:51:23 our world
1:51:25 easy
1:51:27 so if you think about
1:51:30 um
1:51:32 if you buy a car or you buy a house or
1:51:36 you try to get your insurance your
1:51:38 health insurance company to pay their
1:51:39 [ __ ] bills
1:51:42 those forms that they have you fill out
1:51:44 that are like 27 pages of you know
1:51:47 probing up your ass and figuring all
1:51:49 that stuff out they're purposefully
1:51:51 complicated there are whole Industries
1:51:54 built around these ever increasing
1:51:57 complicated systems insurance companies
1:51:59 in particular
1:52:02 make it as complicated as possible
1:52:05 so that you can't figure it out because
1:52:07 if you can't figure it out or you get
1:52:10 frustrated
1:52:11 or or you have ADD and you don't feel
1:52:14 like sitting there for four [ __ ]
1:52:16 hours filling out forms then you won't
1:52:18 submit a claim and they make more money
1:52:20 AI solves that [ __ ] it solves it right
1:52:24 is it solved right now no but you can
1:52:27 take incredibly complicated documents
1:52:30 and just go tell me what this is about
1:52:31 wait there it is
1:52:34 so the same thing's going to be take
1:52:36 this incredibly complicated insurance
1:52:38 policy read it for me
1:52:41 take this rejection letter and this set
1:52:44 of lab work that was done and go get me
1:52:47 my money
1:52:49 and off it will go so I so so like I
1:52:53 don't even know what the implications of
1:52:54 that are
1:52:55 but but
1:52:57 all of the industries that are built
1:52:59 around the complexity of today's world
1:53:02 are going to be completely disrupted I
1:53:04 don't know what they're going to be
1:53:04 replaced by
1:53:06 but I think that
1:53:08 nagging part of our world that is soul
1:53:12 crushing I think it goes away largely
1:53:16 to be fair VR is dope VR is dope listen
1:53:19 don't get me wrong well
1:53:22 no VR I've yet to see VR that
1:53:27 isn't like what um not six degrees
1:53:32 foreign
1:53:33 what was the what was the the VR thing
1:53:36 in like the early 2000s well whatever
1:53:37 that was the it's not it wasn't six
1:53:40 degrees I always mix it up with that I
1:53:42 don't know what it was but anyway like
1:53:44 they all look like that they like like
1:53:46 people walking around in VR as it
1:53:48 currently is is like awkward people
1:53:50 walking around at a Junior High dance
1:53:52 not not knowing what to do or you know
1:53:55 can't look at each other
1:53:57 um the [ __ ] that Apple showed is
1:53:58 actually different like like they're
1:54:00 using
1:54:02 you know
1:54:04 um sort of typical [ __ ] we do in our day
1:54:06 as the on-ramp to VR which I think is a
1:54:08 is a smart choice because if you look at
1:54:11 the AR 8 the VR worlds they just look
1:54:14 like
1:54:17 really bad video games from the late 90s
1:54:22 they do
1:54:23 speculate on what a feature or two of
1:54:26 chat GPT 10 would be
1:54:28 I think that
1:54:31 um I don't think it'll I had a [ __ ]
1:54:33 gpd10 I don't know like I'm trying to
1:54:35 figure out what gbd5 will be
1:54:38 I think I think it will be
1:54:43 I read a like a Twitter post or a blog
1:54:46 or something like that where where the
1:54:49 guy was talking about the concept of the
1:54:51 everything app
1:54:53 where you you'll just have a single app
1:54:56 and it will do everything
1:54:58 so if you wanna
1:55:00 take a photograph in your digital camera
1:55:02 and then you want to edit it you just
1:55:06 open the everything app and go edit that
1:55:09 photo to look like my street photography
1:55:12 that'll be done then you like turn that
1:55:15 into
1:55:17 um
1:55:18 you know 16 social media posts to go out
1:55:22 on my preferred social channels out
1:55:24 it'll go
1:55:26 um create a slideshow with my the the
1:55:28 latest 20 videos but done turn it into
1:55:32 an animated movie done
1:55:35 like write me a soundtrack that you know
1:55:39 um sounds like what the last 10 photos
1:55:42 look like done oh by the way pay all my
1:55:45 bills oh by the way rebook my trip to
1:55:47 Cancun to Madrid because I got a Bonus
1:55:49 like I just
1:55:52 I think we move into this place where
1:55:56 we'll all have this this personal
1:55:58 digital assistant that just [ __ ] does
1:56:02 it
1:56:03 like your job as a human will be to
1:56:05 figure out what you want to do
1:56:08 and you tell the thing and it will go do
1:56:12 it
1:56:13 and and and so what is that so like what
1:56:16 does our role become then I don't know
1:56:18 like I think our role could become
1:56:20 reconnecting with one another
1:56:23 um
1:56:24 like self-expressing at a higher level
1:56:27 contributing to Society at a higher
1:56:29 level because we all have these really
1:56:31 incredible tools supporting us
1:56:33 so then it just becomes about pure human
1:56:36 expression
1:56:39 I don't know
1:56:42 tell me a bedtime story about near
1:56:44 future AI Tech I just did
1:56:46 I I really do like like like
1:56:49 I know it's it's tempting and very easy
1:56:52 to go to the robots will kill us future
1:56:54 there's a plausible other future that
1:56:57 the robots will free us
1:57:00 they'll do the shitty work
1:57:03 imagine a world where the shitty work
1:57:05 just gets done
1:57:09 right and if you look at the humanoid
1:57:11 robots that musk and and Boston Dynamics
1:57:16 and there's like two or three other
1:57:17 companies and I think open AI just
1:57:20 didn't they just announce they've got
1:57:21 one someone else did
1:57:23 these humanoid robots are going to do
1:57:25 the shitty physical work like so we're
1:57:27 going to have humanoid robots doing the
1:57:28 shitty physical work we're gonna have
1:57:30 these GPT tools or whatever they become
1:57:32 in the future doing the shitty Technical
1:57:35 and knowledge work and then we're going
1:57:38 to be free to do what's left which is
1:57:41 what I don't know be human
1:57:44 take a walk
1:57:45 start a concert
1:57:47 make some art I don't know I don't know
1:57:51 I'm looking forward to Moriarty's first
1:57:53 victim on the whole deck that's good how
1:57:55 do you get chat CBT on a PC you go to
1:57:58 that URL just open your web browser and
1:58:00 go to chat.openai.com just go there
1:58:03 that's why it's here I don't get paid
1:58:06 from them that's just there as kind of a
1:58:08 public service announcement and just go
1:58:10 play with that other [ __ ]
1:58:13 oh boy I don't know what I said fully
1:58:15 obsessed with AI now thanks a lot Kyle
1:58:17 yeah sorry about that Anne what are you
1:58:19 gonna do uh anything with front-end Dev
1:58:21 ux I think I just showed that dividing
1:58:23 by zero
1:58:26 um I Heard Google has a generative AI
1:58:28 course available now yes and open AI
1:58:32 just launched one on prompting
1:58:35 um yeah go learn that [ __ ] anything like
1:58:38 whatever's your preferred way of
1:58:40 learning like mine is like build [ __ ]
1:58:42 get Hands-On if you learn better through
1:58:44 courses and stuff like that take it from
1:58:45 the horse's mouth like Google invented
1:58:48 this [ __ ] Googles will likely be more
1:58:50 technical
1:58:51 um open AIS is about getting better at
1:58:54 prompt writing and and they're writing
1:58:56 it so so open AI understands technically
1:58:59 how these things work and so they've got
1:59:01 some best practices for how to write
1:59:03 prompts so yeah I would go do that stuff
1:59:06 absolutely
1:59:08 anything you can do right now to get
1:59:10 smarter about what these tools are
1:59:13 what they're good at what they're not
1:59:14 good at what they make capable and where
1:59:16 they're going is going to serve you
1:59:18 incredibly well
1:59:19 I like mid-journey the most for images
1:59:21 but ffs
1:59:24 oh for [ __ ] sake the interface is
1:59:25 trash yes Karen
1:59:27 I
1:59:29 absolutely agree with you the fact that
1:59:31 you have to make images and [ __ ]
1:59:33 Discord is is just a disaster like this
1:59:38 interface is a disaster the fact that I
1:59:41 have to command line like we're creating
1:59:44 art images right and the fact that I
1:59:47 have to go slash imagine and then I have
1:59:49 to wait for that prompt to load and then
1:59:51 I have to load it and then I have to do
1:59:52 this thing and then I have to know codes
1:59:54 like dash dash AR stands for aspect
1:59:57 ratio 16 colon nine and I have to type
2:00:01 that [ __ ] in it's like really
2:00:05 um there's a like there's a tool out
2:00:07 there called leonardo.ai that's it sits
2:00:10 on top of stable diffusion not
2:00:11 mid-journey unfortunately
2:00:13 um but it but it
2:00:15 you know it's at least the beginning of
2:00:17 an interface where you can click on
2:00:18 stuff the other the other interface
2:00:19 that's really good right now is Adobe
2:00:21 Firefly
2:00:23 um
2:00:24 so so adobe's doing a couple of things
2:00:27 very right they they have trained their
2:00:29 model not on copy written images so and
2:00:34 they just launched this for Enterprise
2:00:36 so they're making this safe for
2:00:40 um they're making it safe for
2:00:42 corporations to use it but if we go text
2:00:45 to image on Adobe Firefly
2:00:49 it's not as good as mid Journey but
2:00:51 there's some cool stuff in here so oh
2:00:53 it's at the bottom okay so
2:00:55 so if I type in here
2:00:58 um let's see what did I say before I
2:00:59 said um
2:01:02 stylized
2:01:04 um
2:01:05 albatross
2:01:08 with
2:01:10 Princess Leia
2:01:16 hairdo
2:01:21 Hardo hair do
2:01:25 did I spell that right
2:01:28 um
2:01:29 standing on a rock so it'll create this
2:01:36 it's pretty quick for one thing
2:01:40 yeah see that's getting there it doesn't
2:01:42 quite understand what Princess Leia's
2:01:44 hair is I I think I spelled Leia wrong
2:01:46 so it just knows princess so that that
2:01:48 one in the upper right starting to get
2:01:49 there but anyway it's got like it's got
2:01:55 um content type photo so if I just click
2:01:57 photo it'll kind of redo those things as
2:02:00 photos
2:02:03 and then if I click graphic it'll redo
2:02:06 them as graphic and then I can and
2:02:08 there's different styles and color tones
2:02:10 and things like that but you just click
2:02:12 on it right so so it did those as
2:02:14 Graphics so you just click on it so you
2:02:16 don't have to type all that [ __ ] in so
2:02:17 the interface for these things is going
2:02:19 to get better mid Journeys will get
2:02:21 better it's just that right now they're
2:02:22 putting their
2:02:23 their time and energy into
2:02:27 um
2:02:28 the image generation not the interface
2:02:31 right which is really frustrating
2:02:34 um I totally agree with you but the
2:02:36 results are [ __ ] mind-blowing
2:02:40 some comments in this live were filtered
2:02:42 don't be nasty to each other please
2:02:45 that makes me cranky when you people are
2:02:47 mean
2:02:49 thoughts on Google AI class I think you
2:02:52 should take it I haven't taken it yet I
2:02:53 don't I don't I don't I only know it
2:02:55 exists how do I use the plugins so I
2:02:58 think I've shared this before but if I
2:03:00 haven't showed it here
2:03:03 I'll go do it quickly so so the thing
2:03:07 you need to do is you need to be paying
2:03:09 for chat GPT Plus
2:03:11 and then you'll get access to them and
2:03:14 then you need to go to your settings
2:03:17 and
2:03:19 go to Beta features so the second choice
2:03:22 down and enable
2:03:24 browse with Bing and plugins so you got
2:03:27 to turn them on
2:03:28 and then when you go to gpt4 you've got
2:03:31 these three options default which is the
2:03:33 way it used to be
2:03:35 browse with Bing is connected to the
2:03:37 internet and then you've got plugins and
2:03:39 then
2:03:40 the plugins are you you've got a list of
2:03:44 whatever you have installed and then you
2:03:46 have check boxes for up to three of them
2:03:48 that you can actually activate for this
2:03:51 session so I might have a set of three
2:03:54 that I use for booking travel
2:03:55 arrangements and I might have two that I
2:03:57 use for reading PDF documents and I have
2:04:00 others that I use for financial [ __ ] so
2:04:03 so you kind of mix and match them
2:04:05 depending on the session right and then
2:04:08 to to find new plugins you go to the
2:04:12 plugin store and you can search and you
2:04:15 can scroll and there's now 400 of them
2:04:16 so there's a lot to go through
2:04:20 um and then basically what you do is
2:04:23 which ones do I have here so I've got
2:04:24 some travel ones
2:04:26 um activated so basically what you do is
2:04:28 I okay I know I've got some travel
2:04:30 things activated so I'll say plan
2:04:34 um a set of daily activities for
2:04:41 my exciting trip to Toledo Ohio next
2:04:47 week
2:04:49 um I'll
2:04:50 be there for three days
2:04:54 so rather than having to hop off to
2:04:56 Google and go find all this [ __ ] I could
2:04:58 just type in that one little sentence
2:05:01 and what it should do is just fire off
2:05:03 it'll figure out which of those plugins
2:05:06 is the appropriate one so it says using
2:05:09 get your guide
2:05:11 and then it says sure here's your
2:05:13 suggested activity
2:05:15 your self-guided driving tour through
2:05:17 Ohio that's cool
2:05:20 and there's a link
2:05:23 you can book your tour here
2:05:25 I will need to find more activities for
2:05:28 you please wait oh so now it's using get
2:05:30 your guide so it actually fired off
2:05:32 two other plugins that fired off get
2:05:34 your guide and it fired off experiences
2:05:36 so
2:05:39 so you you don't so much use plugins as
2:05:43 curate them right you go you go find
2:05:45 plugins that kind of do the thing you
2:05:48 want to do and then you activate them
2:05:50 for a session
2:05:52 and so now it's it's given me a whole
2:05:55 sort of itinerary here using plugins now
2:05:58 where I think these things are going in
2:06:01 the future is
2:06:03 um
2:06:05 [Music]
2:06:07 you won't have to pick them
2:06:09 I think where this is going is
2:06:13 um
2:06:14 you just tell it what you want it to do
2:06:16 and it will figure out do I need a plug
2:06:18 in here
2:06:20 um and if so great I'll do it and then I
2:06:23 think the future future version is do I
2:06:26 need to plug in here and then it'll be
2:06:28 like oh there isn't a plug-in let me go
2:06:31 see if there's any apis
2:06:33 that do this and it'll go find whatever
2:06:35 API to whatever data system and it'll
2:06:38 say okay let me write let me write my
2:06:40 own interface to that API now that I
2:06:42 understand it and if that doesn't exist
2:06:43 it'll say well let me create my own app
2:06:46 to do that thing like I think I I think
2:06:49 in a fairly short amount of time
2:06:51 probably within three years you just ask
2:06:55 this thing for [ __ ] and it gives you
2:06:57 [ __ ] how does it get there I don't know
2:07:00 it invents uses or remixes
2:07:04 any tools it needs needs combined with
2:07:07 any data it needs combined with the
2:07:09 large language model to organize that
2:07:11 [ __ ]
2:07:14 then makes pretty pictures and music and
2:07:16 videos as well
2:07:20 do I think I see it Security will be
2:07:22 impacted by AI
2:07:24 oh I think it's going to be upended
2:07:28 um
2:07:29 if you take this idea of autonomous
2:07:31 agents if you don't know what those are
2:07:33 it's basically like having
2:07:35 chat GPT prompt itself to go use tools
2:07:38 and go execute things to solve a goal
2:07:41 that you give it so you could give it
2:07:43 the goal
2:07:45 um go find me as many computers as you
2:07:48 can with this specific Windows
2:07:50 vulnerability
2:07:52 um exploit that vulnerability and give
2:07:54 me a list of hyperlinks of servers that
2:07:56 we have access to
2:08:00 I would be surprised if
2:08:02 people weren't playing with that right
2:08:04 now
2:08:05 right so yeah so I think it's going to
2:08:07 be impacted dramatically now
2:08:10 the it security teams also have access
2:08:13 to these AI tools so they're going to be
2:08:16 able to do countermeasures and and you
2:08:18 know be able to catch those people in
2:08:21 dramatically more powerful ways as well
2:08:22 so the war will continue between you
2:08:26 know black cat hacker hackers and I.T
2:08:29 security teams uh but it's going to get
2:08:32 a [ __ ] load more interesting okay please
2:08:34 help
2:08:35 best one for techs like chat gbt and got
2:08:38 image creation have a low budget
2:08:42 um
2:08:42 get on the wait list for leonardo.ai
2:08:46 they've they've they give you like 150
2:08:48 credits a day it's really good
2:08:50 um leonardo.ai like Leonardo da Vinci
2:08:54 um
2:08:55 and then chat GPT and if you want to use
2:08:58 gpt4 go to Bing bing.com is chatgpt4
2:09:02 without the
2:09:05 um cost
2:09:07 um
2:09:08 and then use that also right now I think
2:09:12 Firefly get on the Firefly beta from
2:09:16 Adobe
2:09:17 because I think there's going to be a
2:09:18 free version of that I'm not sure though
2:09:20 that's that's probably
2:09:22 I'm speaking out of my ass on that one
2:09:26 um there's also something that's like
2:09:28 mid-journey called Blue Willow so go to
2:09:30 Blue Willow is it bluewillow.com hang on
2:09:34 that is free but it's it's actually
2:09:36 based on stable diffusion not
2:09:37 mid-journey but it is free
2:09:39 um blue
2:09:41 Willow
2:09:43 blue willow.ai free AI art generator so
2:09:47 you go to bluewillow.ai
2:09:50 and then you say join the beta and it
2:09:53 will take you to a Discord server and
2:09:56 you'll be really confused if you don't
2:09:57 use Discord a lot and you're like wait
2:10:00 wasn't I on a website what is this thing
2:10:02 that's Discord that's the nightmare that
2:10:05 we were talking about before with
2:10:06 mid-journey just saw this in my Google
2:10:08 feed hello welcome
2:10:10 you're probably not here anymore because
2:10:11 I'm so far behind on my messages
2:10:15 um
2:10:17 all right can I talk through the videos
2:10:20 that I I jumped all the way to the
2:10:22 bottom
2:10:24 hey Kyle I like that name LOL AI for
2:10:27 retail and Logistics
2:10:29 um
2:10:30 yeah I don't know a ton about logistics
2:10:32 I think
2:10:35 I think AI is going to have a pretty
2:10:37 I don't know what's going on with retail
2:10:39 I don't know enough about that World
2:10:42 um what I would say is if you know a lot
2:10:43 about that world
2:10:44 I would start
2:10:47 um
2:10:48 I would just start exploring it with
2:10:50 chat gbt and with
2:10:52 um there's some plugins right now for
2:10:54 Google Sheets called uh I think it's
2:10:57 called GPT for Sheets is one of the
2:10:59 plugins that I know where you can
2:11:01 actually put calls to open AI into
2:11:04 individual cells within Google Sheets um
2:11:06 I'd started I'd start experimenting with
2:11:08 that because I think you could do some
2:11:09 insane stuff with the logistics side of
2:11:12 things
2:11:14 um I'll and I'll come back I'll do a
2:11:18 demo on how I made the the synthesized
2:11:20 video of myself I'll do that in a bit
2:11:22 let me just go back a few questions I'm
2:11:24 going to jump to the bottom of questions
2:11:25 here I'm going to skip a bunch so I can
2:11:27 get caught up because I think I was
2:11:28 about an hour behind on questions so so
2:11:31 I'm just gonna kind of randomly jump
2:11:34 back about I don't know 10 minutes or so
2:11:36 it's coming fast don't ignore it because
2:11:39 it's not perfect yeah exactly don't
2:11:41 ignore this [ __ ]
2:11:42 seriously this stuff is
2:11:46 here's here's the thing that
2:11:50 as remarkable as all this stuff is it's
2:11:53 as bad today as it's ever going to be it
2:11:55 only gets better from here right like it
2:11:57 only gets dramatically better from here
2:12:00 um
2:12:01 do you mean what you say
2:12:03 most times unless I'm lying
2:12:06 [Laughter]
2:12:08 then I don't
2:12:09 so we pay to improve the machines yeah
2:12:14 I mean listen this whole thing about it
2:12:17 should be free man
2:12:19 um
2:12:21 the amount of money that it costs to
2:12:23 train these models and the amount of
2:12:25 money that it every time you run a GPT
2:12:29 query
2:12:31 I think it's something like 10 cents or
2:12:34 like 20 cent like it's it's a [ __ ]
2:12:37 remarkable amount of money and like a
2:12:39 hundred million people are using this
2:12:41 like when you pay the 20 bucks a month
2:12:43 for chat EBT they're limiting limiting
2:12:45 it to 25 queries for every three hours
2:12:48 because it's [ __ ] expensive
2:12:51 um
2:12:52 so so yeah and they're using us to
2:12:54 improve their things just like Google
2:12:56 used us
2:12:58 to build their entire [ __ ] business
2:13:00 and Facebook used us right like we've
2:13:03 already given away all of our data right
2:13:06 we're already we've already paid
2:13:09 way more than money for all the free
2:13:11 tools so yeah we're paying to improve
2:13:14 the machines these are the most powerful
2:13:16 tools in the history of mankind
2:13:18 it's worth 20 bucks a month in my
2:13:21 opinion but you know
2:13:22 but there's free versions of it even if
2:13:25 you don't think it's worth 20 bucks a
2:13:26 month just go use Bing they got free
2:13:28 image generation and it's gpt4 for free
2:13:31 and it'll be free forever
2:13:33 all right I run black Arch Unix or Linux
2:13:37 cool I don't know what that is but it
2:13:39 sounds cool I don't even like artificial
2:13:41 and artificial intelligence the bird
2:13:44 nests or beaver dams
2:13:46 are bird nests or beaver dams artificial
2:13:49 yeah I agree I mean
2:13:53 like I don't think we're getting away
2:13:55 from AI like we've called it Ai and
2:13:56 movie for years movies for years and
2:13:59 it's just it's just easier to say
2:14:01 I agree with you
2:14:03 I tried for years in the in the early
2:14:05 days of when I was doing agency.com in
2:14:07 the early days of the web I actually got
2:14:10 a New York Times article written about a
2:14:12 concept I didn't think we should call
2:14:14 web pages Pages because when you call
2:14:16 them a page you metaphorically think of
2:14:19 it like a page and you create boring
2:14:21 flat pages
2:14:23 but if you think of it more like an
2:14:25 experience where you're entering a brand
2:14:27 like a theme park you'll create more
2:14:30 interesting experiences and in the end
2:14:32 people called it Pages because it looked
2:14:34 like a page and so that's why the
2:14:36 internet [ __ ] looks like the goddamn
2:14:38 Yellow Pages right now
2:14:40 like the early days of the web were so
2:14:42 inventive and so cool
2:14:45 huh
2:14:47 all right how is gpt4 priced using the
2:14:50 API of creating an app for public use
2:14:52 it's remarkably cheap it's expensive to
2:14:55 train and fine-tune like if you're gonna
2:14:57 if you're gonna create embeddings based
2:14:59 on your data that's pretty expensive but
2:15:02 just calls to like The DaVinci 3 Model
2:15:04 is is really cheap and then
2:15:08 um
2:15:09 I don't have access to the gpt4 API yet
2:15:12 so I don't know how much that is but
2:15:14 like
2:15:14 I but I'm not doing I'm not doing
2:15:17 massive calls like if you're doing tens
2:15:19 of thousands of calls I'm sure the price
2:15:21 gets up there
2:15:23 um but here's the good news
2:15:27 if you're thinking about developing with
2:15:29 open ai's things and you're concerned
2:15:32 about cost just go spin up one of the
2:15:35 open source large language models that
2:15:36 like they're getting really good
2:15:38 go start testing all that open source
2:15:41 stuff because that's all free it's all
2:15:42 free
2:15:44 go down the download the GitHub
2:15:46 repository and tell chatgpt to tell you
2:15:50 how to install that [ __ ] on a Google
2:15:52 collab machine and go make yourself a
2:15:55 thing
2:15:57 and it won't cost you a [ __ ] it'll cost
2:15:59 you what Google co-lab will you know
2:16:01 charge you to run their virtual machine
2:16:04 for a week
2:16:07 will the lottery be easier to win based
2:16:10 on the data it will give well I think
2:16:12 there's a difference between
2:16:14 what is it probability and
2:16:18 I don't know I'm too tired uh no it
2:16:21 won't
2:16:23 it you will still
2:16:25 more likely be struck by lightning three
2:16:28 times in a week and survive it than hit
2:16:30 the lottery
2:16:31 you cannot go faster than Carl Lewis
2:16:33 that's true
2:16:35 I
2:16:36 I don't think I can go faster than Carl
2:16:39 McGillicuddy and I don't even know who
2:16:40 that is how about gi's Mac
2:16:44 not with me this semantics Roger copy
2:16:48 that I don't know what you're saying
2:16:50 there
2:16:51 what would what would you recommend is
2:16:53 the first step to really learn AI That's
2:16:56 why I put these up here first step well
2:16:58 it so when you say really learn AI
2:17:01 I assume you're talking about really
2:17:03 learn generative AI really learn these
2:17:05 new tools that we can just use
2:17:09 um
2:17:11 start understanding what they make
2:17:12 possible go to chat.openai.com I'm going
2:17:15 to go to GPT 3.5
2:17:19 so this is the non-fancy this is the
2:17:22 free version of chat GPT it's incredibly
2:17:25 fast and I'm going to type in like like
2:17:28 the best way to get started just start
2:17:30 doing what I call parlor trick so I'm
2:17:32 going to say
2:17:33 um I have
2:17:36 a fat
2:17:37 black and white
2:17:40 dog named champ
2:17:44 that
2:17:45 Sparks at
2:17:48 everyone
2:17:50 and sings when I play the guitar which
2:17:55 is true
2:17:57 um please write me a country song
2:18:04 um I can sing to him
2:18:08 so start with something ridiculous and
2:18:10 stupid like that now watch this Bang
2:18:17 oh it's slow tonight oh there there it
2:18:19 went
2:18:20 it's thinking
2:18:22 um
2:18:23 it knows the difference between verses
2:18:25 and choruses
2:18:26 well I got a dog named champ a buddy
2:18:28 like no other a fat black black and
2:18:30 white fella a real country Lover he
2:18:32 barks at every passerby protecting what
2:18:34 he's got but I strummed the guitar when
2:18:36 he joins on the spot
2:18:38 oh my champ my loyal friend barking and
2:18:41 Howling from beginning to end we make
2:18:42 music together A Perfect Harmony in this
2:18:45 country song for my champ you see so do
2:18:49 stupid [ __ ] like this and then say
2:18:51 um
2:18:52 um let's see
2:18:54 um make this a short story
2:19:00 Once Upon a Time so now there's a short
2:19:04 story bang done
2:19:06 um
2:19:07 create a table that compares good dogs
2:19:16 with bad dogs like champ
2:19:23 there's a table
2:19:25 comparing good good and good and bad
2:19:27 dogs
2:19:28 um appearance good dogs friendly and
2:19:31 approachable bad dogs intimidating and
2:19:33 aggressive
2:19:39 oh my God training eager to learn quick
2:19:42 to obey bad dogs difficult to train and
2:19:45 stubborn oh my God
2:19:48 this is awesome
2:19:50 um so so it so start with stuff like
2:19:54 this so and then and then I would say
2:19:56 quickly move from parlor tricks
2:20:00 to
2:20:02 um
2:20:03 do a project do a project like for your
2:20:06 community group or for your business
2:20:07 maybe your business is struggling with
2:20:09 marketing
2:20:11 use chat CPT to write a marketing plan
2:20:13 just start learning it just start
2:20:15 getting your hands dirty start watching
2:20:17 YouTube videos start you know doing two
2:20:21 hour long Tick Tock lives at 58
2:20:27 I mean seriously this channel is like
2:20:29 here as much for me to learn as as it is
2:20:31 for me to teach because
2:20:35 it like
2:20:37 it's impossible to know where this stuff
2:20:39 is going
2:20:41 all I know is that it's going to change
2:20:43 everything and I want to be as in it as
2:20:46 I can be like so someone was I was
2:20:49 talking to said
2:20:50 not everyone's a tsunami Surfer right
2:20:53 this is a tsunami that's coming at us
2:20:56 and it's coming at us way faster than
2:20:57 anyone realizes and he said
2:21:00 not everyone's a tsunami Surfer and I'm
2:21:02 like ah I like the metaphor and where I
2:21:05 am is I'd much rather be on the
2:21:07 surfboard
2:21:08 and chance wiping out then be in the in
2:21:12 the Cozy Beach Cabana
2:21:15 with the tsunami coming right so all the
2:21:18 people that are like it's not perfect or
2:21:21 you know it's not really AI it's machine
2:21:24 learning okay there's still a tsunami
2:21:27 coming right
2:21:29 all the people that are like oh I've
2:21:30 probably got three years no there's a
2:21:33 tsunami coming
2:21:34 so so that's so so I think that's the
2:21:39 best way
2:21:40 a special purpose dear internet thank
2:21:44 you for comedy
2:21:46 Nadia you seemed to have a purpose here
2:21:51 dear thank you for your efforts you're
2:21:53 welcome Kyle's purpose is to distress
2:21:55 Nadia winning exactly got her
2:21:59 he used words I didn't like that he used
2:22:02 words why do you use words Kyle
2:22:05 um
2:22:07 Ubi when so Universal basic income
2:22:11 what's fascinating on that one for me is
2:22:13 when Andrew Yang was going all crazy
2:22:15 about Ubi and like AI is going to demand
2:22:18 Ubi I was like that's interesting but
2:22:21 what the [ __ ] is he talking about and
2:22:23 then I had my my little brain Epiphany
2:22:26 with chat GPT YouTube could make money
2:22:28 with chat ubt
2:22:31 um
2:22:33 there's going to be tremendous
2:22:34 disruption I don't think the Ubi thing
2:22:36 is going to be required long term right
2:22:40 I think it can absolutely be a tool for
2:22:43 the next two to three years maybe five
2:22:46 years when all this disruption happens
2:22:48 because what's going to happen is
2:22:50 a bunch of jobs are going to get
2:22:51 displaced and and there's probably going
2:22:53 to be some enough significant disruption
2:22:56 that it's in the same neighborhood of
2:22:59 disruption as covid where it's like the
2:23:01 government's just like oh [ __ ] we got to
2:23:02 do something right
2:23:03 and then it might look like Ubi it might
2:23:06 look like something else
2:23:07 but then what's going to happen is
2:23:10 all the jobs that got displaced are
2:23:11 going to get replaced with a bunch of
2:23:12 new jobs that we've never even imagined
2:23:15 could have been jobs before right so
2:23:17 just like when the Industrial Revolution
2:23:18 happened a bunch of people got displaced
2:23:21 and then all these new jobs were created
2:23:23 so it's going to be a similar sort of
2:23:25 thing like that hopefully the new jobs
2:23:27 that are post AI are not as soul and
2:23:30 body crushing as the Industrial
2:23:32 Revolution but you know who knows
2:23:35 um but but so I don't know that it will
2:23:38 be in the next year but I have a feeling
2:23:40 probably within two you're going to need
2:23:42 to do something
2:23:43 what skills do you need to teach our
2:23:45 kids now with AI prompt engineering
2:23:48 understanding inter understanding intent
2:23:50 yeah I think the big skills moving
2:23:53 forward is essentially critical thinking
2:23:55 so understanding intent what do you want
2:23:57 to accomplish what are we trying to do
2:23:59 what does success look like what's good
2:24:01 what's bad and then understanding the
2:24:04 tools enough to ask for it right so so
2:24:07 rather than saying you know we're going
2:24:09 to try to catch people that are cheating
2:24:11 using chat GPT to write papers bring it
2:24:14 into the classroom and say let's have
2:24:16 chatgpt write the paper and then we're
2:24:18 going to analyze it as a class to see
2:24:20 what it got right what it got wrong how
2:24:23 would we do it different how might we
2:24:25 prompt it in a different way that it
2:24:26 would give us a better result like all
2:24:28 of that sort of stuff that's going to be
2:24:30 the critical skill moving forward the
2:24:32 Tactical execution of of the day-to-day
2:24:35 mind-numbing [ __ ] that we all have to do
2:24:38 every day let the robots do that
2:24:41 Elevate our brains and our students
2:24:45 brains
2:24:46 to to a higher level of Engagement
2:24:49 critical thinking
2:24:51 like I you know I feel like we're we're
2:24:53 entering this revenge of the liberal
2:24:55 arts major era right
2:24:58 have a broad view be able to look at
2:25:01 things objectively and say is that good
2:25:03 is that bad what's that based on I don't
2:25:06 know my world experience critical
2:25:08 thinking understanding the problem right
2:25:11 not just being a cog in the [ __ ]
2:25:12 wheel
2:25:14 when will mid-journey get plug-ins when
2:25:16 will mid-journey get a [ __ ] interface
2:25:19 um it's a billing code
2:25:21 I surface read that article that AI is
2:25:25 keeping us in a sort of bull market what
2:25:27 do you think
2:25:28 I think that there's
2:25:32 a non-zero possibility I've I've heard
2:25:35 that from some Engineers recently I
2:25:37 think it's a great term there's a
2:25:38 non-zero possibility
2:25:41 that so so one of the things I've heard
2:25:44 is that Tech shops engineering shops
2:25:46 that are that are fully embracing AI for
2:25:49 their coders are seeing a 30 to 40
2:25:52 percent productivity increase that's a
2:25:55 lot in business a 30 to 40 percent
2:25:57 productivity increase with an existing
2:25:59 team is a [ __ ] ton it's a lot
2:26:03 so
2:26:05 the the pessimistic view of this right
2:26:08 the doomer's view of this is if you're
2:26:10 30 or 40 or percent more efficient you
2:26:13 fire 34 30 to 40 percent of the people
2:26:16 or
2:26:18 you have a 30 to 40 percent increase in
2:26:21 productivity that you dump back into the
2:26:23 economy
2:26:25 so it's possible there's a non-zero
2:26:27 chance that this AI stuff if if the
2:26:30 impact comes
2:26:32 swiftly which I think it will
2:26:34 and if the private Equity firms don't
2:26:37 just bleed
2:26:39 bleed all the resources out of companies
2:26:41 to to you know skim more money off which
2:26:44 they're gonna do that but hopefully more
2:26:47 companies you know increase their
2:26:49 productivity so I think it's possible
2:26:52 um insurances use different calendars
2:26:55 insurance is use every game in the book
2:26:57 to to screw with us
2:27:00 our phones listen and yet an info
2:27:02 marketing Cog we go mint Y2K okay can I
2:27:06 come up for a five minutes I have
2:27:09 something to show and see if you can
2:27:11 explain
2:27:13 um possibly I don't know how to bring
2:27:15 you up let me see if I can find you in
2:27:18 here I probably can I don't see you in
2:27:21 here I guess raise your hand or do
2:27:24 whatever you need to do Mexico
2:27:26 I'm happy to bring you up the problem
2:27:27 with bringing you up is I lose my dual
2:27:29 screen so that thing's gonna go away but
2:27:31 I think that's important but it may not
2:27:33 be
2:27:35 Vic thank you very much Oculus what
2:27:38 about Oculus
2:27:40 um what's the difference between an llm
2:27:42 and human reasoning and or Consciousness
2:27:45 so an llm
2:27:48 and llm has the consciousness of a
2:27:50 calculator
2:27:53 it has none now
2:27:55 it behaves
2:27:58 like it's got Consciousness and it
2:28:00 behaves like it's got knowledge and it
2:28:02 behaves like it's got empathy
2:28:05 but it actually doesn't like what it's
2:28:08 doing is it's predicting one word after
2:28:10 the other you put in your prompt there's
2:28:12 a mathematical equation that's created
2:28:14 they dump that mathematical query into a
2:28:20 massive number database and the numbers
2:28:23 basically say there's a pro the highest
2:28:26 probability of the next word that I
2:28:28 should respond with is this word
2:28:30 and then it does it again and again and
2:28:32 again so so it it writes all these
2:28:34 individual words that are just
2:28:36 mathematical
2:28:38 um
2:28:38 predictions
2:28:40 and then we as humans read it and we're
2:28:42 like oh
2:28:44 it must have Consciousness because it
2:28:45 said It felt something it said it was in
2:28:47 love with that New York Times Reporter
2:28:49 and he should leave his wife
2:28:53 but it it wasn't saying that it just
2:28:56 typed those words right so the
2:28:58 difference is human beings
2:29:01 can make sense of those words the large
2:29:04 language model is just putting them out
2:29:05 there
2:29:06 there will be a version in the future
2:29:08 where it puts them out there and then it
2:29:12 analyzes what it did and then it you
2:29:15 know so there's a version of this in the
2:29:17 future where those two things get much
2:29:20 closer right
2:29:22 but you know why these large language
2:29:23 models are good why they work so well is
2:29:25 they've they've mimicked the human brain
2:29:27 how it learns so it learns like a human
2:29:30 brain but it's not processing like a
2:29:32 human brain
2:29:34 all right cursing for
2:29:38 this I don't know what that word is
2:29:40 dislodges
2:29:42 dialogues Maybe
2:29:45 officially obsessed with AI now I agree
2:29:48 AI can free us up for The Human
2:29:50 Experience work is an escape for many
2:29:53 yeah and I think you know work is also
2:29:55 one of those things where
2:29:58 there's a there's a hopeful like there's
2:30:00 there's just going to be some really
2:30:01 [ __ ] pain coming which I hate
2:30:05 but the the Silver Lining that I'm
2:30:07 hoping is that people are going to be
2:30:10 displaced from jobs that they
2:30:12 hated but they wouldn't admit that they
2:30:14 hated and then once they're gone and
2:30:16 once they realize that job's not coming
2:30:18 back
2:30:19 that they might go you know I [ __ ]
2:30:22 hated that job you know what I've always
2:30:24 wanted to do cookies let me start a
2:30:27 [ __ ] Cookie Company like I think
2:30:29 there's going to be a lot of reinvention
2:30:30 and and also all the displaced people
2:30:33 also have access to these tools right so
2:30:35 they're going to be able to be competent
2:30:37 and masterful at things that they
2:30:40 historically would not have had the
2:30:42 decade or the time to put in to get good
2:30:45 at because the tools are going to give
2:30:47 them a leg up
2:30:49 I worry a bit that a lot of people want
2:30:52 to do routine work I worry a bit that a
2:30:55 lot of people want to do routine work
2:30:57 yeah not everyone's an entrepreneur I
2:30:59 agree with that
2:31:00 there will still be routine work it'll
2:31:03 just be different work right if your
2:31:05 routine work is sifting through
2:31:07 insurance forms to find typos or to do
2:31:12 data entry that kind of shit's going to
2:31:15 go away but there's going to be other
2:31:16 kind of routine work like maybe
2:31:18 maintaining the AI
2:31:21 I don't know but Library I don't know
2:31:23 who knows
2:31:25 but there are there are different kinds
2:31:27 of people not everyone wants to be an
2:31:28 entrepreneur AI can change a person's
2:31:30 entire life right yes
2:31:31 I agree like so for example
2:31:35 um Stephen on on
2:31:37 um like a week and a half ago there's a
2:31:40 guy in Italy that uh was paralyzed in a
2:31:42 motorcycle accident
2:31:44 and they created a Bluetooth bridge over
2:31:46 the injured part of his spine they used
2:31:48 a large language model to read his brain
2:31:51 brain scan his MRI I brain scan had him
2:31:54 think about moving his legs so the large
2:31:56 language model learned this is what his
2:31:58 brain does when he's thinking about
2:32:00 moving his legs it then sent Bluetooth
2:32:02 signals down to electrodes in the lower
2:32:04 part of his spine and he's walking
2:32:08 he's not working great like it ain't
2:32:10 perfect but like that happened that
2:32:14 happened last week
2:32:15 so so I think I think the level of
2:32:17 impact you know to humanity is going to
2:32:20 be profound I think
2:32:22 it's going to do things like solve
2:32:23 genetic diseases and it's gonna do
2:32:26 things like cure cancer and it's going
2:32:27 to do things like solve the energy
2:32:29 crisis and solve the you know climate
2:32:32 crisis and there's a lot there's a lot
2:32:36 that it's going to change so I think on
2:32:38 an individual level it can be
2:32:40 transformative and then on societal
2:32:42 level I think it can be transformative
2:32:44 and it could just kill us all
2:32:48 oh my God the discourse thing's so
2:32:50 annoying oh Discord thing yeah it really
2:32:52 is
2:32:54 um making a note to send Kyle a black
2:32:57 vest to wear during our lives I know my
2:33:00 I like these fancy shirts but then you
2:33:02 know you can't read the text sorry about
2:33:04 that does anyone know how he does the
2:33:06 second camera I can tell you how to do
2:33:08 the second camera when you go live
2:33:10 there's an option that says dual camera
2:33:11 you turn it on that's it like it's just
2:33:15 it it's got this camera on and that
2:33:18 camera on at the same time and it puts
2:33:19 the back camera in the little window and
2:33:21 then but I'm not but I'm not looking for
2:33:23 them so if you got an acting geek for
2:33:25 for me I'd love to do one I've actually
2:33:27 thought about getting back into that I
2:33:28 haven't done any acting since last time
2:33:30 I did a play was I did Picasso with the
2:33:33 lapana Zeal in Ridgefield Connecticut in
2:33:38 2003 or four
2:33:40 that was the last acting I did I played
2:33:42 Elvis
2:33:43 that was fun
2:33:48 I like my question it was simple and
2:33:51 honest and straightforward
2:33:54 it was it was good solid solid question
2:33:56 I missed your response to anything for
2:34:00 front end WX oh I showed you that cool
2:34:02 thing
2:34:04 so
2:34:05 I just went to Google
2:34:09 I'll have to do it again I typed in mid
2:34:13 Journey
2:34:17 um ux
2:34:19 prompts
2:34:21 and then it popped up the UI design
2:34:26 and then when I went to the site
2:34:29 there are examples and prompts of all
2:34:32 these people doing really interesting ux
2:34:34 stuff
2:34:36 with mid-journey
2:34:38 and then what I talked about was you
2:34:39 know now these are just images so come
2:34:42 on Focus you dumb dumb come on dumb dumb
2:34:45 come on dumb dumb there you go
2:34:47 um
2:34:50 these are just images but they're
2:34:52 they're images that are you know really
2:34:54 interesting fun ux stuff that you can
2:34:56 take into figma but then what I talked
2:34:58 about was go to
2:35:02 go to uh
2:35:05 futurepedia.io that's a list of that's a
2:35:08 directory of a bunch of AI tools I'm
2:35:10 sure there are ux specific things
2:35:12 because I'm pretty sure there's there's
2:35:14 lots and lots of things right now that
2:35:15 are text to like build me a mobile
2:35:18 interface build me a web interface
2:35:20 that's actually building like layered
2:35:22 figma files and [ __ ] like that I'm
2:35:24 pretty sure they're out there but if you
2:35:25 just want to do the ideation doing ux
2:35:28 and mid Journeys really really
2:35:30 pretty cool
2:35:34 we need a plugins manager filter we we
2:35:36 need so many things
2:35:38 we need I need the ability to search my
2:35:42 previous prompts because since I've been
2:35:44 doing these lives I go into my prompts
2:35:46 and I do all these prompts for the lives
2:35:48 and I can't find the [ __ ] that I did for
2:35:49 work so
2:35:51 yeah it's like chat GPT right now it's
2:35:54 it's just it's really raw and new
2:35:58 it just is what it is right now Google
2:36:01 AI class is worth doing very short and
2:36:03 good to understand how involved they've
2:36:05 been with llms yeah I mean listen Google
2:36:07 invented the Transformer the thing that
2:36:09 powers all of this cool [ __ ] that we're
2:36:12 playing with was invented by Google in
2:36:13 2017.
2:36:15 how do you do you know how to get access
2:36:18 to code interpreter no but if you figure
2:36:20 it out will you tell me
2:36:22 so what it's going to be is so I've seen
2:36:25 the interface
2:36:27 so when you're in chat GP
2:36:30 there's this
2:36:32 option called beta features and I've
2:36:34 seen the interface so right now I've got
2:36:37 access to browse with Bing and plugins
2:36:39 at some point there's going to be code
2:36:41 interpreter right here
2:36:43 but it ain't there and so that makes me
2:36:47 cranky
2:36:48 because some people have access to it
2:36:50 but I don't have access to it and it's
2:36:53 just sad
2:36:55 I'm so sad
2:36:58 I want code interpreter
2:37:04 um so no I don't
2:37:08 why is he such an idiot I don't know
2:37:14 because I've been doing this for two
2:37:16 hours my ass is numb what do you want
2:37:18 from me people
2:37:20 apparently Nadia wants me to have a
2:37:22 purpose
2:37:28 remind me where I can paste The Prompt
2:37:31 into Discord Okay so
2:37:34 if you if you go to poe.com Mid Journey
2:37:38 it takes you to this mid-journey bot
2:37:41 and then you have to go to
2:37:45 the mid-journey Discord or any Discord
2:37:48 that's got the mid-journey bot installed
2:37:50 in it this is the AI bot but you you
2:37:53 have to have an account with mid-journey
2:37:55 where where you actually you you either
2:37:57 have free credits
2:37:59 or you've or you've paid for it you've
2:38:01 got a subscription and then once you do
2:38:03 that you come to a mid-journey channel
2:38:07 in Discord and you type slash imagine
2:38:11 and see It'll see it pops up imagine
2:38:13 right here so I just hit return and it
2:38:16 will type it down here and so there's a
2:38:18 little prompt window and now I can
2:38:22 copy this prompt and paste it in here
2:38:25 and hit return
2:38:27 and now it will go make me that image
2:38:31 ever notice that GPT 3.5 is DaVinci
2:38:35 002 look at the URL I thought it was 003
2:38:38 I think it's 003
2:38:41 but it might be two I don't know
2:38:43 I I haven't really paid attention to
2:38:45 that
2:38:49 there's there's our Alba try
2:38:53 albatrosses
2:38:55 that doesn't seem like the right plural
2:38:56 there
2:38:58 can you check out my mid-journey
2:39:02 my mid-journey prompt pobot slash MJ
2:39:06 Captain sure let's go look at it now
2:39:09 let's all go look at it m MJ Captain so
2:39:12 we're gonna go po.com
2:39:16 slash
2:39:20 e m j captain
2:39:27 a box that will generate optimized
2:39:29 mid-journey prompts you can request
2:39:31 either a photo painting or design and
2:39:33 the bot will create
2:39:35 a prompt that combines different styles
2:39:37 and words welcome okay cool oh sorry
2:39:40 there you go
2:39:42 that's MJ captain
2:39:45 this is bird dog 9000 awesome name
2:39:50 what's your name bird dog 9000 why
2:39:54 um okay so let's say photo of a
2:39:59 brand new
2:40:01 1969 GTO
2:40:05 judge in an abandoned Factory
2:40:13 um the brand new 69 judge parked in an
2:40:15 abandoned Factory rusted pipes in the
2:40:17 background and the style of Ansel Adams
2:40:19 black and white that'll be cool Diane
2:40:21 Arbus high contrast photography
2:40:23 photography vintage car 1960s 4x3 aspect
2:40:28 ratio yeah this looks cool so let's grab
2:40:30 this I think the V 5.1 in style oh no I
2:40:35 miss never mind
2:40:37 there was there was no issue there so
2:40:39 that looks that seems that seems
2:40:41 promising and so I'm gonna go because
2:40:44 the that that other mid-journey bot that
2:40:46 I used seemed a little over verbose I
2:40:48 like that yours is a little simpler I
2:40:51 found that mid-journey can get
2:40:53 um
2:40:54 can get a little if you overdo it with
2:40:56 the prompts it doesn't it I think it
2:40:58 gets confused
2:41:00 all right
2:41:02 so so bird dog 9000 here's your big test
2:41:05 like the prompt is one thing but does it
2:41:07 make a cool image is the real thing uh
2:41:11 looks like they might be in color they
2:41:13 might be in black and white
2:41:15 don't know yet
2:41:17 uh uh looks like muted colors for sure
2:41:23 let's see if it got the 69 GTO Judge
2:41:25 right no not even close
2:41:28 that's not your fault though
2:41:31 that's mid-journey
2:41:34 those are kind of cool those kind of
2:41:36 look like um
2:41:38 some sort of future
2:41:40 like future reimagining of
2:41:44 like that's like a Mad Max reimagining
2:41:46 of it of a late 60s muscle car it's
2:41:49 pretty cool
2:41:50 but it's black and white all right we're
2:41:52 at 78 percent
2:41:56 yeah pretty cool
2:41:58 nice job yeah creating the bots on Poe
2:42:01 is pretty easy and it's it's fun so so
2:42:04 how you do that let's go I'll go show
2:42:06 you how to do a bot
2:42:07 so if you go to poe.com
2:42:10 anybody can do this by the way so you
2:42:12 got you got all these different large
2:42:14 language models up here and then you've
2:42:15 got your Bots down here and then you
2:42:17 you've got this thing called create a
2:42:19 bot and and so you say create a bot and
2:42:22 then you basically name it so he named
2:42:24 his MJ Captain which is a good name and
2:42:27 then he gave it a description and then
2:42:29 he chose is this going to be Claude
2:42:30 instant or chat gbt I use cloud instant
2:42:33 because it's fast and then you write a
2:42:35 prompt so this prompt might be act as a
2:42:37 mid-journey prompt expert and hear the
2:42:41 different you know attributes of a
2:42:44 mid-journey prompt and you basically
2:42:45 dick around with your prompt until you
2:42:47 get it doing what you wanted to do and
2:42:50 then the intro message is what the bot
2:42:52 says when you logs in log into it and
2:42:54 then you can upload an image
2:42:57 I think right here so you can make an
2:43:00 image in mid-journey to make your Bot
2:43:02 and then you say create bot and then you
2:43:05 have now created a bot I use this to
2:43:07 create Sling Blade Yoda
2:43:11 what else have I created Shakespeare
2:43:13 Emily Dickinson I am Steve Jobs I am
2:43:15 Jay-Z I am Maya Angelou Mary Shelley
2:43:18 Virginia Woolf Edgar Allan Poe Mark
2:43:21 Twain so it's fun it's it's a good way
2:43:24 to practice your
2:43:27 um your prompting your prompt
2:43:28 engineering your prompt writing
2:43:34 cross your fingers tomorrow I am asking
2:43:35 Dollar Tree for a job finally someone
2:43:38 will hire me
2:43:39 I hope so good luck
2:43:44 um how long did it take you to do the
2:43:45 course okay you're talking to each other
2:43:47 cool
2:43:50 ever notice that GPT 3.5 is oh we've
2:43:54 talked about that one
2:43:56 um
2:43:58 [Music]
2:44:01 Charlie Rose had a guy warning people of
2:44:03 universal AI because robots then
2:44:06 pandemic happened and
2:44:09 something Universal income
2:44:14 well yeah
2:44:18 I think there's going to be big
2:44:19 disruption in the employment Market
2:44:21 and I think someone's the government's
2:44:24 gonna have to do something about it and
2:44:26 not just ours
2:44:27 because this is not Just Happening Here
2:44:32 did you know that um those GPT models
2:44:35 um know essentially all the languages
2:44:38 they're really good at translating
2:44:40 languages
2:44:42 we also behave as if we have empathy
2:44:45 that's a great Point part of the matters
2:44:48 um core free self-will is so second
2:44:51 Millennium that's good mid-journey is
2:44:53 working on a web interface next
2:44:55 according to their office hours well
2:44:57 they should
2:44:58 you know listen you've got the best
2:45:01 you know image generation tool I think
2:45:03 they have two things they need to be
2:45:05 working on one is their legal department
2:45:06 because they're going to get their asses
2:45:08 sued from here to Oblivion the other
2:45:10 one's an interface
2:45:12 we need our own AI you can go get them
2:45:14 they're everywhere now they're just look
2:45:16 up open source large language models go
2:45:19 to go to Reddit search for open source
2:45:22 large language models run down that
2:45:24 rabbit hole you can have your own AI
2:45:28 like tonight
2:45:30 there's one called GPT for all it's not
2:45:33 gpt4 but it's GPT for all but you put
2:45:36 the four in there
2:45:37 um and it will install it on your local
2:45:39 machine and then you can pick different
2:45:40 large language models and it's all
2:45:42 running locally
2:45:43 you can go do that tonight you can
2:45:45 probably do that in the next half hour
2:45:47 anybody can be an entrepreneur
2:45:48 entrepreneur wait anybody can be an
2:45:51 entrepreneur but not everybody can be an
2:45:53 entrepreneur
2:45:56 everybody is free to be an entrepreneur
2:45:58 but yeah some people are not
2:46:00 entrepreneurial that's what I learned
2:46:02 when I did agency.com is that some
2:46:04 people just want jobs they're like tell
2:46:06 me what to do and I'll [ __ ] do it I
2:46:08 don't want to invent anything I don't
2:46:10 want to contribute to the [ __ ]
2:46:12 mission of the company like I'm a
2:46:15 copywriter let me write copy just tell
2:46:17 me what to do I'll sit here and shut up
2:46:20 and businesses need those kind of people
2:46:22 like those those are people that are
2:46:24 needed now
2:46:26 my question is are those the kind of
2:46:29 jobs that get automated out I don't know
2:46:31 possibly
2:46:33 and then what do those people do I don't
2:46:35 know that's where we might come that's
2:46:38 that's where everybody swing back to
2:46:39 Universal income because not everybody's
2:46:42 an entrepreneur
2:46:45 and not everybody should be or we
2:46:47 wouldn't have any existing businesses
2:46:49 like no one would serve you your you
2:46:52 know food
2:46:58 [Music]
2:47:03 AI versus subconscious data entry yes
2:47:07 exactly
2:47:10 we also behave as if we have empty of
2:47:12 I've already seen these sorry I lost my
2:47:15 place again because I'm just not good at
2:47:17 the tick tock
2:47:18 you know he's okay he's he just seems
2:47:21 old he seems old and tired and I don't
2:47:23 think he has a mission I don't think he
2:47:26 has a mission have you asked him if he
2:47:28 had one because I don't think he does
2:47:30 I mean what's his purpose what do you
2:47:33 think his purpose is
2:47:34 why is he here on tick tock
2:47:39 I just don't know
2:47:41 all right
2:47:44 all right
2:47:45 work change things started for years now
2:47:48 already work changed things started for
2:47:51 years now already
2:47:52 Sunny Aces Pet Retreat that English not
2:47:57 really as is a sentence
2:48:01 but I think I know what you mean
2:48:04 uh I'm trying to find a web-based AI app
2:48:07 that'll fix motion blur on a face
2:48:10 I'm pretty sure
2:48:13 that photoshop has that built in already
2:48:16 it's it might be under their AI tools
2:48:24 basically go to hang on
2:48:29 um
2:48:40 go to neural filters
2:49:04 it might just be I think it's just a
2:49:06 filter under under blur
2:49:09 it might be under neural filters which
2:49:11 is their AI filters but I'm pretty sure
2:49:19 oh it's not blur it's sharpened
2:49:23 go look at neural filters I'm pretty
2:49:25 sure that exists already in Photoshop
2:49:29 Kyle is cranky and sad no I'm not I'm
2:49:31 just tired I've been I've been doing
2:49:33 this for a while so
2:49:35 you have to log in as a developer to see
2:49:37 code interpreter I am logged in as a
2:49:40 developer wait
2:49:43 um maybe that would give me
2:49:46 access to gpt4 as well
2:49:51 um nostalgic flicks if you're still here
2:49:54 tell me how to how to log in as a as a
2:49:57 developer because I've got an API key so
2:50:01 I'm pretty sure I'm logged in as a
2:50:02 developer but if there's if there's a
2:50:04 super secret way to log into openai that
2:50:06 I can get a code interpreter and gpt4
2:50:10 please tell me
2:50:12 that would be nice when do you usually
2:50:14 go live whenever I have a wild hair up
2:50:16 my ass to go live
2:50:18 um how can you keep images private mid
2:50:21 Journey pay them 60 bucks a month
2:50:23 hi what carrier opportunities
2:50:27 oh career what career opportunities in
2:50:29 AI Industries I don't know
2:50:33 it's it's more like
2:50:39 I I don't I don't know I mean there's a
2:50:42 lot of talk about prompt engineers and
2:50:44 prompt engineering pays these wild
2:50:46 six-figure salaries I don't think prompt
2:50:49 engineering is going to be
2:50:51 um
2:50:53 I think a lot of that's going to get
2:50:55 automated into the background really
2:50:56 really quickly I think that's a
2:50:57 short-term blip but we'll see
2:51:00 um
2:51:03 I would say take what you're good at and
2:51:05 learn how to do it with AI start there
2:51:09 because every industry every career
2:51:12 every job is going to be
2:51:15 um
2:51:16 is going to be
2:51:19 impacted by it so you should be one of
2:51:21 the people that knows how to do what you
2:51:24 do well with AI
2:51:25 and you'll you'll find a career there
2:51:28 somewhere
2:51:31 retrofuturistic is this Zelda we collect
2:51:33 Poe
2:51:36 um can you create a chat GPT plugin
2:51:39 on PO dot AI
2:51:42 I don't think so I don't think Poe's
2:51:45 supporting
2:51:47 um
2:51:49 I don't think Poe's supporting
2:51:52 plugins right now but it but I could be
2:51:54 wrong I don't know
2:51:55 am I missing the link to to his Discord
2:51:59 so
2:52:00 if you want to link to the AI Salon
2:52:02 Discord go to my profile there's a link
2:52:05 tree there
2:52:06 the salon.ai is the website that's got a
2:52:10 link tree and in that is the invite to
2:52:12 the salon Discord you can also just go
2:52:15 to the mid Journey Discord if you want
2:52:16 to do things there
2:52:18 I hope we can get more young people in
2:52:19 other countries to access generative AI
2:52:21 I do too I think it sucks that a bunch
2:52:24 of countries are blocking it all
2:52:25 together but now that we have opening
2:52:27 Source large language models that shit's
2:52:30 gonna end
2:52:37 all right can you repeat to have your
2:52:41 own AI bot yeah go to
2:52:44 what's going on here that's a menu
2:52:47 go to
2:52:49 po.com
2:52:53 and then in the left hand
2:52:55 column
2:52:57 go down to create a bot
2:53:01 and then when you do that in the right
2:53:03 hand column you can put a picture of
2:53:05 your Bot you name it give it a
2:53:07 description
2:53:08 choose the model
2:53:10 write the prompt and this is just like
2:53:12 writing a prompt in chat gbt you're just
2:53:15 you're just telling it to be a certain
2:53:16 kind of Bot so it's going to rewrite it
2:53:19 like like I did the Shakespeare one I
2:53:22 told it it was Shakespeare and that it
2:53:24 should always respond to niambic
2:53:25 pentameter and then you give it a little
2:53:28 greeting message and you build it and
2:53:30 you're done and then
2:53:32 you can tell someone to go to poe.com I
2:53:35 am Shakespeare or I am mid-germy or I am
2:53:37 MJ captain
2:53:40 [Laughter]
2:53:41 [Music]
2:53:42 looks like we got some new people just
2:53:44 joined um I've been doing this for like
2:53:46 two and a half hours now so I am bloody
2:53:50 eyed
2:53:51 um my butt's numb
2:53:53 and uh I I'm I'm a little moist and uh
2:53:58 but this is this is the AI learning lab
2:54:00 my name is Kyle Shannon I'm an
2:54:02 entrepreneur I answer questions about AI
2:54:04 if you've got questions pop them down
2:54:06 below if you're new to ai go to these
2:54:08 websites the top one is the official
2:54:10 chat GPT website I'm not affiliated with
2:54:12 it they should [ __ ] pay me
2:54:13 commissions because I'm like a giant ad
2:54:16 for them every night I do this but
2:54:18 they're not
2:54:20 um they're just they're like that's the
2:54:21 official one and then bing.com is gpt4
2:54:25 for free and po.com you can do a bunch
2:54:27 of [ __ ] including making your own Bots
2:54:29 which I just showed you so
2:54:31 that's who I am and what this is
2:54:34 Kyle's here for Nadia you're funny
2:54:36 Emilio's wife all right thanks for your
2:54:38 lives enjoyable educational thank you
2:54:40 what is a local machine Okay so
2:54:44 right now when you use chat EBT
2:54:47 if you type in
2:54:50 um write me a poem about X
2:54:53 um it sends that prompt to
2:54:55 Microsoft servers that are running AIS
2:54:59 models and then it sends you back an
2:55:02 answer running it local means you're
2:55:04 running it on your local machine using
2:55:06 your machine's own processing power so
2:55:10 when you type in your prompt it just
2:55:12 processes it processes it locally with a
2:55:15 local large language model and gives you
2:55:17 back the answer just locally so nothing
2:55:19 goes away from your computer so that's
2:55:22 the that's the advantage there so
2:55:25 um
2:55:26 you you just have much more control you
2:55:29 can use open source large language
2:55:30 models that don't have safety guard
2:55:32 rails on them for example things like
2:55:34 that that's why people want to do that
2:55:38 oh and you already answered each other
2:55:39 local machine equals your hard drive
2:55:41 okay so sure I could have said it in
2:55:43 three words but I chose to use 642 words
2:55:46 to answer it because I have diarrhea of
2:55:49 the mouth any legit way to integrate
2:55:53 with mid-journey API I don't think so
2:55:55 they don't have an API yet they don't
2:55:57 have an interface
2:55:59 it's incredibly convoluted
2:56:02 but it's just the [ __ ] best
2:56:06 they need an API they need an interface
2:56:12 mid journey is very frustrating right
2:56:14 now
2:56:15 because when you have a tool that's that
2:56:17 good and also the the CEO I've seen a
2:56:19 couple of interviews with him he's a bit
2:56:21 of a douche which that does not
2:56:23 I just there's parts of that
2:56:26 organization that just yeah rubbed me
2:56:28 wrong but God damn the quality of the
2:56:31 images is so good
2:56:35 um what's the best platform to generate
2:56:36 music in the style of a specific
2:56:39 musician with certain mood and Tempo I
2:56:41 don't know there's so many of them
2:56:43 that's why I have futurepedia.io up here
2:56:46 um I can't keep up with the tools
2:56:48 anymore at all like and I have a
2:56:50 newsletter like it's it's part of my
2:56:52 self-dis self-imposed job is to learn
2:56:56 all this [ __ ] Google just released their
2:57:00 music LM so you can go to Google they've
2:57:02 got a text to music thing
2:57:05 um
2:57:06 I I would just go to futurepedia.io go
2:57:10 to the music section and just start
2:57:11 clicking on links
2:57:14 and then you're gonna find like there's
2:57:16 gonna be I don't know
2:57:18 there's gonna be 150 music tools that
2:57:20 all claim to be the best thing since
2:57:22 sliced bread there's gonna be two of
2:57:23 them that actually don't suck so it's
2:57:26 going to be time consuming and annoying
2:57:27 but I guarantee you
2:57:29 there are one or more tools out there
2:57:31 they're gonna [ __ ] melt your face off
2:57:33 and you'll just be like really I'm so
2:57:36 screwed
2:57:37 and that and then so so here's the deal
2:57:40 if you have that moment where you're
2:57:42 like I'm [ __ ]
2:57:43 don't run from it like lean into it and
2:57:46 just now start learning that tool like
2:57:49 get really good at it
2:57:50 because here's my theory the the AI tool
2:57:53 that scares you the most is the one that
2:57:55 you should be learning because it's the
2:57:57 one that's the most the the biggest
2:57:58 existential threat to who you are and
2:58:00 what you do
2:58:02 and my theory is this you're going to be
2:58:05 better at using that tool than someone
2:58:06 who doesn't isn't as good as that [ __ ]
2:58:09 as you are right if someone just comes
2:58:11 and doesn't know anything about music
2:58:13 they're going to make music with that
2:58:14 tool it's going to be okay you're going
2:58:16 to be able to do things of a different
2:58:18 level so anyway
2:58:20 um Good Luck jump in
2:58:23 get scared
2:58:25 um if you're if you're if as you dig
2:58:28 into this stuff
2:58:30 you can't figure out if you're excited
2:58:32 or scared
2:58:34 that's the appropriate response
2:58:38 thanks for checking out Poe it's based
2:58:40 on prompts I based it off prompts from
2:58:43 the describe feature in mid-ring oh
2:58:45 that's cool this is really cool so I
2:58:47 forget how it works do I go
2:58:51 copy link
2:58:54 you go slash describe
2:58:59 you're gonna slash describe
2:59:02 oh and then and then you drag an image
2:59:05 there so let me go grab an image off my
2:59:07 hard drive okay we'll grab my this is
2:59:10 the image I did for this is the Genesis
2:59:13 image for my Kyle's Shannon dreams art
2:59:16 project so I'm going to drag that onto
2:59:18 this thing in mid-journey
2:59:21 and I'm going to hit return and now it's
2:59:24 going to read that mess it's going to
2:59:26 read that image and it's actually going
2:59:28 to write a prompt for me
2:59:30 or four of them
2:59:32 so first prompt the man wears blue
2:59:35 goggles and a suit in a city setting in
2:59:38 the style of neon infused
2:59:40 digitalism realistic hyper detailed
2:59:44 portraits security camera art movie
2:59:46 poster
2:59:48 Jason Middlebrook futuristic Gothic AR
2:59:51 8x11 so I'm going to copy that
2:59:54 and now I'm going to go slash imagine
3:00:00 and I'm going to paste in so so it read
3:00:03 the message gave us the prompt now I'm
3:00:05 going to put that prompt back into mid
3:00:07 journey and have it generate an image
3:00:08 based on the description only so this is
3:00:11 not using the image at all just the
3:00:13 description it gave and we'll see what
3:00:15 it came up with it usually does you get
3:00:17 you get really interesting things this
3:00:19 way so wait here
3:00:22 um
3:00:23 gotta go to bed I'm old and it's 12 30
3:00:26 here I know it's it's 11 30 here thanks
3:00:28 for all the info you're fabulous thank
3:00:30 you
3:00:31 that's that's all people got to stick
3:00:33 together I gotta go to bed soon
3:00:37 um oh this is starting to look cool
3:00:39 uh
3:00:41 is stable diffusion comparable quality
3:00:44 to mid-journey it's getting close and
3:00:46 there there's some things where stable
3:00:48 diffusion can be better there's
3:00:51 there's other things where it's not the
3:00:53 the advantage of stable diffusion is
3:00:55 stability AI open sourced it and the
3:00:59 developer so the problem with mid
3:01:00 Journey right now is it's all closed
3:01:03 it's all behind closed doors so
3:01:07 whatever developers they have are how
3:01:10 they're advancing their technology which
3:01:12 is why they don't have a good interface
3:01:13 and a good this and a good that like
3:01:15 they only have so many resources what's
3:01:17 happening with stable diffusion
3:01:19 is that it's opening um it's it's it's
3:01:22 open source so the entire development
3:01:25 Community every time there's something
3:01:26 new from stability AI put out in the
3:01:28 world the developer community pounces on
3:01:31 it and creates all these tools so stable
3:01:33 diffusion has a lot more tools and a lot
3:01:35 more flexibility and multiple visual
3:01:37 interfaces in front of it and there's
3:01:40 leonardo.ai is sitting on top of stable
3:01:42 diffusion and I don't know there's a
3:01:44 bunch of them out there so stable
3:01:46 diffusion is just it's just a much
3:01:47 bigger ecosystem but quality of image
3:01:51 mid-journey still has it in my opinion
3:01:53 so here's these images that it created
3:01:55 you know based on
3:01:57 the description of that it made those so
3:02:00 it's in the neighborhood pretty cool
3:02:04 all right
3:02:07 what's the best use of these sites which
3:02:10 industry example accounting or
3:02:12 photography which cites these
3:02:16 the top three are really just chat GPT
3:02:20 like start learning
3:02:22 what these large language models are and
3:02:24 can do the prompts one helps you prompt
3:02:27 there and then future PD is just here's
3:02:29 all the tools you want to dig into this
3:02:31 [ __ ]
3:02:32 they're they're it's almost like every
3:02:34 tool at Future pedia is a rabbit hole
3:02:38 that you could spend the next week on
3:02:40 and there's thousands of them so that's
3:02:44 what's going on in the industry if if
3:02:46 you're feeling overwhelmed and behind
3:02:48 welcome to the club
3:02:50 your information is very interesting
3:02:52 thank you Bonnie I appreciate that which
3:02:54 AI is good for it wait I lost it damn it
3:02:57 which AI is good for creating children's
3:02:59 book characters just all of them you
3:03:01 just need to learn to prompt tell it to
3:03:03 make you a children oh well so there's a
3:03:06 weird Okay so
3:03:08 this will get better but it's really
3:03:10 shitty right now
3:03:12 so one of the things that's actually
3:03:14 really hard for these things is if you
3:03:15 create a character in mid-journey that
3:03:17 you really like and then you want to
3:03:19 carry that character forward and put it
3:03:21 in different scenarios but have it be
3:03:23 really consistent that's really hard
3:03:27 in stable diffusion there's this project
3:03:29 called control net which allows you to
3:03:33 do that but it's really complicated to
3:03:35 set up so I would go Google or whatever
3:03:41 control net for stable diffusion
3:03:44 um and and it will help
3:03:47 all right I'm an educator and would like
3:03:49 to help my students learn how to use
3:03:51 language where can I start that's why
3:03:54 these are here start with chat gbt just
3:03:57 start with chat EBT
3:03:59 have them toss you out ideas you type it
3:04:02 in they watch the results then have them
3:04:04 type them in
3:04:06 it's you'll see
3:04:08 like their eyes will get big
3:04:11 it's fun stuff
3:04:15 um all right I'm gonna go soon I'm gonna
3:04:16 go which AI translated various which is
3:04:19 AI translated various languages chat EPT
3:04:22 understands all the languages just tell
3:04:24 it to translate say paste in something
3:04:27 in English and tell it to translate some
3:04:28 language that you know it's it's really
3:04:31 quite good this has been the best
3:04:33 quality time I've ever spent on Tick
3:04:35 Tock good Lord sorry about that uh but
3:04:38 thank you I appreciate the kind words
3:04:41 um have you ever used chat GPT to
3:04:42 generate mid-journey prompts yes and I
3:04:44 and we just did this before we use Poe
3:04:46 to do that as well
3:04:48 um name your character and help the AI
3:04:51 maintain the same character in your book
3:04:53 yeah that's the thing that's really hard
3:04:55 control net for stable diffusion helps
3:04:58 that
3:04:58 Adobe will crack the nut on that
3:05:01 um as will others it's a real pain in
3:05:04 the ass right now so you you
3:05:07 there's not a good answer to that one
3:05:09 unfortunately well there is wait a year
3:05:12 what are your thoughts on Jasper so
3:05:14 Jasper Jasper has been around for longer
3:05:17 than chat gbt has so they're built on
3:05:19 top of
3:05:21 um gpt3
3:05:23 um so it's basically Jasper and copy Ai
3:05:25 and all those things are are basically
3:05:28 pre-written prompts like every little
3:05:30 button that you push on like the the
3:05:32 blog writer and the Tweet writer and
3:05:34 this and that those are all just prompts
3:05:36 sitting on top of gpt3 so you could do
3:05:40 the equip the The Identical thing that
3:05:43 they do by going into chat GPT and
3:05:46 coming up with a prompt that
3:05:47 consistently gives you good tweets or
3:05:49 good blog posts the difference with
3:05:53 those what you're paying for there is
3:05:54 they've taken the time to write all
3:05:56 those prompts and turn them into pretty
3:05:57 buttons
3:05:58 so I think they're good so I mean some
3:06:00 of them are quite good Jasper's
3:06:03 um quite good but that's that's what it
3:06:05 is from a technical perspective it's
3:06:06 it's actually quite a quite simple
3:06:10 um prompting layer on top of the same
3:06:12 tool that you have access to
3:06:15 um do I think AI is progressing too fast
3:06:17 possibly
3:06:20 yeah I think AI is progressing way too
3:06:23 fast but there's no stopping it so
3:06:25 uh it's it's coming way faster than
3:06:27 we're prepared for thanks so much for
3:06:28 your time I've learned a lot okay I'm
3:06:30 gonna go I appreciate all your time I've
3:06:33 been here I think three hours or
3:06:34 something like that uh my butt's numb uh
3:06:37 I'm lightheaded and I'm tired and sweaty
3:06:40 and my hair is unfortunate so uh thanks
3:06:44 all for your time I'll be back at some
3:06:45 point peace