aiLL 9/3/23 - New AI Tools & How to Use Them Today

Transcript

1. 0:29welcome everybody Dr J was shaking Matthew Craig official what's happening welcome welcome Happy Sunday I think it's Sunday
2. 0:42I've been working my this filter is way wrong I look I I look if you've ever been to like a newscast where you see
3. 0:51someone who is put in makeup this is what they look like when you run into them in the hallway hang on let me let
4. 0:57me change the effect oh wait that was not the right one let's see enhance enhanced effects no that's not right uh and hands
5. 1:08are enhanced filters let's say let's do this no no no no that's not too bad yeah that's good that kind of kind of more like skin
6. 1:24vintage yeah yeah yeah youth Twilight we'll go Twilight Twilight twine twine
7. 1:39it was a Jim Gaffigan joke got it gotta go down the hardware store to get twine [Laughter] oh welcome everybody
8. 1:53did you know there's a Santa Claus filter you can pretend to be Santa maybe if we get close to December I'll I'll go
9. 1:59I'll go all Santa so uh welcome everyone oh holy crap there's a lot of people here my name is Kyle Shannon this is the
10. 2:06AI learning lab uh here's a whole bunch of cool pictures that uh I made in ideogram if you don't know ideogram
11. 2:12there's uh some cool stuff there you can actually make images now with words on them words words words so what we do here
12. 2:23my name is Kyle Shannon I like talking about this generative AI stuff um we got some some some folks show up here
13. 2:32regularly they're called the Irregulars they're they're down in the comments they'll help at some point I'm going to
14. 2:37make them uh moderators uh I like to talk about all this uh generative AI stuff I'm an entrepreneur I've started
15. 2:44lots and lots of companies uh my current gig is I'm founder and CEO of a company called storyvine and um
16. 2:52in the mid 90s I co-founded a company called agency.com and I bring that up just because the early days of the world
17. 2:59wide web and the early early days of this generative AI stuff there's a lot of parallels it's feel it's feeling very
18. 3:05deja vu all over again so so we'll talk about that uh if you have questions and comments pop them in the comments below
19. 3:12I do my best to answer everything um the other thing that we'll do is we'll sort of kick off tonight talking
20. 3:19about if you're new to this AI stuff if you're just sick of people like you know you should try Ai and you're like I
21. 3:26don't want to try AI what is this AI anyway what is internet anyway um we'll walk you through that we'll get
22. 3:34you up to speed that's part of what this channel is about part of it's just about talking [ __ ] uh I like to make funny
23. 3:39voices uh I don't know we have fun we have a good time here so we'll hang out it's Sunday night you're not doing
24. 3:46anything else might as well hang out here right okay so if you're brand new to chat EBD thank you Matthias um
25. 3:53go there go to chat.openai.com that's the official chatgpt website go there and if you don't quite know what to do
26. 4:01there you can sort of ask it anything uh but you can start out with what I call parlor tricks have it write you a
27. 4:06country song about and describe your dog or a significant other or have it write you a marketing plan or just just tell
28. 4:14it to do something that seems random and like it couldn't do and watch what it does and then go to this thing
29. 4:21prompts.chat that's a document and that will teach you how to talk to these things these are called large language
30. 4:27models that one's chat EBT it's right now it's the best uh but there there are a lot of uh companies nipping at their heels right now
31. 4:37um but it's good I'll talk about these other ones in a bit um a couple other things to update I think now just to be clear I'm an old guy on
32. 4:49Tick Tock right so I'm figuring this [ __ ] out I think I got a store set up so that there's products so I don't know
33. 4:56what your interface looks like mine has a little button called shop on it and um there's a women's t-shirt a men's
34. 5:04t-shirt and a tote bag I just put them in there kind of as a test there's only one design right now uh it's uh
35. 5:13um I'm not weird I'm irregular so it's for the Irregulars so if the Irregulars are here if you would just go check the
36. 5:18store out and let me know like is it there does it look functional you don't have to buy anything right now if you
37. 5:24want to fine but but um so anyway that's there the other thing that I did if you go to the YouTube channel YouTube uh AI
38. 5:33learning lab Dash TT uh regular at AI learning lab was taken so the dash TT is for tick tock um that's archives of these videos I got
39. 5:44all the way through June is uploaded there and I actually figured out the prompting for generating chapters so I got in like I don't know
40. 5:5525 or 30 um descriptions and chapters in there so you can go see that um okay store looks great awesome terrific hey Kyle hey Irregulars hey
41. 6:04Stevo with shaking Man story looks great awesome yeah it's working fine okay good awesome um so I guess I know what I'm doing I
42. 6:11guess I guess now I need to figure out some products um I'm gonna do a you can make money with chat gbt huh ha
43. 6:22uh if you're wondering how to make money with you at gbt we can talk about that feel free to ask comments below
44. 6:28um okay let's see let's let's jump in and talk about this stuff okay so as I said before that's the official chat GPT website
45. 6:42the reason I have three of these here these tools right now a lot of these AI tools as sexy as they are and as
46. 6:49powerful as they are they're what I like to call it's a technical term janky pieces of [ __ ] um because we're very early in whatever
47. 6:59this why is that not zooming smoothly there we go um we're just very early so so even though machine learning and ai's been around
48. 7:11for a while this generative AI this this idea of chat GPT and things like it relatively new so a lot of the software is very very um
49. 7:19simple and doesn't do a lot of sophisticated stuff and and because of that you you kind of have to play with
50. 7:25different ones so they're good at different things so chat gbt is is um really good at writing so gpt4 in
51. 7:34particular if you pay the 20 bucks a month here you get gpt4 um you also get this thing called Advanced data analytics which is
52. 7:41mind-blowing you also get plug-ins and they're going to give you early access to new technology as they roll it out
53. 7:46they're they're the sort of um 800 pound gorilla right now others are coming but it really is the best Claude is a bunch of ex-openai people
54. 7:56that started this company called anthropic what's cool about Claude is it's got a hundred thousand token
55. 8:01context window which is about 75 000 Words which basically means this you could paste a 50 000 word novel into the
56. 8:09prompt and tell it to tell it to write you an epilogue for your novel it'll just do it um claude's pretty remarkable
57. 8:17um so so if I if I have big documents or I have something big I want to go through I use Claude um Bard is Google's entry into this
58. 8:24space they just made the top three um what are the criteria for the top three I don't know get a wild hair up my
59. 8:30ass and I decide what's in the top three right now these are the three that are the most useful for me um who's at the Kids Table kids table
60. 8:37still useful still worth playing with um and exploring Bing is Microsoft's uh large language model it's it's actually
61. 8:48built on top of gpt4 so it's it's like chat GPT but it's it's connected to the internet you can make images there you can
62. 8:57um analyze images there perplexity.ai is a really good research tool labs.perplexity lets you play with
63. 9:06metas llama 2 models and their new code llama model which in terms of coding actually just beat gpt4 so if you're having
64. 9:17frustrations with gpt4 with coding try out code llama then Pi dot AI they have an app and a website and it's what we
65. 9:26call here in the lab chatty GPT it will just talk your ear off so if you want to get work done don't use Pi but if you
66. 9:33don't have a conversation it's very very interesting and then poe.com's got a lot of bots and things like that that you
67. 9:39can play with I use Poe today because I got rate limited out of Claude because because I was doing three hour
68. 9:48YouTube or Tick Tock transcriptions YouTube transcriptions and I got booted out so I was using Poe to do it uh togas
69. 9:55thank you very much I appreciate that um so so Poe can be very very useful then finally the pictures the pictures
70. 10:01but how do we make pictures Kyle tell me how to make pictures I want to make pictures it's all about the pretty
71. 10:07pictures please please stop talking for the love of God shut up just shut up can I just shut it zip it all right
72. 10:19um Mid Journey's the best ideogram is eight days old maybe nine days old it does text this is a startup they raised
73. 10:2816.5 million dollars led by Andresen Horowitz it's an X Google I think it's the next Google person started that and they figured out how to
74. 10:40do text before the big players like stable diffusion and mid-journey mind-blowing and that's that's what's
75. 10:46going on in this business right now we don't know where the Innovation is going to come from and all these companies are
76. 10:52you know sort of fighting to one-up each other so we're the beneficiaries of that Leonardo and clip drop are both based on
77. 10:59top of uh stable diffusion Leonardo's just a really good website uh where you don't have to install your own [ __ ] and
78. 11:07clip drop is from stability AI those are the people that make stable diffusion there's some pretty cool tools there
79. 11:13and then prompt book will teach you how to talk to these things and and and sort of teach you how powerful they are and
80. 11:19then finally futurepedia is a big long list of uh what do you what do you call it um AI tools I know I never remember that
81. 11:28by the way for the Irregulars uh drinks tonight uh a cheap wine we bought at Safeway hold grocery store wine
82. 11:40oh yeah it's solid and then uh Rocky Mountain tap water filtered Rocky Mountain tap water all right so let's get to the questions
83. 11:52pop your questions in below and I will do my best to answer them I will get behind that's one of the things
84. 11:59I do I also cuss I make funny voices I don't know who knows who knows let's see what we got I'm hesitant wait um
85. 12:09let me flip this around let's get let's get things set up here all right all right yeah my lactators um all right I'm headed hesitant to
86. 12:25discredit anyone because I lost more money in cryptocurrencies than I care to admit story looks great
87. 12:31um have you heard of Kit or pay hip for cheap e-commerce store and easy setup I have not I'm using printful right now
88. 12:39because I want to do print on demand because I don't want to I like I just I can't have the inventory I'm not I'm not
89. 12:46running that business but I thought it would be cool to have something there um so right now I'm using printful but
90. 12:52if I if I decide to ever sell these things I gotta figure that [ __ ] out so at some point I'll figure it out
91. 13:02um all right oops sent twice that's okay Kyle in a regular store looks great hey Marie deal shaking Matthias
92. 13:11ptld as Portland I suppose uh this stuff reminds me of Lincoln Logs Erector Set Legos Radio Shack 64. uh in one chemistry set
93. 13:22what stuff reminds you of it oh like all the AI tools right now they're totally like that they're totally just janky ass
94. 13:28you gotta glue them together and it's they're the most like um what was the Erector Sets right yeah the Erector Sets
95. 13:36because because they have the little nuts and bolts in the little cheap wrench and when it would slip you would
96. 13:42cut your knuckle open on the Erector Set that's what using AI is like right now you're like oh this is cool that sucks
97. 13:56oh good lord lord lord lord hearing truth what's happening Emilio's wife was shaking please folks most of you
98. 14:04probably know but don't let anyone guide or help to buy crypto yeah so so we had a scammer so the AI
99. 14:13learning lab had someone that copied the channel and then was trying to sell members crypto stuff if anyone is
100. 14:19pretending to be me or anyone Reach Out reaches out trying to sell you crypto [ __ ] or even just getting you to log in
101. 14:25with your crypto wallet to their stupid [ __ ] site don't do it um so I don't I don't know if that's going on again but it but yes what
102. 14:33Winston says just be very careful it just crypto stuff is if if you think AI tools are jankier than [ __ ] crypto tools are
103. 14:43really [ __ ] special you have to be very technically Savvy and even then you know I know I know of a lot of people that
104. 14:55had a lot of money just vacuumed out of their accounts because they clicked on one link and you're like but Kyle it shouldn't be
105. 15:04that easy to steal people's money I agree and what the crypto Bros will say is not your wait not your keys not your crypto
106. 15:13um yeah so there's something yeah anyway just anyone tries to sell you crypto [ __ ] do not do it it is not the road to riches well except
107. 15:25for them it is it's the road to riches for them all right let's see what questions do you have Hi boss how are you today well
108. 15:34Johnson royalty I'm doing well I'm doing well I was busy all day I was downloading videos uploading videos
109. 15:41using po to Poe and Claude to write YouTube descriptions with chapter things that was pretty interesting I'll talk about that in a bit
110. 15:52um then I was working on the e-commerce stores uh store well just whatever all that [ __ ] so I was busy all day today and then I mowed
111. 16:01and then I went to Costco because I know how to live and then what else uh I don't know watch the little US Open
112. 16:10tennis Djokovic not so dramatic today one in straight sets three Americans are in the quarterfinals so that's cool
113. 16:26someone approached me today with what first I thought was a helpful tip and then he offered to help guide
114. 16:33yeah Winston sorry about that that's probably the scammer got a bunch of people's um you know handles even though they
115. 16:41deleted his account he's probably just got a spreadsheet that he's working his way down so yeah anyone wants to guide
116. 16:49you to crypto tell them to [ __ ] off I will not do that if we're gonna do a pyramid in scheme in a pyramid scheme in
117. 16:58this place I'm going to call it a pyramid scheme will be transparent all that stuff pisses me off all right but thank you for sharing that
118. 17:11that happened to you Winston and and for letting people know yeah exactly um how welcome three times is the number
119. 17:19of times I subscribed in six seconds with my mini phone I I I hope you didn't subscribe three times uh
120. 17:35picant but not too sassy um what do you know about DMSO I forget what DMSO is hang on
121. 18:12dimethyl sulfoxide no I don't I assume that's not what you're talking about um I don't know what you mean uh
122. 18:20explaining it and I'll I'll see if I know anything about it I don't think I know anything about it how meaning ho ho
123. 18:25ho ho welcome Hey regular share bear with shaking okay farmer is here what AI can create a metaverse well there's um
124. 18:38what the hell is it called something labs hold 58 year old brain trying to remember something oh please
125. 18:53Gen X meltdown um
126. 19:10dream Fusion blockade Labs that's what it is hi box all right so so this isn't exactly creating a metaverse but this is pretty
127. 19:21[ __ ] cool so I'll show you this I have read and I accept your terms by the way oh you wanna you wanna see
128. 19:27something fun so I went to so where I am is I'm at Skyblock skybox.blockadelabs.com and then they've got a what's new since you last logged
129. 19:36in and they have a I have read and accepted the terms of service and the privacy policy I I am not a manual reader I have never
130. 19:44in my life read terms of service and I assume most people are like that some people do I don't um you know what I do now I go to that site
131. 19:55and so there's the terms of service and I'm just going to copy them all I'm gonna copy them I'm going to go over to Poe
132. 20:10I'm going to say new conversation so pose just a large language model but I've got the Claude 2 100 000 100 000 token so I can just paste
133. 20:19that whole damn thing in there so let's see um [Music] um based on these terms of service
134. 20:35colon paste there's all the terms of service please give me a concise bulleted list of what I am
135. 20:53agreeing to that is material in your POV keep each bullet
136. 21:08two seven to ten words all right feed this thing for us not that I really give a [ __ ] I'm still gonna accept it but this is just kind of
137. 21:23a cool use it's one of the things that these tools are good at is understanding uh text and reading it
138. 21:32you own any original assets created with the service blockade can use and share your content for promotion so you own
139. 21:39them but you give them the rights to play with them you must be 18 years of old fees are charged monthly or yearly
140. 21:45and are non-refundable so don't do the yearly thing in case something better comes along Services can be terminated or suspended
141. 21:54at any time you indemnify blockade against any losses from your use blockades liability is limited to 50 bucks or fees paid
142. 22:04um disputes must be resolved in individual arbitration class action waiver must uh arbitrate claims individually blockade can change terms
143. 22:13at any time so it doesn't seem too egregious there you go bang so so that's a fun thing to do um all right so we're going to accept those
144. 22:23and I do not want to show this message again all right so so we have a cool 3D World here right
145. 22:38well what you can do at blockade Labs is you can just say started you can just say um let's see dystopian
146. 22:53hellscape in a water logged um dry lake bed [Laughter]
147. 23:10um um elements from a um hippie Festival
148. 23:24are strewn about and then we're gonna do let's see techno Arts I will we'll do realistic we'll see we'll do realistic all right
149. 23:37so we're going to generate so this thing will now generate this 3D World so this isn't exactly generating a metaverse but
150. 23:43this is this is the beginning of the kinds of things that we're going to be able to do where you're going to
151. 23:49literally be able to just talk something into existence um I know that um that there's some folks working on
152. 24:00um rolling these kind of things into well that's boring as [ __ ] but you know gives us what we want so
153. 24:08let's do let's do something else we'll do um [Music] infinite Cosmos full of exploding
154. 24:24planets and spaceships and then we'll do [Music]
155. 24:38Tech Noir cyberpunk will do that this one should be more interesting um so what what blockade Labs actually is
156. 24:47is this is just creating a kind of a flat image that is is done in such a way that they can map it to a sphere
157. 24:55um so so it makes this thing that looks like you're in a 3D space but you're actually not and then there's our cool spaceship
158. 25:03looking thing with exploding planets better give me an exploding planet nah no exploding planets I want an exploding planet
159. 25:17I I want AI to be perfect it's not but anyway still pretty flipping cool so this is skybox.blockadelabs.com um
160. 25:30you can take these and put them into Unreal Engine and put objects in them and then you're able to navigate them
161. 25:37right now it's not super sophisticated it's not creating a 3D world but I know that folks are working right now on
162. 25:44being able to put together 3D what are they called procedural like procedural scenes within Unreal Engine so so this stuff's going to get
163. 25:56very real very fast so blockade Labs is one to play with the other thing that I would do is I would go to because I'm not really keeping up
164. 26:05on this stuff because I can't go to futurepedia.io and look up text to 3D or just look up 3D or worlds just find a
165. 26:14category in there that looks like the kind of [ __ ] you're looking for and uh yeah there you have it but pretty cool isn't it
166. 26:23pretty cool pretty cool you know what I mean it's all right it's all right it's all right it's all right right right right
167. 26:31all right let's see how can we use chat EBT to make 400 returns in the stock market uh have it teach you how to do technical analysis
168. 26:42and start doing day trading and pray you don't lose your ass when you start leveraging your your trades
169. 26:51don't listen to me I have a degree in acting buddy you know life is tough it's tougher when you're stupid all right see I'm getting
170. 26:59better I'm getting I'm not getting quite so thrown when someone throws the damn hat on me all right what did you do with the chat for coder
171. 27:11huh what did I do with it like why is it screwed up and not as good as it was I didn't do [ __ ] with it I don't work for open AI
172. 27:22um I haven't noticed it's gotten worse but I haven't been uh playing with code code with it for all day every day like
173. 27:30some people are and saying it's getting worse but I don't know I don't know what's going on with it um what I can tell you is that these
174. 27:38tools all of these tools are they're kind of almost all of them they're flying the plane as they're building it so when
175. 27:54[ __ ] just changes it's cause [ __ ] just changed because they had to deal with a lawsuit or they had to deal with
176. 27:59something broke or they had to deal with you know whatever might be happening crypto Joe just joined welcome crypto Joe um
177. 28:09please do not get any help from anyone with cryptos online agree with Winston again uh bakala gift shop it's annoying people
178. 28:16are scamming and pretending to be others than themselves it's very annoying vodka gift shop very annoying but we did have
179. 28:24that we did get rid of they had a site that cloned all of my videos and looked exactly like mine and then they would
180. 28:31hang out on lives and when I would shut down my lives they would reach out to people that were on the live and say hey
181. 28:37it's Kyle from the live [ __ ] [ __ ] so I think I gotta turn the air conditioner back on it's getting a little moist in
182. 28:46here if you know what I mean uh most interesting AI stocks um how's the wine it's okay it's called Seven Deadly zins so I suppose it's seven
183. 28:58zinfandels it's fine isn't it I'm not too sophisticated when it comes to wine so it's not awful um
184. 29:13like I said I have a degree in acting so I I can't tell you what stocks to pay attention to I'll just tell you things
185. 29:19that I think are interesting right now um and and some of these are private companies so you can't even buy stocks anyway um
186. 29:28I think in videos fascinating I mean I don't think I don't think their stock is at a discount they should a trillion dollar market cap
187. 29:38uh and I forget they had a hundred 101 percent year-over-year earnings so that's interesting um Nvidia seems to have successfully
188. 29:49cornered the [ __ ] Market on AI chips they were doing game chips and then the crypto Bros figured out it
189. 29:59was good for doing blockchain [ __ ] and then they were doing all that and I think all the other chip manufacturers
190. 30:04are just like [ __ ] it and they just backed out and then and then you know AI was like oh this is really good for AI stuff and
191. 30:11what what happened with when the Transformer came out that Google invented in 2017 it became clear that the more more data
192. 30:20you threw at the at these tools and the more gpus you threw at them the better they got so companies like open AI just
193. 30:27said [ __ ] it let's get as many gpus as we can and throw as much data at us as we can that's why they cut a deal with
194. 30:33Microsoft because they have racks and racks of servers with these big gpus in them um and so they're interesting
195. 30:42um about a week ago people were coming in the thing going well you think open AI is going bankrupt there was an
196. 30:48article saying they're going bankrupt they're spending the seven hundred thousand dollars a day and I was like
197. 30:52that's a lot of money but no I don't think they're going bankrupt well they just came out with an earnings
198. 30:56report that they just they just uh hit a billion dollar billion dollar annual uh run rate for Revenue billion seven
199. 31:08hundred thousand dollars a day is about 250 million in expenses billion dollars in Revenue I think they're gonna be okay
200. 31:16now they're probably plowing a lot of that into um other [ __ ] like gpt5 and dealing with lawsuits and trying to cut deals with
201. 31:28the New York Times so they don't get their ass sued by them after Google just cut a hundred million dollar deal with
202. 31:33the New York Times you know there's stuff going on so it's not like they're rolling in cash over there but a billion
203. 31:38dollar run rate ain't bad for for uh a company that's spending 250 million dollars a year um all right let's see break waitbrifacki.com
204. 31:55beef oh wait beef be fake ai.com all right I'll go look Matthew Craig I'll bite let's go look be fake is this
205. 32:06some deep fake ai.com is this some deep deep fake sort of thing be honest we're not always our creative
206. 32:20best and not every moment is ready to share so give your reality an AI lift [Music] be fake is an effortless and fun
207. 32:28introduction to AI that allows you to transfer yeah that's like it's like lens uh and it's like the the new one that
208. 32:34people are doing headshots with where you upload your photos so it's basically doing like a Laura model in stable
209. 32:40diffusion that's probably what it's built on they've got some sort of you know uh dream Booth like Tech on the back end that's training on
210. 32:51how to do that that's pretty cool um there's going to be a ton of those I'm assuming the wine is good it's not
211. 32:57bad Costco oh dang it dang it Jim Tick Tock just jumped me to the bottom uh yes yeah masako
212. 33:12um let's see um if you're the first time here put one in
213. 33:27the comments if you've been here before two and a regulars put three oh cool I like that Matthew Craig this stuff reminds me oh okay we did
214. 33:39that before [Music] now I went too far back AI can create the metaverse what did I do with chat coder
215. 33:54how's the wine I assume the wine is good Costco on the weekend Kyle what's wrong with you I'm an idiot you know that
216. 34:07I'm not the smartest seller you know when I go to the Costco I like going to the Costco we go there I take the uh I
217. 34:15got the Ford Flex it's like a hearse but more like a station wagon it's like a cross between station wagon and her so
218. 34:21take the Ford Flex and I go there because you know when you need 16 jars of 43 pound peanut butter you go to
219. 34:29Costco right and then I go there but then every now and again I just walking down the aisle I'm like yeah I need a 762 piece wrench set
220. 34:39have you ever been to Costco and not spend 400 that place is just [ __ ] wrong any AI tool that can help me
221. 34:48learning I.T networking yes all of them [Laughter] all of them um chat GPT is good at it you might want to go to
222. 35:05bard.google.com because it's connected to the internet it's also pretty good at coding um and just tell it that you want to learn um
223. 35:17you want to learn networking uh it networking and especially the ones that are connected to the internet Bard Bing and Pi Pi dot AI
224. 35:29so Bing and Pi are also connected to the internet so what you can do with those is you can say hey go find the latest you know an
225. 35:38article a blog post or or documentation on it networking have it design you a custom course tell it what
226. 35:47you know tell it what you don't know tell it um that you're looking for resources it might also be interesting to go to
227. 35:54perplexity.ai and see what it can find just go to these things and start asking them what what will actually kind of
228. 36:00help help you get your head around how these things work go to prompts.chat that will talk to you this is just a
229. 36:08document that that's a kind of a primer on how to talk to these large language models most of that document is telling
230. 36:15it to act like a certain kind of persona so act like a screenwriter act like a a consultant act like an I.T professional
231. 36:24right an I.T networking specialist uh and and when you tell it to do that it gets better at doing that kind of
232. 36:31stuff so go there and kind of learn how these things work and basically just tell it to tell it what you're trying to
233. 36:37learn and have it create a custom uh curriculum for you do you want to learn it in three in four weeks four months four days
234. 36:45how deep do you want to go do you want to specialize in any area of it do you not even know what's possible have it
235. 36:51tell you what's possible and say oh I'm interested in this and I'm interested in that tell it you know are you really
236. 36:57good at project management or are you good at ideation or are you good at you know sitting there and sort of knocking
237. 37:03out just you know technical um you know things technical drudgery some people are really good at that if you're good at that
238. 37:11tell this thing that and then say what jobs within that would I be good at and it will tell you and then design a course for you to do it so
239. 37:22um any good in video dot IO competitors they only give me 10 minutes for free per week all the video tools are really
240. 37:31expensive so let me go look at it in video I haven't played with it in video dot IO is that right
241. 37:45create publish worthy videos oh this is the one that does like stock stock video assembly you write the script and then
242. 37:53and then it cuts together all the stock stock video stuff um [Music] there's videos all these things are expensive there's video
243. 38:15um there's the script d d e s c r i p t d script dot descript AI what's the URL
244. 38:31descript.com um again what I would do I don't because I'm not following this stuff on a day-to-day basis I would go to
245. 38:42futurepedia.io and look up their video tools um they're decently well organized and I'm I guarantee you there's you know 20 or
246. 38:5330 tools out there from what I can gather all of the video tools are decently expensive right now because I
247. 38:59think the compute time you know if you figure what they're doing is is some version of you know 30 or 60 frames a second and if they're
248. 39:07generating it even if they're just editing it it's it's it's probably decently pricey from an API standpoint
249. 39:14but go to futurepedia.io and that'll help all right hang on I gotta go turn on the air I'll be right back
250. 39:30your live is different than your videos gosh because I'm live yeah my videos are like hey it's Kyle Shannon with the AI
251. 39:38learning lab here's some [ __ ] I learned and I try to package it in a tight little thing here I'm just rambling this
252. 39:42is just more like a conversation um I'm figuring these lives out kind of on the Fly and you know what I want it to
253. 39:50be what the community wants it to be where's where the value is so yeah sometimes they're fun sometimes they're
254. 39:56serious sometimes I don't think they're ever serious but sometimes I get a little I get a little soap boxy I get on
255. 40:02my soap box and I go I just go nuts I go nuts I go mwah so yeah I could probably act more like an idiot on my videos but I don't know
256. 40:13it's just they started as a different thing foreign you see the Deep fake DeSantis yikes AI I have not seen it
257. 40:34to Santa's campaign uses deep fake images to attack Trump what's depict of Ron desantis's Michael Scott that one same leadership is about to
258. 40:50fake leadership is about fooling the voters it's about book deals and globalism Leadership is about getting ahead no
259. 41:02matter who you stab in the back in my America yeah I mean it sounds very monotone it sounds very synthesized here's the thing um
260. 41:11that's kind of a crappy deep fake but they're going to get better and better and better I think the 24 election is
261. 41:18going to be absolute Insanity with with deep fakes and misinformation so um I I think I think we are just scratching
262. 41:27the surface in terms of what's coming it's gonna be it's gonna be something else it's gonna be something else special
263. 41:36gonna be something else special is there currently an AI tool that can create a metaverse we did that one
264. 41:42just try to buy a t-shirt by clicking on the laptop in the little window uh okay so yeah so uh for any Irregulars here I
265. 41:55I created three products and they're in a tick tock shop they're in the tick tock shop here there's a it's it's the it's this graphic it's
266. 42:06that graphic I'm not weird I'm a regular so the people that are regulars here they've Nick they've they've dubbed themselves the Irregulars
267. 42:15um so I turned that into a t-shirt so there's a women's t-shirt a men's t-shirt and then a tote bag um you can go check them out it's it's
268. 42:22uh I I just got it set up so I I assume it works I hope it works I'm not sure if it works so please let me know um
269. 42:30yeah the the uh yeah so so we'll keep going all right blink AI video editing good glad you're helping each other
270. 42:37because I don't know those tools very well um what AI can create an app well there there are a lot I would go again I
271. 42:48would I'll point you to futurepedia.io there's a lot of app building tools it depends what you want to do
272. 42:56there there are some that build like you know site maps and UI um there are other things that build um ux and some code there are some that
273. 43:07do code there are some that purport to do it all um I think a lot of them are kind of janky um zapier has AI built into it so you can
274. 43:21make zapier automations by just describing them and it will put together um an automation for you you still have
275. 43:27to you know put in your information for all the calls um and and zapier also has a new beta of something called interfaces where you
276. 43:37can create you know like a website application um hmm bubble.io is is kind of like zapier and you can make apps with it but bubble dot
277. 43:52IO is complicated enough that it's you're almost learning you're almost doing programming so I assume what you're looking for is
278. 43:59something that you can just say hey I want an app that does XYZ um from what I've seen all of those things right now are not quite ready for
279. 44:06prime time so I'd go to futurepedia.io look up app development or programming and just go start playing see if see if
280. 44:14the any of them are any good A lot of those tools are super sexy for the demo and you can like get something close but
281. 44:22then when you try to actually build something functional for yourself they kind of fall on their face because it's complicated
282. 44:27to do good apps but keep exploring those and if anyone finds anything good that's a good AI like text to application Builder let me know
283. 44:36because that's an area that I'm interested in as well stable video new generative AI text to video Stable videos it from stability
284. 44:47Matthew Craig bring in the uh bringing the good stuff um stable video
285. 45:16the decoder this new paper
286. 45:34stable video is a research project exploring generative AI to edit existing videos frame by frame rather than generating videos from scratch
287. 45:42okay interesting so it looks like in this example they have a boat and they can change the boat and it's sort of
288. 45:54modifies objects within within a video if that's the one you're talking about yeah see I've seen that before but I
289. 46:01haven't I've not played with that one I'm like I'm like useless tonight everything everyone's asking me like I
290. 46:07don't know I don't know you're just gonna have to go research that on your own just go to the internet and just go
291. 46:13to the internet and it'll be there on the internet somewhere riverside.fm transcriptions very cool is
292. 46:21there a long wait for code llama too no there's no way it's open source and free right now go play Abdul and if and if
293. 46:31you if you want to just go play with it for free without having to install anything go to labs.perplexity.ai
294. 46:43and that's perplexity.ai they're hosting um they're hosting llama2 and code code llama so you can just go there right now and
295. 46:56play for free now if you want to install it and do do all that sort of stuff you'll do that on your own
296. 47:02um do you know do you know AI is just an API call 95 of the time I do know that [ __ ] squid that's what we call we so most of
297. 47:13most of futurepedia.io is what I call wrapper apps which are basically just what you said it's someone's created a clever prompt and they
298. 47:22you know dump it off to a stability Ai call or a um an open Ai call yeah most of them are that it doesn't mean they're not of
299. 47:33value right it doesn't mean they're not a value if someone's figured out an interesting way to fine-tune the model
300. 47:39to give you really good legal analysis of terms of service documents then you want to pay five bucks a month
301. 47:47for that or ten bucks a month for that more power to you but yeah a lot of the apps that are out right now this is very
302. 47:54very very very very very very very very similar to the early days of the World Wide Web where you know a new kind of piece of
303. 48:02technology would come out people would pounce on it they'd make little micro websites to do certain things and then
304. 48:08they try to charge a lot of money for them and then some real company would come along and that would just become a
305. 48:13feature of their software gonna be the same [ __ ] with this stuff so yeah that's very much what a lot of these things are right now so
306. 48:20um and that's why it's one of the reasons that I have these things on screen for most of my lives like these are the base models right
307. 48:29Chachi PT Claude bard um Bing perplexity Pi perplexities you know I think they're doing uh open AI calls but they're doing some
308. 48:41interesting stuff with research and citations but Pi is inflection Poe's got their own model but they're using other people's models so yeah
309. 48:49most of the stuff I have here is kind of core stuff and that's that's why I have them here I think this early in the industry it's
310. 48:58important to learn to use these base models rather than just sort of a point-and-click tool that someone pre-prompted something for you
311. 49:05that was cool solid content thanks for the education you're welcome Ron Mage dot space text to image with sdxl very
312. 49:13cool yeah I've seen that before what AI was that the AI with the images was uh skybox.blockadelabs.com
313. 49:22dystopian hellscape what did I stumble upon camcatkin camcat can hear weather on the sevens I'm Kim catkin all right
314. 49:37you're a funny guy Kyle thank you very much girl mail sing blockade um add negative prompts what app or website is that that was
315. 49:50blockade Labs offer to help Okay um see what is the difference between embedding and fine-tuning on open AI you
316. 50:01asked this the other night Tech audit TV so so and fine tuning is just if if you okay
317. 50:16so when you come to let me just make sure I don't have this turned on right now I mean I do have it turned on Save
318. 50:24okay so when you come to chat GPT and you type in um I want to know how to do it networking it says certainly networking is a broad
319. 50:40topic um it can refer to both computer networking and social professional networking then it explains the stuff so
320. 50:50it's generally helpful it's generally nice and if I say um you misunderstood I meant networking like flying a spaceship
321. 51:07let's see if it apologizes I apologize I misunderstood um it sounds like you're interested in learning about the skills and knowledge
322. 51:16required to operate and navigate a spaceship in a space-faring context so Chachi BT is fine-tuned to be helpful
323. 51:24and so if you tell it it got something wrong it's like oh I'm sorry and it'll try to get it right and it'll try to do that Within
324. 51:32um chat gbt they have now added this new thing called custom instructions and so you can fine tune chat gbt to act
325. 51:42like whatever you wanted to act like and sounds like fireworks are going off here I hope uh um oh and I lost my double camera that was
326. 51:57weird okay I'll get it back um so so what I'm going to say is um
327. 52:13you're an overbearing pompous ass who uses large words that even you don't understand as a way to intimidate those that you talk to period
328. 52:28okay so I'm gonna I'm gonna do that and then I'm gonna say whenever anyone asks you anything I want you to use an obscure literary reference
329. 52:44to belittle their idea and then redirect the conversation to talk about yourself all right and so I'm going to save that
330. 52:59so now so what I've just done is I've just fine-tuned chat gbt to be a compass ass so I'm going to say I
331. 53:07want to learn about computer networking ah a commendable aspiration though one wonders if you're prepared to Traverse
332. 53:20the Labyrinth of Court Labyrinthian corridors of networking intricacies it brings to mind the words of T.S Eliot we
333. 53:28shall not cease from exploration in the end of our exploring will be to arrive where we started and know the place for
334. 53:35the first time but I digress as for computer networking the vast expanse of protocols architectures and topologies
335. 53:42awaits your inquisitive mind now speaking of intellectual Pursuits have you ever pondered the allegorical significance of the protagonist's
336. 53:51journey in James Joyce's Ulysses it's a rather captivating subject much like the networks that interconnect our digital
337. 53:57Realms right so I fine-tuned Chachi BT To Be A pompous ass that's fine-tuning um hang on let me how do I get this back
338. 54:08oh I know how I get it back I go enhance yes dual camera do that boom there I am yay hi hi hi thank you Kenny hey Kyle Kyle what are
339. 54:23you doing Kyle um embedding embedding is where you take lots of data and you essentially the the simplest way
340. 54:35to put it is you gptize it you you vectorize it so you you take some data you embed it which turns it into it it
341. 54:44puts the words into semantic clusters in a vector space there you go it just turns them into math you know life's
342. 54:52tough it's tougher when you're stupid um I'm totally Pavlov's dog um so so think of embedding as as kind of larger scale um
343. 55:05it's not really fine tuning it's a different thing it's embedding is what's going to happen when chat GPT and all these companies
344. 55:14Microsoft and Google make it really easy to take your company's data and make it gptable that's that's embedding where
345. 55:21you're where you're getting large data sets into this um crazy ass Vector space that makes all this stuff possible so I hope that helps
346. 55:30and then you just learned about custom instructions in chat GPT and you can make it a pompous ass so you
347. 55:37could make it a coach life coach lives tough it's tougher when you're stupid ah someone give you two hats in a row so
348. 55:45I could finish that that's tough definitely stupid thank you Stevo Cena y'all y'all are just y'all just picking
349. 55:54on me now you just break yourself [Laughter] hey thank you though for the hats hats are nice they were swell they're very nice I like taters
350. 56:11well thank you all right let's see uh Irregulars are up to 14 800 on the likes [Laughter] uh dystopia is like Twilight Zone
351. 56:24parallel world when I click too fast on the iPhone Mini um code interpreter is data analytics now yeah exactly Beth they um
352. 56:36I when they first came out with code interpreter I'm like this is a remarkable piece of technology and it's a horrible name
353. 56:42and so they fixed it which is good it's now Advanced data analytics it doesn't exactly roll off the tongue but it at least says what it is
354. 56:54you know I would have called it data analyst in a box but you know that's probably Politically Incorrect or something
355. 57:01I'm a data analyst you're not gonna put me in a box why would you say that I feel offended I'm triggered can you explain like I'm five how
356. 57:16chatgpt answers your prompt versus a Google search yes Bob I can I have a I have a chart you want you want me to you
357. 57:23wanna you wanna do charts and graphs um I can explain exactly how it works well I can explain my understanding of how it works I'm not
358. 57:33a mathematician I'm not a data scientist I'm I'm a business guy that likes geeking out on new technologies I think it's really important
359. 57:41for me to understand how these things work because I want to understand what they're good at what they're not good at
360. 57:47I want to understand why they lie I know now why they lie because I understand how it works um so I'm going to do it in two ways I'm
361. 57:54gonna I'm gonna explain it first just how it's different and then I'm gonna show you actually how it works
362. 58:01which I find mind blowing when you see what these things do and then you actually understand how they're doing
363. 58:07what they're doing it's [ __ ] more remarkable that they do what they do did anyone figure this out in the first place
364. 58:15so Google has indexed all of the internet right they've got these Bots that go out and they say this website's
365. 58:20about this that and the other right and then there's this thing called search engine optimization where you can trick
366. 58:25the Bots to rank your things and so when you search for something in Google they want you to give a nice concise you know
367. 58:31just tell us just the facts what do you want I want to find out about minimalist running shoes and then it basically goes
368. 58:39into its index and it says based on you know all of this knowledge that I have here's a list of links what these things do is
369. 58:48they've been trained on as much of the public internet that in the case of chat GPT open AI could get their hands on
370. 58:57so so so think I don't know whatever eighty percent of the public internet for the past 80 years it might not be
371. 59:03that high it might be higher but just a massive amount of public web pages and public data sources we're we're trained into them and the
372. 59:16way they do is they break all of those things down and all of the words get grouped into these semantic clusters so
373. 59:23you might have the word dog and it lives in one area where it means a pet and it lives in another area where it means um
374. 59:31like uh the guy that did his girl wrong is a dog right that that and and so when you type your prompt it creates a
375. 59:40probability weighting of what the next word should be um and so it's writing the GPT in chat GPT the G stands for generative so it's
376. 59:51actually generating original content one word at a time based on probability so when you ask it to write you an original
377. 59:58country song it's actually writing you an original country song live video violation you have not been promoting wait
378. 1:00:11hang on I'm just learning this stuff you have not been promoting products for a while during your life we encourage you
379. 1:00:17to introduce and promote linked products in your life to keep the audience engaged and interested all right well if I have to do that then
380. 1:00:25I'll probably turn this [ __ ] off confirm okay um I have a store if you click on it I got stuff you can buy it's t-shirts that
381. 1:00:34say I'm not weird I'm a regular they're Tick Tock hope that helps ah capitalism um so
382. 1:00:49so one of the misconceptions about large language models is They're copying and pasting content from different documents
383. 1:00:56and presenting that to you it's not how it works so I'm gonna actually show you how it works it's fascinating I think um
384. 1:01:07what's going on here do I not have I don't do
385. 1:01:29okay Okay so
386. 1:01:47how chat GPT works okay you can see that right okay so like I said GPT stands for um generative pre-trained Transformer
387. 1:01:58generative means it's generating stuff pre-trained means it's been pre-trained on all the internet you don't need to bring your own data
388. 1:02:05and Transformers the technology that allows it to be trained on massive data sets so Google invented the Transformer
389. 1:02:12and everyone else is using it where all the beneficiaries of that technology if you sometimes will hear the term llm
390. 1:02:18large language model that's also what these things are called um I guess GPT is the name for it but large language model is the category and
391. 1:02:27it's basically a large language model they're getting all the words and all the different meanings and they're making a model out of that um
392. 1:02:35you will sometimes hear people talk about the latent space and so the latent space is a mathematical space so what
393. 1:02:44I've got here is a dramatically oversimplified version of visualization of the latent space this is just a two-dimensional
394. 1:02:52um graph of different clusters all these dots represent a vector and a vector is a string of numbers that place
395. 1:03:01that Vector in a particular space on this x y coordinate the real Vector spaces are like a thousand Dimensions this is two
396. 1:03:11Dimensions so I can't quite get my head around what a thousand dimension dimensional Vector space might look like
397. 1:03:19but it doesn't [ __ ] matter it's more complicated than any of us can understand so here's a two-dimensional version
398. 1:03:26each Little Dot represents a new a number a series of numbers that that give it a point in space and then these clusters are semantic
399. 1:03:35clusters so so you might have you know different versions of the word dog living in these you know different
400. 1:03:41clusters and this is all based on it looks at all these documents and it figures out the meaning of words as they
401. 1:03:47relate to other words in all these different documents um so that's what's sitting up in the in The Ether somewhere and then the way
402. 1:03:55these large language models work is that based on your prompts the large language model predicts the next word that that
403. 1:04:02it should respond with and it does that over and over and over again so I'm going to do a really simple example if
404. 1:04:08the prompt is the quick brown fox jumps over the lazy blank we as humans likely know that that word is dog right
405. 1:04:16it doesn't know that all it knows is based on what you type there there's some mathematical probability that that
406. 1:04:24somewhere in this Vector space some of these vectors have a higher probability of others in being the right word so you
407. 1:04:33know these little dots might all be animals right and and so within those vectors there's one of those vectors
408. 1:04:40that has the highest probability and if you look at what that actually is it's actually just this long string of
409. 1:04:46numbers that is its point in Vector space and then that longest string of numbers is associated with the word dog
410. 1:04:53and then the large language model puts the word dog into that sentence and so when you see large language models
411. 1:05:01writing you you know a full letter of recommendation when you say could you write me a letter of recommendation for
412. 1:05:07this job that I want to help my friend get and it goes it's doing it essentially one word at a time with mathematical probability it
413. 1:05:17has zero zero zero comprehension of what it wrote and this is one of the reasons people get confused by this because when it
414. 1:05:29writes it looks like it's a sentient being but it's actually it's got the sentience of a calculator right it's just doing
415. 1:05:37this probability game and and so that's one of the reasons it's confusing this is also one of the reasons that it does
416. 1:05:43what's called hallucinating why does it lie to you all the time because it's it's writing you know these probability
417. 1:05:50sequences one after another after another it doesn't actually know what's in there and so it'll say you know
418. 1:05:57it'll write an article and you're like great write me some citations and links to that and it'll go sure it wants to
419. 1:06:02please you so so it just it writes citations they may not actually exist because it's doing it one word at a time
420. 1:06:09and that's why sometimes it's just got stuff that doesn't resemble reality um the other thing that's important to
421. 1:06:16know about this this latent space this this Vector space one of the reasons that um one of the reasons that
422. 1:06:27telling chat gbt to act like a certain thing works right if you tell chat gbt I want you to act like a Stanford mathematics professor
423. 1:06:37is what it would effectively end up ends up happening is if you know this sort of semantic space right here is where
424. 1:06:45you know Stanford mathematics professors words are it'll kind of filter out all the other [ __ ] so you know out here might be
425. 1:06:54um students that that did the math problem wrong and you know uh reporters who wrote about the math problem and
426. 1:07:01described it out here and but but it's it's really in here where the the math professor's work is so if you say I want
427. 1:07:07you to act like a Stanford professor and I want you to help me understand this problem it'll do it better if it's got a
428. 1:07:15narrower semantic space that it's that it's querying basically so that was a lot I know listen it was a lot it was a lot for me it's a lot
429. 1:07:25there's a lot I know I know it's very is this stuff is I like I I mean not for nothing but is it not like
430. 1:07:38[ __ ] miraculous that someone figured out how to do that and I mean it's been decades in the planet right like people
431. 1:07:45have been figuring this out for a while the the Transformer was invented in 2017 and you know they did GPT one first and
432. 1:07:52then they did gpt2 and then they did gbt3 and that was the one that made you know a little bit of headlines that this
433. 1:07:59thing is starting to act like a powerful tool and then it was kind of a year year and a half after that that you know they
434. 1:08:07started getting their [ __ ] together and GPT 3.5 came out and then chat GPT came out or you know something in that neighborhood
435. 1:08:17um so the thing that we see right now that we're like oh this is this brand new thing they've actually been figuring out for a while but uh
436. 1:08:25I I just find it mind-blowing that that's what's happening and then what it does just feels like [ __ ] Magic
437. 1:08:33it's just like magic so I hope that helps Bob does that help does that answer um here's the deal that took me a long time I'm more than
438. 1:08:4199 messages behind I'm gonna jump to the bottom I'm probably gonna skip a bunch of questions right after Bob
439. 1:08:48Hamilton's question about how does this [ __ ] work s so I'm gonna just jump to the bottom and pick a random spot here and start
440. 1:08:55answering questions again uh to share bear tomorrow 9 p.m shark we're gonna hit you with Hats giddy up oh cool
441. 1:09:08oh oh oh that's that's why you had all the hats okay I get it I see what's happening now I'm understanding
442. 1:09:15blockade Labs wants 576 dollars a year build annually for unlimited use and it's in my opinion it's not ready for prime time yet so
443. 1:09:25my my advice on all of these tools all of them all of them not that one hang on whether it's open AI or Claude or Bard
444. 1:09:38or all of them or blockade Labs or Runway ml I don't know if it's my advice this is what I'm doing you do you do you
445. 1:09:49I'm not going to subscribe to anything annually for at least 18 months because all of this [ __ ] is going to be
446. 1:09:57made obsolete by something else within three months six months so you're gonna you're gonna lock you in
447. 1:10:05for a year for the latest state of the art thing and then Google's gonna come out with a free tool that kicks the [ __ ]
448. 1:10:10out of the thing that you just paid for a year for and it's gonna make you cranky so just go month to month for now yeah
449. 1:10:18you're not going to save the 20 or the 10 by doing the annual thing but just again janky pieces of [ __ ] they're all
450. 1:10:25janky pieces of [ __ ] until further notice um and that just means there there's just gonna there's there's just too much
451. 1:10:33is too much is changing too fast to buy annual subscriptions right now um
452. 1:10:55apps were a ux Improvement to websites bot Personnel llms will improve them both I predict I agree with that
453. 1:11:04apps for a ux Improvement to websites yeah well the web was the ux Improvement to the internet when it was command line right
454. 1:11:13and then apps are improvements to to just raw hyperlinks yes and then chat gbt was a ux Improvement to command line machine learning stuff
455. 1:11:26and then and then as these tools get more sophisticated I think you're absolutely right coruno I think it's
456. 1:11:33going to be like this I think it's going to be um you've got all your complicated application layers down here you've got
457. 1:11:40Salesforce and slack and just all these [ __ ] overly complicated pieces of [ __ ] software I'm I'm sorry very
458. 1:11:47professional useful software that's completely worth paying for but it's overly complicated and these
459. 1:11:55large language models are just going to simplify all that right it's just it's going to be natural language I want to
460. 1:12:01update this in Salesforce it goes and figures it out I you know I want to notify these three different slack
461. 1:12:09servers this message in the appropriate channels and it'll figure out what the appropriate channels are so yeah totally agree
462. 1:12:17um that's that's I I think that's going to be one of the initial killer apps for um for these things is just simplification of the overly
463. 1:12:28complicated I should be working on a password app but Kyle's too funny so I'll keep the endorphins flowing
464. 1:12:38sorry Winston yeah I wouldn't be working on high Security application development while I'm acting like an idiot you'll
465. 1:12:45just type in there you can make money with chat gbt wait what that's not in the code what am I doing [Laughter] [Applause] oh
466. 1:12:57just uh the cheese t-shirts pretty cool looking those t-shirts by the way this this graphic was done with um
467. 1:13:05uh ideogram ideogram.ai Winston thank you so much for the coinage sorry for screwing up your productivity um
468. 1:13:18just to be sure that you're the real one wow is it Tick-Tock or is it big brother oh Tick Tock is totally big brother they've
469. 1:13:26got their little words they know everything we're saying everything we're doing they're like you enable the store and
470. 1:13:31you're not here talking about your products 24 7. that's how we designed it and that's what we want you to do so if
471. 1:13:37you don't do that we're going to shut you down so I will shut that [ __ ] store off if that's the rule you hear that tick tock
472. 1:13:48you got Gen X on the line don't screw with this we got a good thing going here um was that an impromptu ad
473. 1:14:03I don't know maybe did I do an ad you can make money that's not impromptu but it started as an impromptu thing
474. 1:14:15because I just like when I first started doing these lives people you know are asking me how do I make money with this stuff and
475. 1:14:22and it's just those Tick Tock channels that have that those people on there that say that you know you can make money in real estate you
476. 1:14:33can make money yeah you can make money that's what Tick Tock wants me to turn this channel into catch catch you get your AI learning
477. 1:14:44love t-shirts now no not gonna happen not going to happen foreign
478. 1:15:01for anyone who wants to create an app with your ideas okay cool yeah I've heard of that one before I think I played with it a while ago
479. 1:15:09let me look at it again happy pie epipi no code development platform the challenge with no code development platforms create your app
480. 1:15:23if I have to [ __ ] sign up for this before I can create my app Builder that builds any business name all right we'll do storyline Inc
481. 1:15:33choose the business category um this feeling like marketing uh [Music] all right let's say business pick a color scheme you like
482. 1:15:48we'll go deep ocean Mac iOS test your app you must create an account first all right this looks like the right
483. 1:16:02thing so whoever was asking for uh an app builder before like this that I I do remember going through this and playing with it before
484. 1:16:12so it's going to make me I think I even signed up for Google with this one hang on let me see if I did 10 million happy clients hey well
485. 1:16:3010 million people signed up all right it's not letting me in uh anyway yeah that was that one's worth that one's worth exploring
486. 1:16:44[Music] [Music] all right let's see what do we want to do the Multiverse yay confetti
487. 1:17:00it's not that complicated Pete welcome back sir paint Mater by the way great name paint Mater
488. 1:17:15for those of you in the thing that are wondering what my um qualifications are I have none Pate does Pate works at Google
489. 1:17:27Pete's got an actual PhD I just play a PhD on the tick tock I don't even play a PhD I'm just a dude trying to figure this [ __ ] here's the
490. 1:17:38deal to figure this stuff out just like you are and because I've kind of been through one of these Cycles before with the World Wide
491. 1:17:51Web I started one of the early digital agencies in the mid 90s I know the opportunity that's sitting here this shit's way more profound than
492. 1:18:02the World Wide Web was and that changed almost everything so um that this channel is about just trying to [ __ ] figure it out and
493. 1:18:12that's that's really what it's about right now so the inflection point in my opinion is November 30th 2022 is the day that chat GPT launches
494. 1:18:24and that's the day that you know there was a you know a diverging in the roads right on one side are all the
495. 1:18:31technologists and the engineers and the researchers and the phds who have spent the past three or four or five or seven decades building
496. 1:18:40all of the infrastructure that has allowed us from this day forward to take these tools and incorporate them into our
497. 1:18:48businesses and into our lives and do all this remarkable [ __ ] as individuals and as businesses without having to have a
498. 1:18:56Fortune 500 you know it budget and Tech budget right we can just incorporate this [ __ ] into slack using zapier or we
499. 1:19:05can just go to chat gbt and have it write us some [ __ ] like we're in this remarkable time in history the challenge is these tools are so
500. 1:19:13profoundly powerful none of us zero of us from the research scientists to the brightest Minds in business no one
501. 1:19:25knows how these things are going to impact us moving forward no one that's what we're trying to figure out and so
502. 1:19:32the only way I know to do that is to get in get your hands dirty it's impossible to describe these things to someone
503. 1:19:40but the minute you show show it to them and they have their Kevin McAllister moment this moment then they go oh [ __ ] this is something
504. 1:19:52different I need to learn this yeah you do we do we all do so that's what this channel is about writing in French
505. 1:20:03who will guard the guards themselves who will also guard the guards themselves I don't know who's gonna guard the guards
506. 1:20:11I mean now that meta has gone on ahead and dumped you know dumped open source llama on us no one's guarding them
507. 1:20:26I think your question's a valid one if you're talking about you know Google and open Ai and Microsoft and
508. 1:20:33inflection and cohere and you know the companies that are creating guard houses but now this shit's open source so no
509. 1:20:44one's guarding the open source so no no and you may ask what are the implications of that Kyle well they're profoundly good and profoundly
510. 1:20:57bad you know although I would argue that the the large language models that are out there right now as powerful as they are
511. 1:21:07we ain't seen nothing yet there was I was listening to an interview with Sam Altman the other day and um it was from a couple of months ago
512. 1:21:16and he said oh there's some [ __ ] at open AI we will never release it's like yikes in the language hyperspace yes that's correct
513. 1:21:29like when they show electronic pulses traveling through the brain yep that's the best explanation I've heard thank you I'm I'm I thank you
514. 1:21:42I appreciate that um it's just maths and probabilities exactly that's that's the thing that's so remarkable about this right the
515. 1:21:50Arthur C Clarke quote of any sufficiently advanced technology is indistinguishable from Magic the fact that I understand how that
516. 1:22:00large language model works is so completely distinct from what they actually do they're [ __ ] magic I'm
517. 1:22:08sorry it's it's magic it's [ __ ] Magic all right that's not entirely you they use that's not entirely true they use
518. 1:22:21beam search most of the time so I assume beam searches what do you pay is what you're talking about they're not doing one word at a
519. 1:22:32time they're doing um chunks of words I think that might be what you're talking about um this is artificial intelligence it's
520. 1:22:40the loaning us yes where multiple words words are being considered at one time for multiple token slots at some distance yeah so so
521. 1:22:48yeah I know I just tried to I just try to simplify it um not all of them are doing the beam search stuff but it sounds like maybe
522. 1:22:56they are now um when I put that thing together I think that was mostly accurate but I've heard that as well that they're
523. 1:23:03doing clusters of words at a time but point point is still the same where it's it's a mathematical probability of that
524. 1:23:09cluster of words rather than that single word is still [ __ ] remarkable um Kyle then wouldn't all the letters of
525. 1:23:17recommendation be identical as each word triggers the next no because of a thing called temperature so Dr J so if if you ever play in the in
526. 1:23:29the um in the playground there's a thing called temperature that you can slide between it used to be between zero and one now
527. 1:23:36they changed it to zero and two for some unknown [ __ ] reason um where if it's if it's all the way up to like one or two
528. 1:23:47um then it's going to be really creative it throws it effectively throws some randomization in there if it's down at
529. 1:23:52zero it makes it more precise so if you move it all the way to zero you can get it to consistently um answer things the same way but
530. 1:24:01the the default for chat GPT I think was like a 0.6 out of one so it's sort of just over you know halfway between not
531. 1:24:11creative and fully creative um so it will do some variations um then it picks the best overall path so it can do groups of words not just
532. 1:24:20one token at a time thank you for clarifying that Pate um which app should I use for what advising me not to use okay you're
533. 1:24:32talking to each other that's cool 9 15 in Denver and it is now 9 32. I'm not so far behind I am 99 Plus messages behind
534. 1:24:45um so maybe I'll jump down are all or most llms use math models to find single or clusters of data based on the query is that a question
535. 1:24:57all or most llms use math models to find singular cluster of data based on query I think that's a statement
536. 1:25:04yeah that's right that sounds right oh let's see Tick Tock just jump me to the bottom probably because I'm not selling enough
537. 1:25:14you can make money with chat every day you can make money with selling T-shirts in The Tick Tock shop [Laughter] oh good lord who's that
538. 1:25:29hey Kyle and a regulars hey major gift fundraising what's happening um [Music] I'm gonna just go back and grab a
539. 1:25:44question or two can you share the slides I don't know where I can share them there's so that explanation that I gave of the
540. 1:25:57how the large language model works that exists um the the I think it's the second most recent video I did is a Champion wants you to know
541. 1:26:11he's protecting us um the second to last video I did I did a tick tock video in my channel that that uses that slot oh no it doesn't use
542. 1:26:21that slide never mind I use a slide like that for images um I'll tell you what I'll do I can put that slide in um in Discord so
543. 1:26:36hang on so if you go here the salon.ai so this is a group I started called the AI salon and we've got a Discord so if you go
544. 1:26:48there we we have every other week we have a meet uh meet up and we meet in person here in Denver and we meet online
545. 1:26:55and then we also have a Discord and I've created a private channel for the Irregulars for people from this channel
546. 1:27:01so I can post those slides in the Discord there so if you go sign up there and just ping me that you want those slides
547. 1:27:09um I'll get them to you or just reach out to me directly my contact information is in my profile so you can
548. 1:27:15just reach out and I'll send you the slides I'm happy to do that 22 000 or up to thirty thousand now this Sayers why can't
549. 1:27:28GPT fact check itself so there there are companies right now looking to do that so um perplexity.ai doesn't quite check itself
550. 1:27:38but it does do citations so it's doing some weird combination of like a large language model search and a web search and then
551. 1:27:48putting its citations in there I don't know which one it's doing first it's sort of doing that the the logic that you're talking about
552. 1:27:58uh this this Sayers is um it makes perfect sense right like half check GPT generate an answer then check that answer against reality
553. 1:28:13and if anything's wrong like the citations are [ __ ] fix them right when I when I say that these tools are
554. 1:28:23janky pieces of [ __ ] that's exactly the kind of stuff I'm talking about it doesn't mean you shouldn't use them it
555. 1:28:30it means you should understand where the limitations are on these things right because they're janky pieces of [ __ ] and
556. 1:28:38they don't things do things like fact check themselves you need to know that so so there's a pretty famous uh if the
557. 1:28:45event that happened where a lawyer had chat EPT write a legal brief and he handed it to a [ __ ] judge without
558. 1:28:52proofreading it and the judge looked at all the legal citations and said um these do not exist what are you doing
559. 1:29:03and the lawyer looked like an idiot that's not on chat GPT that's on the lawyer thinking he could get one over on the
560. 1:29:13system or whatever he was having a bad day um but it says to me that he didn't really understand the tools right and he
561. 1:29:21didn't understand where the boundaries are so it's super important right now because we're so early that you don't just say
562. 1:29:31how do I use chat GPT to make money you can make money with that gbt you actually learn what these things are how they work
563. 1:29:42the the things they do that are [ __ ] brilliant I mean there's things that these things do that just blow my [ __ ] mind and then there's
564. 1:29:51other things that they do that you're like they couldn't have [ __ ] solved that do you know the chat GPT this is like
565. 1:29:59this this this is mind-blowing to me it's been out for 10 months now yeah we're in September came out November 30th nine months
566. 1:30:13you cannot search your prompt history in chat GPT you can in the app the iPhone app you can search your prompt history in the
567. 1:30:24web in the web app you can are you [ __ ] kidding me I've been prompting this thing for nine months and if I want to find the thing
568. 1:30:36where I made a tweet thread I can't [ __ ] search for that so yeah these things are all and I love Chachi VT I think it's you
569. 1:30:47know I'm like I think I think Sam Altman's a really interesting CEO really interesting dude actually thoughtful
570. 1:30:54understands the implications of what he's doing and if I met him at a conference I'd say sam uh when you bring in search to the
571. 1:31:07uh prompt Library there buddy couldn't you throw a junior engineer on that I mean it exists in the app we know how to do search
572. 1:31:24[Laughter] puppy makes me chatty too um I appreciate the explanation as I am just a nurse so listen Debbie here's the deal first of all
573. 1:31:35just a nurse thank you for what you do um holy crap just knowing what I know about the hospital system right now and how
574. 1:31:47understaffed they are like really thank you for what you do um something I that that's really really important right now is
575. 1:31:59you know I talked before about that inflection point of November 30th you know before November 30th 2022 it was
576. 1:32:06all about the Geeks and the researchers and the scientists and after it's about the rest of us don't apologize at all for wherever you
577. 1:32:14are on the on the learning curve of this stuff the good news is that thanks to chat GPT and tools like it
578. 1:32:25you don't have to be a data scientist to use these remarkable tools like we're we're an amazing time in history
579. 1:32:34what you do need to do is be able to think critically like what's the problem I'm trying to solve and can I get these
580. 1:32:42tools to help me solve that problem the answer is actually probably yes you don't have to write an application
581. 1:32:49you don't have to use it to write code you don't have to do all the fancy [ __ ] you don't have to do automations you
582. 1:32:56might just say hey I'm going to take the the PDF of our nursing schedule and I'm going to pop it into this thing inside
583. 1:33:04chat GPT called Advanced data analytics even though you might not be a data analyst person like I'm also not and I'm
584. 1:33:13going to have it help me um come up with a better way to um you know organize the nursing schedules because that's a problem you need to
585. 1:33:25solve start with the problem start with the thing you're trying to do and then you can say how can these things help
586. 1:33:31rather than saying you need to become expert at AI or expert at chat EBT or expert at prompting no no no no no no no no no no
587. 1:33:41will it benefit you if you get better and better at this [ __ ] sure but you don't have to start with that don't put that burden on yourself
588. 1:33:50the the the these are tools and this is so I make this I make the joke about these people right and here's why I make fun of them
589. 1:34:05how you make money with chat gbt is identical to how you make money with Photoshop and how you make money with
590. 1:34:19a real estate education program and how you make money with a shovel these are tools you're not making money with chat gbt
591. 1:34:33you're making money by solving a problem that someone else is willing to pay for and you just happen to be using these tools to do it
592. 1:34:42and and that's that's the real scary thing about these is that they're so sexy that it's really easy to get Hoodwinked
593. 1:34:50into thinking you have to become expert at these things no the expert at the [ __ ] you want to do and the stuff you want to solve
594. 1:35:00and then know enough about these things to go oh I think I could use Chachi BT to solve this part of this equation
595. 1:35:09yeah let me go talk to it and see oh yeah I can oh it gave me a new idea oh and it gave me a new idea but you're
596. 1:35:18still solving that problem right it it almost has nothing to do with this it has all to do with you so anyway
597. 1:35:29that's why I get crazy about those [ __ ] people because they're they're they're perverting how we use technology these are our tools
598. 1:35:40we get to use them how we want not the other way around the robots are not using us we're using them they're amplifying our abilities
599. 1:35:51so anyway thanks for what you're doing um I'm glad that helped awesome answer thanks great Bob cool very cool explanation
600. 1:36:03but what was the magical gift attack I don't know Tick Tock told me I had to talk about the store more
601. 1:36:10it's it's like it said it said like like uh Tick Tock live violation it's like you've got a store and you
602. 1:36:19weren't talking about it for the entire time you were talking whatever like I said I'll turn that [ __ ] off
603. 1:36:29we'll figure out some other way to get merch into the world I gave Bart a run then went back to chat GPT Bing's version told me to stop
604. 1:36:40asking how old the llm was uh see me now so yeah so Bing is Bing officially got kicked to the kids table I moved Bard up to the top three
605. 1:36:51um this is really for me it's the top two but if I need if I need uh a large language model connected to the
606. 1:37:00internet I'm now using Bard Bing is really [ __ ] annoying at this point the the safety guard rails on Bing are just
607. 1:37:07it's it's just bad it's just bad so I'm right there with you chat you that's why chat GPT is still on top it's the best
608. 1:37:14Claude can do a hundred thousand tokens so that's a 75 000 word context window which is sometimes you need that so
609. 1:37:21those top two are the biggies that I use if I need something connected to the internet I'll use Bard now and for the
610. 1:37:27most part I just let Bing sail off into the sunset it's at the kids table with the other [ __ ] um things move fast uh for that
611. 1:37:36commitment to a single API yes yes exactly any ideas why all the big ai's Technologies seem to have become mainstream at the same time
612. 1:37:53so crypto Joe Google invents this thing called the transformer in 2017. that allows for all of these companies to be able to
613. 1:38:08train large language models machine learning models on massive data sets Google in particular Pate probably knows this better than I
614. 1:38:22do they invented the [ __ ] they've got a lot of really powerful technology they effectively mothballed it they sort
615. 1:38:29of put it off in a corner I think because they recognize that if they build a tool that answers questions it fundamentally [ __ ] with
616. 1:38:38their business model their business model is predicated on you clicking on lots of links to get the answer if they
617. 1:38:45just give you the answer you don't click on the links and they don't make money so they kind of Set It Off to the side
618. 1:38:56meta effectively did the same thing I don't think Apple had um a big a big you know bet in the game they're they're a hardware company they
619. 1:39:07sell they sell you know they make their money primarily on Hardware um Microsoft when when gpt3 came out Microsoft I think they saw an
620. 1:39:20opportunity and they said you know on on seeing what gpt3 did you know as it goes I think Sam Altman went over to like like uh
621. 1:39:33the Microsoft CEO's house and I think Gates was there like I I basically some folks at Microsoft saw what the [ __ ] was
622. 1:39:42going on with gpt3 and I think Microsoft saw an opportunity here to maybe get relevant again in the search game
623. 1:39:51and so they made this big investment in open AI as they were about to announce their first billion dollar investment Sam
624. 1:39:59Altman said hey we don't really have anything that we can point to point people to that's really easy to use
625. 1:40:07in the documentation of gpt3 the the developers version of this thing in the documentations were instructions for chat GPT
626. 1:40:19no one had built it for like a year and a half and so Sam Altman I guess back in October this time last year
627. 1:40:28said to his developers hey we're making this big announcement with Microsoft you got two weeks why don't you just go build chat GPT
628. 1:40:36and they did and they launched it and he assumed no one would give a [ __ ] because they didn't really give a [ __ ] with gpt3
629. 1:40:45like developers thought it was cool and Microsoft thought it was cool and they were going to give them a billion dollars
630. 1:40:51chat apt went from zero users to 100 million users in six weeks it got to a million users in five days it was absolute [ __ ] insanity
631. 1:41:02so so what happened was this caused a red alert at Google that said oh [ __ ] we need to deal with this um I think it did a similar thing at
632. 1:41:12meta where Zuckerberg was like okay we gotta pivot off the metaverse thing that's that ship has sailed even though
633. 1:41:18we named the company meta let's pivot away from that and over to AI um Apple accelerated things like Amazon accelerated things
634. 1:41:31um all of these companies like inflection and cohere and um and uh anthropic is clawed um all had this stuff sitting in the
635. 1:41:46wing so so they've been developing on this Transformer model for like six years but what happened was Chachi BT
636. 1:41:53got so popular so fast that it forced everyone to go [ __ ] and they just shoved all of their technology forward so
637. 1:42:02although the technology that was sitting kind of in The Ether off in the background like well we'll take our time
638. 1:42:07and and we want to make sure that it's safe and we don't want to hurt people and chat CPT does what it does and everyone
639. 1:42:16[ __ ] panics and throws all the technology out there including meta going here's we're gonna open source to
640. 1:42:23the developer Community llama one but it's for research only and then when llama 2 came out they're like [ __ ] it
641. 1:42:30here it is for commercial use so there's there's now a incredibly powerful um open source commercial llm out there
642. 1:42:42stability AI is doing the same thing so the developer Community is going that [ __ ] crazy on this stuff the big
643. 1:42:50technology companies are going batshit crazy on this stuff Apple just said that they spent the bulk of their research budget which
644. 1:42:57is 24.6 billion dollars a year they spent the bulk of it on generative AI this year so so they're coming so so
645. 1:43:07while it seemed like it materialized out of nothing this was all being worked on in all these companies but then the
646. 1:43:13popularity of chatgpt forced everyone's hand to say we've got AI Google's first meet first meeting the
647. 1:43:21the thing in Paris they did where they they lost the demo phone and it was it was pretty bad um but like basically what they said in
648. 1:43:29that meeting was uh we invented this stuff we've got Ai and everything which they do but they're on their heels because
649. 1:43:38they weren't leading this conversation someone else was and it was Chachi BT and Microsoft and you know the C the CEO of Microsoft
650. 1:43:48was fairly giddy that he was [ __ ] with Google that was pretty funny he basically said I'll lose money on Bing chat on you know gpt4
651. 1:44:02um for for like like even a if I can get one user to come over from Google like I'll lose money on that thing um
652. 1:44:13my ex didn't sign up for a year it was made obsolete in three months yep six months exactly how about investing in
653. 1:44:21the one AI startup I don't know it I would I'm I'm not going to invest in anything right now I don't think around
654. 1:44:30this stuff it's this is way too volatile one AI generative AI API for your product
655. 1:44:46so I guess this looks kind of like Lang chain commercial version of Lang chain maybe I don't know maybe I don't know things are moving too
656. 1:44:57fast and it's like like one of the I don't know if you saw there was a there was a paper that came out or not a paper
657. 1:45:04it was like an internal Memo from a Google employee that was like four months ago was talking about um time is it going on 10. um
658. 1:45:16it was talking about the the title of the piece was we have no moat and it was basically talking about large language
659. 1:45:23models and how the open source models are getting as good as or close to as good as the really expensive models that
660. 1:45:31the big tech companies are building and there's a similar sort of concern that any piece of technology that you
661. 1:45:37build any any business that you build and launch because of generative AI someone else could knock it off in a weekend
662. 1:45:49someone without technical chops could knock it off in a weekend and so everyone right now is going to be scrambling to find what is their
663. 1:46:01differentiator in the marketplace and I don't know that it's going to be on software features I think the software features are just
664. 1:46:10going to become ubiquitous where the the interfaces English language I want software that does this and it'll
665. 1:46:17just make that software for you why would I need to pay someone to do that now because the tools of janky pieces of
666. 1:46:24[ __ ] right now it doesn't do that right now but the writing is on the wall that when they start incorporating these
667. 1:46:31things together and overlapping them and if if Google's Gemini comes out and it is as sophisticated as it sounds like
668. 1:46:39it might be able to do things like interface and code and ux and marketing and and and and right or it becomes the foundation for that
669. 1:46:51so I would be reticent to invest in anything right now unless unless you're doing stuff kind of at the at the
670. 1:46:59like the Frontier Model the the the core research the the sort of underlying [ __ ] where these where the big the big boys
671. 1:47:07and girls are competing like those are interesting things because whoever wins that race is gonna you know they're
672. 1:47:15gonna have like infrastructure right they're going to be selling the picks and shovels to the miners all of these software applications are
673. 1:47:24I've got a mine let's go get some gold in the mine so I can't I can't [ __ ] keep up with just even using this stuff much less
674. 1:47:37investing in it what's the mid-journey prompt line battle I don't know what that means togas write it in French maybe we can
675. 1:47:49um maybe I can translate it and it'll make sense um time to do the store and Link um love kids saying it I love Turtles
676. 1:47:59I don't know what you're talking about micro um [Music]
677. 1:48:16back here live at the waterfront village with my friend the zombie Jonathan you're looking good Jonathan just got an
678. 1:48:23awesome face paint job what do you think I like turtles all right your great zombie good times here at the waterfront
679. 1:48:31Village open for the next 11 days all right laughs
680. 1:48:51uh some comments were filtered to protect the community's experience don't be jerks and comments no no
681. 1:49:01stop it stop it zip it zip it zip it zip it zip it zip it thank you Cheryl thank you for the cruise
682. 1:49:17should have been terms of service for tick tock with the AI yeah exactly I actually don't want to know what's in
683. 1:49:22The Tick Tock terms of service but I'm sure my family doesn't belong to me anymore can some of my friends here
684. 1:49:28please tell me when I can get duet AI so if you haven't seen it so um so what Anne's talking about is duet
685. 1:49:37is Google's branding for their version of chat gbt being rolled out into all of The Office Products of theirs and Google
686. 1:49:47meet and all that sort of stuff they're calling it duet they're saying it's generally available but you have to go
687. 1:49:53ask for it so let me see where the maybe it's an obvious place Google duet General availability
688. 1:50:07now it will be available to all of its 10 million users Google workspace now available August 29th
689. 1:50:23yeah this is it so go to workspace.google.com blog and then it basically it's the article now available duet AI for Google
690. 1:50:36workspace and then somewhere in here yeah the second the second hyperlink in the first paragraph is get started now with a no cost trial
691. 1:50:49so it's free initially and then I think it's going to be 30 bucks a month with so many AI powered applications out
692. 1:51:02there is any is there any room for any Innovation holy crap yes holy crap yes so here's the deal uh what's your what's
693. 1:51:11your name Thai tyrellius zakarius okay that's a cool name I can't pronounce it but it's super cool tyrellius
694. 1:51:21zuccharius I like it solid 8.5 out of 10. only because I can't pronounce it this is probably more on me than you um so the innovation
695. 1:51:39the tools are going to do remarkable things so there's there's so many places to innovate one of the places to
696. 1:51:44innovate is how are you using these core large language models you could you could kind of go upstream and you could
697. 1:51:51say hey I want to train you know um these image recognition models on a particular kind of Imaging right that's
698. 1:51:59what's happening in healthcare where they're training them on how to read X-rays and pet scans and things like that um
699. 1:52:08you could build application layers on top of large language models you could use something like zapier if you're not a
700. 1:52:16a coder and do automations like one of the things that I've built in zapier are content generation engines where you ask
701. 1:52:24people a few questions they answer those questions and it generates a bunch of social media content or it generates an
702. 1:52:29article or whatever it might be um so you can do it there so so there's there's sort of the whole world of application development that
703. 1:52:38that can be super technical all the way up to you know fairly kind of front-end easy kind of stuff then you've got
704. 1:52:47all of the business process innovation how are we going to do project management moving forward how are we
705. 1:52:54going to do HR how are we going to do customer support right you know imagine taking these tools and completely Reinventing the customer
706. 1:53:05support experience with synthesized voice actually picking up the phone and answering questions and being competent
707. 1:53:12and having the authority to to do stuff right that's going to require obviously some some technology stuff but even if
708. 1:53:19someone builds that tool there's all sorts of innovation around how do you roll a tool like that out across a company that does customer
709. 1:53:28support support the way they do it today every single function whether it's copywriter or art director or consultant
710. 1:53:37or researcher or engineer or anyone nurse anyone who's dealing with any computer any computer mediated job which is
711. 1:53:50essentially in today's society it's almost every job even if you're a mechanic you know working on cars all
712. 1:53:57day long how do you take money with computers how do you schedule people with computers how do you send messages
713. 1:54:03with your phone with computers right so almost all of our Workforce today is computer mediated anything that's
714. 1:54:15computer mediated is going to be transformed by these tools how we deal with that transformation and and what that transformation is that's
715. 1:54:25all the Innovation that like like every single thing is going to be Innovation even though these things are doing
716. 1:54:32remarkable things it's up to us to figure out how we want to wield them to make the change to make the difference
717. 1:54:40so [ __ ] massive opportunity massive massive massive Upstream Downstream change management holy [ __ ]
718. 1:54:53people that are like I can help you yeah so imagine you're working in an industry like the advertising business which I know pretty well
719. 1:55:03and you've spent the last five decades six decades charging time and materials basically the way you get paid is you tell people
720. 1:55:13I'm going to have these 50 people working 40 hours a week at 120 an hour to write you copy for all of your advertising
721. 1:55:22well now a tool comes along that can just do that so rather than 50 people at 40 hours a week I might be able to do
722. 1:55:32the same amount of work at the same level of quality with 10 people working 20 hours a week what happens to that advertising executive
723. 1:55:46that spent the last 50 years with a business model that just got obliterated they're going to need help they're going
724. 1:55:56to need to figure this out so there's a role for someone to just figure out all of the [ __ ] that just exploded how
725. 1:56:04do we tape it back together how do we keep these companies still running so it's like Innovation is going to be everywhere
726. 1:56:11and with Innovation comes pain and you know suffering and weirdness and awkwardness and resistance and fear and excitement and
727. 1:56:23young little upstart agencies that are going to come in and say hey I'm going to charge a flat fee for
728. 1:56:29advertising and we're going to deliver things twice as fast at half the price you want some and they're gonna become the next big thing
729. 1:56:39and the big old agency that resists that Change is Gonna potentially fall off the Wayside Fall by the wayside some of the big
730. 1:56:47agencies will make the transition some won't and I think about every single [ __ ] business is going through that everyone
731. 1:56:57are we there yet no is it coming yeah how fast I don't know it ain't gonna be slow it ain't gonna be five years
732. 1:57:10it's gonna be a year two it's already happening so yes much much Innovation coming we're in the middle of it that's why I'm
733. 1:57:21so excited this is like these are the you don't get to live through these kind of transitions very often in a generation if ever
734. 1:57:34in my life the three that I that I think are them are the personal computer I was too young for that the World Wide Web and generative AI
735. 1:57:49and after I did the World Wide Web thing I didn't think I'd see another one of these in my lifetime and then Along Came this thing this
736. 1:57:56thing it's crazy it's crazy it's crazy all right I'm gonna jump to the bottom of more than 99 questions behind there's 96
737. 1:58:06people in here 92 now and I'm going to answer a few more questions and then I'm going to go to bed about Sunday night we got a labor
738. 1:58:15day tomorrow we got do you like barbecue is Memorial day but is barbecue also Labor Day I don't I don't know I don't know I don't
739. 1:58:24know how to celebrate this particular holiday although I do find it curious that on Labor Day we've got this this tool
740. 1:58:34amongst us that is going to transform how we work in pretty dramatic ways so that's interesting well ain't that
741. 1:58:46special and that's special yeah let's see I wonder who who invented AI who Could That Be on huh oh I know who it was was it Satan all right
742. 1:59:01I hate being too perplexity is amazing just use Dolly through Bing that's all yep that's I use it for that if I want
743. 1:59:07to do it really quick but now that now that um uh this exists by the way go to ideogram dot a i i d e o like the
744. 1:59:19company Ido but it's not them graham.ai and it's just like one of these sites that has like a bunch of images on it
745. 1:59:29but notice what the images have they all have words and so you can just make words um so I can put in
746. 1:59:44make money with chat GPT exclamation point and then I'll go um let's see Roger black
747. 2:00:02Style poster and we'll do poster and we'll do typography and we'll do vertical and we'll do generate
748. 2:00:32make money it doesn't always get it right but against it right now like that's not even close make make s's with money chat GPT make
749. 2:00:46this with chat you BT um but yeah you can do all sorts of stuff you can do words you can do pictures so this is the one that I use now
750. 2:00:55instead of uh instead of being or anything quick because it's just it's super simple that's all it is you you
751. 2:01:01pick vertical horizontal or Square you put in a prompt the quality of the image rendering on this one is like Dolly it's not super great
752. 2:01:10um it's maybe a little bit better than Dolly but it's not mid-journ if you want good images go to Mid Journey but if you
753. 2:01:17want something like I love Comic Sans well that one should be in comic sense shouldn't it um but like there's a Netflix logo it
754. 2:01:25does logos pretty good I had to do a couple of NFL logos and it got pretty close a pig saying eat more chicken cute
755. 2:01:37I just baked some cupcakes some [ __ ] up cakes that's awesome so yeah that's ideogram.ai
756. 2:01:54pretty slick ain't it ain't it pretty good and pretty good a pretty good yeah they're pretty good mm-hmm I like them taters
757. 2:02:08all right let's see 945 in Chen land thank you Emilio's wife which AI can you use to help me with my business um
758. 2:02:16all of them really but but just quite honestly the the one to really just pay attention to right now if
759. 2:02:22you're just getting started with this stuff is just go to chat gbt it's got this thing called um Advanced Data analysis
760. 2:02:32um which you can use to analyze business spreadsheets and things like that and all sorts of data but but just learning
761. 2:02:40chat GPT for your business start with what are you trying to do within your business and then start asking chat GPT
762. 2:02:46about the things the problems you want to solve and just wait for it to blow your [ __ ] mind and then you can also go to this thing
763. 2:02:54prompts.chat that little bottom thing there that will teach you how to talk to chat GPT and it'll give you a lot of
764. 2:03:00ideas for how you can use it that's where I would start there may be more specialized tools like Jasper or copy AI
765. 2:03:08for marketing or or things that are more specialized for doing numbers or doing whatever websites or marketing what all
766. 2:03:15there's there's going to be all sorts of specific tools almost all of them are based on the core underlying technology here so if you get
767. 2:03:23your head around what these things make possible then you can start to look for specific tools that solve problems if
768. 2:03:29you can't solve them yourself 30 at 38 000 we're up to 49 000 likes you all are nuts you get nuts and I love
769. 2:03:39it I love it I love it I love it all right check out factaverse.ai for free fact checking it's been a while since I
770. 2:03:46used it but bookmarked very cool we've seen Facebook um each platform has its own strengths you said Apple
771. 2:03:57um yeah Apple's Gonna Be Coming Out Swinging with this AI stuff they're gonna do it right and they're gonna be
772. 2:04:03good and here's the thing about Apple is a very good reputation for not [ __ ] with our privacy you know refusing to open back doors for
773. 2:04:14governments and doing things like having you know keeping private Keys like locally locked on your Hardware um so Apple's got a really good
774. 2:04:25reputation for privacy and when they do this AI stuff they're going to do it locally if you paid attention to the Vision Pro
775. 2:04:34announcement one of the chips in so a couple of things within the M2 chips there's all sorts of machine learning
776. 2:04:40[ __ ] that can be done in the M2 chips they also have this thing called the the R1 the the reality chip
777. 2:04:47which is taking the input from all the sensors on the goggles and doing real-time machine learning on device on that chip so Apple's gonna
778. 2:04:57be [ __ ] phenomenal at local machine learning processing which is basically means your data's safe right you're not sending it to them
779. 2:05:09you're not sending it to open AI you're not sending it anywhere you're going to be able to take your data GPT the [ __ ] out of it and feel
780. 2:05:17confident that all of that stuff is within your own little data bubble um so I think and and they're also they're also just really good at
781. 2:05:25defining new categories so I think they're just going to be good at that [ __ ] they won't be good at it
782. 2:05:32any other are llms anything more than innovation in HCI what's HCI human computer interface um they are so so yeah JD
783. 2:05:52I mean I would I would suppose that technically you could abstract it up to it's an improvement in human computer interface
784. 2:06:01but here's the big difference the way I think about it historically if I want to understand some data I've got to find the data set
785. 2:06:17I've got to look at all the data I've got to clean up the data like like if I've got if people's addresses are typed
786. 2:06:23in there in inconsistent formats where some of them split the data out into different fields and some of them are in
787. 2:06:29the fields I've got to clean up all that data and then I as a data analyst or as a programmer need to think about okay
788. 2:06:37I've got the following data elements and I want to accomplish the following test task how do I need to logically think about
789. 2:06:46how I might combine and analyze this data in such a way that it gives me some desired result and I have to do all of that logical thinking
790. 2:07:00you can now just ask this [ __ ] to figure it out for you or you can just say hey so here's another example here's how Google works
791. 2:07:17Google you say I want to learn about the Mesopotamian War and Google's like awesome I got all the links of all the
792. 2:07:26people talking about the Mesopotamian War for all of Internet time here's all your links and then the first seven links are ads
793. 2:07:35to buy Mesopotamian t-shirts right they look like articles but they're really just ads for t-shirts they're like God damn answer
794. 2:07:45and then and then over time you click on enough links and then you go read the articles and you think okay I got the
795. 2:07:54Mesopotamian word started over this and then it did this and and you're you're learning right and this is taking you
796. 2:07:59however long it's taking you a week two weeks and then you learn enough about that war and you're like hey I wonder what the what the
797. 2:08:07comparison is of the evolution of that war with World War II because I know a lot about World War II
798. 2:08:15so you know what would be really cool is if I compared them so I'm going to take the starting points of each of them so
799. 2:08:22I'm going to go research this and I'm gonna go research this and then I'm gonna put it in a in a Google doc
800. 2:08:28like start of war and this one started here and World War II started here and then and I'm gonna okay with these things you go here
801. 2:08:39I'm gonna do GPT 3.5 because it's just really [ __ ] impressive and I gotta turn off my custom instructions because
802. 2:08:45I think it's acting like an idiot right now yeah I'm gonna turn that off okay new chat GPT 3.5
803. 2:08:59you're a history Professor that specializes in the analysis of major Wars period period
804. 2:09:14make a table comparing the major historical Milestones of the Mesopotamian War and World War II [Applause]
805. 2:09:34you could say to me that that that's just an improvement in human computer interface but this is something different right
806. 2:09:44this is knowledge on demand Google was information on demand but it was up to us as humans to learn the [ __ ] to find
807. 2:09:51the [ __ ] to sift through it to think logically about how one thing compared to the other this just does it time period 3000 BC to
808. 2:10:02539 BC World War II 39-45 Geographic scope primarily centered in Mesopotamia Geographic scope Global right so it
809. 2:10:12understood what I was asking for it gave me things now that was without even trying without really knowing anything about it if you knew
810. 2:10:19something about it you could do incredibly sophisticated analysis of these two things um essentially instantly now might it
811. 2:10:29hallucinate and give you wrong things yeah it might will it do that two years from now no it won't right so yeah it's
812. 2:10:37a fundamentally different thing it's a fundamentally different way to interact with computers so it's
813. 2:10:44not just an improvement in in interface in my opinion but I don't know I could be wrong
814. 2:10:58all right I've got 10 minutes what do you think about this what is it as good as the hype says what oh idx.dev
815. 2:11:08I have not heard the hype on it but let's go look at it idx dot uh it says it couldn't reach the page
816. 2:11:23idx.idx.dev it must be wrong I like turtles hi Kyle hello Robert Rossi Kam katkin la la la la
817. 2:11:39camp cat kitten pleased to meet you actually cam catkin would be a great name for someone selling um a class on making money with GPT hi
818. 2:11:51Kim cat can hear you can make money with your gbt I'm Kim catkin this is get rich with chat EBT large language models are your friend
819. 2:12:04the first class is free [Laughter] would you prefer Starbucks or oh that's cute I'm not gonna say it because I'll probably booted off Tick
820. 2:12:16Tock best businesses um chat EBT helped me write a cool book nice um it's how good you are at props how to monetize um
821. 2:12:30you got to figure out a way to use these tools in a way that people are going to pay you to use them clip is hilarious sorry
822. 2:12:39a couple of years at least it's in the mail let's see I'm jumping to the bottom here because I only got a
823. 2:12:48few more minutes totally agree I've been in sales Business Development accounting and HR I see Untold potential yup finish the summer party
824. 2:12:59oh that's what Labor Day is yep the future of AI is coming like a freight train with Ray Kurtz Violet the helm yeah I just said there's an
825. 2:13:08interview with Ray from about three months ago or so uh where he's talking about you know the singularity and they
826. 2:13:14were talking about when is it coming and he said his prediction years ago was 2029. and a bunch of AI people just sort of
827. 2:13:24lowered their estimate from like 20 40 or 2038 to 2030. so people just sort of lowered their estimate from like 20 40
828. 2:13:32or 2038 to 20 30. so it seems like in the next in the next handful of years we're going to be seeing some pretty remarkable [ __ ]
829. 2:13:44I reckon I like taters all right what about llama llama big deal llama big deal llama two and code llama so if you want to go play
830. 2:13:59with llama you can go to labs.perplexity.ai if if you're not geeky and don't feel like installing [ __ ] in Virtual machines or on your
831. 2:14:07machine lab stop perplexity and thank you for the Llama Winston labs.perplexity lets you play with llama 2 7 billion 13 billion 70 billion
832. 2:14:20and then code llama and code llama just beat gpt4 for coding so shit's getting better and it's free and it's open source and it's it's not
833. 2:14:34really open source it's commercially licensable but you can use it without paying them a penny all the way up until
834. 2:14:41you have more than 700 million customers so as long as you have less than 700 million customers you're good to go it's a big big deal
835. 2:14:51Thelonious sorry if I appeared condescending whatever double check it's not always accurate uh I did okay thank you for that
836. 2:15:06have you spoken about Chachi BT ask chat gbt oh Chad GPT no is It Like a Dude someone fine-tuned it to act like a dude we could do that
837. 2:15:18we could we could make it act like a surfer dude uh sorry it's not called Twitter anymore Nancy it's called X now I don't know if
838. 2:15:27you knew that but now it's called X okay Nancy all right let's say Chad GPT or yeah GPT gets it
839. 2:15:47is it an account hey bro you heard heard you're looking for big gains oh Chad GPT has you covered
840. 2:16:02here oh so it's like a crypto bro oh here it is Chad GPT with a check mark what's up Bros yeah perfect awesome [Laughter] that's awesome
841. 2:16:17oh hilarious hilarious what AI double learning lab is still on Tick Tock is it really cabrino uh are you serious I thought
842. 2:16:39all right so if that's the case it's back but it doesn't look like it has content
843. 2:16:57all right let me we'll do a little Service Announcement here all right so this sucks if it's back assistant little [ __ ] isn't he
844. 2:17:04so this is us this is this this is an [ __ ] who is a scammer so if anyone reaches out to you purporting to be me asking you to invest
845. 2:17:16in some crypto [ __ ] check really closely it's really hard to miss that there's two lowercase L's together I put them large just so you
846. 2:17:25can see it um that sucks so anyway just be careful out there be careful out there folks all right we did
847. 2:17:34we're gonna go we're gonna get on out of here listen I I think Tick Tock would not be happy with me if I didn't say so we've got a
848. 2:17:43um a store in here now so I've got three pieces of merch I just put it up tonight so I'm just trying to figure out if I can get it to work
849. 2:17:51um so that's there you can go check that out I'll probably add more stuff to that but if Tick Tock keeps telling me I've got
850. 2:17:58to talk about the store more I'll just pull it down and put it somewhere else we'll just we'll just make a store
851. 2:18:03somewhere else and I'll just point people to it um all right kavuno I had it make a whole bunch of content this morning cool all
852. 2:18:11right listen folks happy Sunday night been cool hanging out with you um have a fantastic evening I'll see you
853. 2:18:19tomorrow sometime have a fantastic Labor Day and uh let's do our little Romper Room Roundup here I see let's see I see Nancy
854. 2:18:30and I see Winston and I see Riff Raff and Matthew Craig official be careful of Matthew Craig not official it's not the
855. 2:18:40same it's not official harsh weed three cabruno peace Amelia's wife good night good night Tobias um let's see Stevo Firefly Robert Rossi
856. 2:18:53Chris Rock rocker Omar domain night from us oh cool we got Australia in the house is it morning there it must be morning there uh down
857. 2:19:05in the mud all right everybody cam catkin this is camcatkin checking out you can make money with judge EBT I'm Kim catkin see y'all