aiLL 7/9/23 - 2 - Jump In! AI Is Here and It's Mind-Blowing

Transcript

1. 0:00foreign
2. 0:21name's Kyle Shannon I'll lead the adventure this evening this is my second live today and uh just had some ribs and
3. 0:39did a little yard work hooked up the hose because it it's been raining so much here in Denver we haven't had to water
4. 0:45anything so it's all watered now looks like the next week's gonna be sun so we'll see welcome welcome welcome feel free to
5. 0:55share the live and hit the like screen let people know we're here make them believe like you're enjoying this and
6. 1:02then we can we can fool them we'll get them in here and then you know and then we'll extract cash from them like we do every night right
7. 1:11God man I'm [ __ ] rolling in the dough from these lives you you have no idea oh my God it's fantastic I think I think I'm up to
8. 1:233.4 cents an hour so at this rate if I had a nickel for every hour I'd be getting a raise anyway uh this is the AI learning lab
9. 1:36we'll wait for some more folks to come in looks like the uh looks like the algo Rhythm there he is hey Silver Fox what's
10. 1:43happening congratulations on uh building a thing for a someone with your restaurant knowledge that's really exciting that's really flipping
11. 1:54exciting um and I I put this in in uh in the comments but I think it's the right move I think starting with service is is a
12. 2:03good way to go because all of this stuff is going to change so fast that if you spend too much time building something
13. 2:09you're going to build something that's obsolete in three months and then you're gonna be cranky and pissed off anyway
14. 2:14welcome everybody my name is Kyle Shannon this is the AI learning lab all these URLs around my head here I'll flip it in a second
15. 2:22but here's why these are here if you have not played with chat GPT I'll talk about what the what the channel is about
16. 2:28and who I am if if that's of interest but if you haven't played with chat GPT while I while I chat here Sunday night class
17. 2:37um go to chat.openai.com that's the official chatgpt website and just start playing with it if you don't have any
18. 2:44clue what to do there go to prompts.chat that document down there and that'll that'll give you instructions on what prompting is and
19. 2:52how to talk with this thing it kind of looks on the surface like a Google search bar but it goes way deeper these
20. 2:58are some deep Waters deep deep Waters uh that's chat EBD and then bing.com is Microsoft search engine but if you click
21. 3:06on the chat button that's also chat GPT but it's gpt4 connected to the internet and it's free and you can make images
22. 3:13there you just say make me an image of a penguin and a superhero costume and it will and then poe.com gives you access
23. 3:20to six different chat GPT large language models or not Chachi well chatgpt like large language models including chat gbt
24. 3:28okay and then futurepedia.io that is uh just a big old directory of AI tools so let me let me flip these
25. 3:37so so while I ramble go there go there now and go there until you have your Kevin McAllister moment play with it until this happens
26. 3:49and then you're ready to get started okay so the way this works pop any questions or comments you have uh down
27. 3:59in the uh comments below you can ask questions about AI I'll do what I can to answer them or point you in directions
28. 4:07um I do snarky commentary I cuss occasionally so if you don't like that get on out and uh what else uh uh I'm an entrepreneur
29. 4:19I've started 12 or 15 companies the uh the re the this channel exists to talk about generative Ai and um the I I sort of see this this
30. 4:31inflection point of the stuff I talk about started on November 30th 2022 so I'm not a machine learning guy I'm not a
31. 4:39data scientist I'm not an engineer I'm an entrepreneur but I love technology and I love transformative technology the
32. 4:47last time I was even remotely this excited about anything was in 1994 when I stumbled upon this thing called the World Wide Web
33. 4:54I started one of the first online magazines I started one of the first digital agencies and I didn't think in my lifetime I
34. 5:02would be alive to see another thing that transformative thank you thank you for the hearty heart uh something that transformative and
35. 5:13then Along Comes generative AI chat GPT and mid-journey and stable infusion those things and it's going to change
36. 5:20everything in fact my contention is it's already changed everything it's just no one knows it yet well 100 100 million
37. 5:27people a couple hundred million people know it um but uh it's coming and it's coming fast and I think faster than all of any of us are
38. 5:36prepared for so that's what this channel is all about um what do I like most about AI technology um it is
39. 5:50somewhere between having magical powers and having super powers Arthur C Clarke says any sufficiently
40. 6:00advanced technology is indistinguishable from Magic it feels like magic um it felt like magic the first time I had
41. 6:09it write a python application for me and I don't write python um and it still feels like magic today and then as the technology advances and
42. 6:20gets more and more sophisticated I get more and more excited normally with stuff like this like when I was playing
43. 6:25around with blockchain and nfts and things like that I got super excited about it initially and the more I dug in
44. 6:32the less real it felt the technology felt real but the companies the projects the crypto Bros just all that stuff was
45. 6:39just very surface level this is completely the opposite um the harder you push it the better it gets the deeper you dig the more
46. 6:49interesting it gets the more profound it gets um and I've never felt more creatively alive and free as I have in the past six months because
47. 7:02this removes creative blocks like like the idea of writer's block or creative block or or any of that like disappears with this
48. 7:12stuff so so that's it for me it's kind of like it takes who you are and amplifies it and it lets you to take
49. 7:19little insignificant ideas you have and you kind of put them in these machines and they come out like bigger than you imagined much faster
50. 7:29than you imagine and then you get to start playing with them and refining them and so yeah I think that's the thing I like
51. 7:38most about it all right Sunday night class bard ee hello Tobias what's happening
52. 7:54yeah feel free to if you want to get other folks in here feel free to share the live double tap tap the like but
53. 8:00like button looks like we hit a a bunch of people and they all they all they saw my face and they were like yeah yeah and
54. 8:08they took off so let's get some more folks in here if we can feel free to pop any comments down down
55. 8:16below I had my moment today using Vector Journey awesome I don't know Vector Journey your home alone moment
56. 8:24what's Vector Journey let me go look at that vector Journey AI
57. 8:39oh I don't know what Vector journey is you mean mid-journey or is there some other tool I'm missing
58. 8:53where do you think AI will be in a year I'm glad you didn't say five years because five years I don't have a [ __ ] clue a year okay
59. 9:03in a year so this time July 2024 gpt5 will be out it will be truly multimodal meaning
60. 9:22you can input audio video image words and it will output all of those as well we will likely have the first iterations
61. 9:37of personal agents that are trained on our data so rather than like having to go to chat GPT and figure out what you want to
62. 9:45prompt it to do and do I need to add plugins like all the [ __ ] it is now will will be more used to it and and
63. 9:58the the tools will have more autonomy so we'll just say like go book me a trip or go get United Healthcare to pay for my
64. 10:08doctor's appointment or go reschedule my this or go book a whatever haircut and it'll just go do it so I think that's gonna happen I think
65. 10:20generative multimedia like generated video and generative music and all that sort of stuff will be incredibly sophisticated um
66. 10:35I think that will be five months out from the 2024 election and so I think the the internets will be full of
67. 10:49misinformation deep fakes crazy [ __ ] like that oh and I think that it so so whatever chat gbt becomes will be
68. 11:05incorporated into all of Microsoft products all of Google's products anything you do on AWS deeply integrated Within meta's products
69. 11:18and apple will have launched some AI something probably personal agents is my guess with Apple I think that's oh and then the other thing is
70. 11:31within businesses AI will not like right now people are like well I don't know if AI is real and you
71. 11:40know is it really ready for business and I don't know what I'm doing back to bed to bed those are the CEOs I'm making fun of um
72. 11:51I think most businesses will will be will either jump in with both feet and have ai strategies or we'll be forced to
73. 12:01have ai strategies I don't think I don't think there will be many businesses a year from now that can ignore this I
74. 12:08think it's just going to be too too far into what we do that was a lot good Lord does this man ever shut up I don't know why are you
75. 12:20watching him I have no idea hey Bob what's happening hey could you grab me a cold one awesome this guy again oh good lord I absolutely
76. 12:38love Bing but I haven't tried chat gbt yet it so so you have Bing chat is gpt4 it's just that it's it's uh it's
77. 12:50Microsoft's version of it basically so you've got the open AI chat.openai.com is chat GPT that's the original and then
78. 13:00Bing chat is built on top of gpt4 so it's the same underlying uh large language model that powers both of them
79. 13:09they just have slightly different implementations and yeah I mean a couple of things about Bing right now
80. 13:15um the the official chat GPT they pulled their connection to the internet because it was um scraping content from behind paywalls
81. 13:23and apparently it did it to Fortune Magazine and they time Inc didn't take a shine to having their content stolen by
82. 13:29a large language model apparently so they have to shut that pumping down so bing.com right now is the only one of
83. 13:37these large language models directly connected to the internet it also you can generate images with bing.com you
84. 13:44can also upload images to bing.com and it can read them like you can upload a picture of a unicorn and it will tell you it's a
85. 13:52unicorn with sunshine and rainbows and Skittles so yeah Bing's Bing's pretty solid you are correct wait what was I correct
86. 14:03about God dang it tick tock it just jumped me to the bottom again this guy seems like he doesn't know how
87. 14:10to use tick tock shut up Marge he's talking about AI he's talking about AI God lord God lordy was that like Minnesota Irish
88. 14:26all right where was I I had my moment today where will I be in a year I became number one wait someone became number 41 member on the team
89. 14:38congratulations Susie Beach toes you're number 41 on the AI learning lab team I have no idea what that means but that's awesome I think
90. 14:46[Laughter] uh there's I love Bing you are correct about the oh here we go the Creator Creator block statement yeah uh telepo the um
91. 15:01there's there's just something about Chachi PT's ability to instantly take vague ideas and just go I got you you're like I kind
92. 15:14of have an idea generally in crappy English with typos and you're like you got any thoughts on that of course I've got thoughts on that
93. 15:27and then somewhere in there you're like oh yeah that's it that's what I was thinking but that's not quite right do it like
94. 15:32this do it like that and it it is I like it's a weird emotional thing so okay so before if I had an idea for
95. 15:46something and and the next step was okay in order to bring that idea to in any form of existence I gotta go to Google I gotta
96. 15:59search for some sort of Wikipedia page or some sort of something I gotta read it I got it in it like if it had more
97. 16:08than five steps it had to be like a radically compelling idea that like kept me up at night to motivate me to go do that horrible
98. 16:17[ __ ] work right like the the blocking and tackling work of creative ideation and now with this thing you kind of have
99. 16:25that Vapor of an idea and you stick it in there and it's just like and you're like oh that's good a little more
100. 16:42that's just if you haven't used it why do I have this stuff on screen while I talk I'm not getting paid by them
101. 16:51if anybody knows anyone on open AI this channel is like a [ __ ] ad for them I'm not being serious about that I don't care if they pay me um
102. 17:02but that's why I have it on screen if you haven't played with it go there go play with it any one of those three
103. 17:10how do I feel about AI becoming a catch-all for ML and other things I think it is I I think machine learning is more accurate
104. 17:22um but AI is what they call it in the movies and tough [ __ ] ai's got a better PR firm than ml does I don't really care
105. 17:33I you know I think I think I I think those distinctions are important for the people building the technology and for the researchers and
106. 17:42the scientists to understand the difference between those but for the 150 or 200 people million people that are using chat GPT that
107. 17:52don't know the history of machine learning and AI doesn't matter to them what it is to them is it's the magic
108. 17:58thing that writes their project plans now or it's the Met it's the thing that helps them write their Great American
109. 18:03novelers the thing that helps them organize their Garden they don't give a [ __ ] what it's called um like for example open ai's new
110. 18:17mind-blowing face melting tool is called code interpreter because from a technical perspective what it's doing is called code interpreting
111. 18:28human beings don't know that it's a shitty name it's a shitty name for a product that's got a consumer product or a consumer user base
112. 18:39it should be called like my magic programmer or [ __ ] [ __ ] anything other than code interpreter which makes it sound like good Lord do I
113. 18:50take a nap now or when I start using it but when you use it it's face melting you upload a file to it and say what is this and it starts
114. 19:05writing python code to figure out what it is executing the python code and then writing its own instructions to help you
115. 19:14understand what's in the [ __ ] file that's insanity well we've named it code interpreter because from a technical perspective
116. 19:24that's the most accurate description of what's Happening from from the sense of uh you know it's also quite limited it's
117. 19:31just a python code at this point I mean of course in the future it could could of course be expanded to include other
118. 19:37languages and it could do things Beyond interpreting code of course and right right now the the the modalities are who gives a [ __ ]
119. 19:48give it a good name and tell people there's a there's a little magic box inside that thing that is going to change everything
120. 20:03maybe that's what we should do tonight is is go to chat gbt and come up with a new name for code interpreter because
121. 20:09it's the worst [ __ ] name I bitched about that somewhere and someone made a comment made a comment to me well it's the accurate
122. 20:18description of what it is so I don't care it's the answer to your question sorry about that I know an awful little bit there but you know
123. 20:30I just had ribs I'm supposing I'm supposed my blood pressure's a little spiked um have I experimented with the pi AI I
124. 20:45have um installed it you install that one yeah I think I think it's an app it's a yes it's an iPhone app
125. 20:54and I talked to it and it responded to me but that's about that's about as far as I got um I'm definitely paying attention to it
126. 21:03because they just got funded for 1.3 billion dollars for their series B that's a relatively significant series B if even in the world of of
127. 21:13ridiculously overblown series B valuations that's that's right up there um so I'm paying so I'm now paying and I
128. 21:21think they're called inflection is the name of the company behind pie so I'm paying attention to them I'm
129. 21:26paying attention to anthropic which anthropic I like a lot although the sense I have about anthropic is I
130. 21:35bet they're building some amazing [ __ ] we haven't seen yet and they have no marketing department because you just don't hear anything
131. 21:41about them but like the poe or the Claude 100K model that you can play with at poe.com 75 000 word context window holy [ __ ]
132. 21:52so I have a feeling anthropic's gonna be big I'm paying attention to them inflection open AI Microsoft we'll see what Google does
133. 22:04I figure Apple will show up in a year Apple should probably I let's see so they do their thing in June
134. 22:13and then when's the next thing September June is the developers conference September is the new products is that
135. 22:20how Apple does their thing or October um we might not hear anything from Apple on AI this next time but but probably by
136. 22:31next June they'll be a pretty significant announcement from them or or they'll just be something built into the
137. 22:36operating systems what AI program Graham can I use to copy a website um there are I would go to futurepedia.io
138. 22:49and go to there's there's probably a web design section there I know there's some that copy websites there's some where
139. 22:55you can you pop in a URL it scrapes the website and then separates it into a figma a figma design Library based on that website so I know
140. 23:06there's some that do it I just I haven't done it so um I would go to futurepedia.io that bottom one there that's why that's there
141. 23:15futurepedia.io its purpose on this list is not that I love it it's just that it keeps track of all these tools because I can't
142. 23:23so I think it's what the guy the guy's name is Matt wolf that started that he's got a YouTube channel um the ability to aify yourself is
143. 23:33coming very soon upload your digital footprint getting it AI yeah I agree Dave I think I think that's going to be
144. 23:39I think I think that's kind of the next big thing right where it somewhere between autonomous agents and personal
145. 23:45personal agents or you know personal assistants or whatever we're going to call them pis um yeah I think I think that's next because because
146. 23:56like I only give so much of a [ __ ] about outside data unless I'm a data scientist or unless I have a client that's doing
147. 24:03it but everyone on the planet would like to be able to leverage like if you think about depending on how long you've been on the
148. 24:13planet I've been on the planet for a fair [ __ ] time and I've been on the internet in a pretty significant way since late 80s early 90s
149. 24:25in some form or other and wouldn't it be amazing to be able to just be rewarded for all of my digital hoarding
150. 24:35right I've got all these hard drives full of data I've got all these I've got a MySpace page somewhere I've kind of
151. 24:42[ __ ] you know Facebook for a decade and a half right we all have all this [ __ ] everywhere wouldn't it be amazing to just be able to
152. 24:52point this tool to all that and have it kind of synthesize it and assimilate it into some usable thing that can that can amplify my
153. 25:02contribution that would be amazing so yeah I agree I think that's coming I think that's going to be big
154. 25:09have you heard of people deleting their negative comments about AI in case it ever takes over the future I have heard
155. 25:17that I think it's I think it's listen here's what I here's what I appreciate about that it's at least critical
156. 25:24thinking like I the thing that you know I evangelize the most on this channel is get curious about this [ __ ]
157. 25:32like understand what it is understand how it does what it does and what it makes possible and all that but in the
158. 25:39end I just want people to be curious about it and so if someone's curious enough to go you know if this thing gets
159. 25:45powerful enough and I say shitty things about Ai and it doesn't like that um yeah I'm gonna go get rid of that [ __ ] [Laughter]
160. 25:57but then listen if if you're gonna go down that paranoia rabbit hole then you got to get rid of your Alexa because you
161. 26:04know you got to go delete all your Amazon data and your Siri data and your Google data like all the [ __ ] machines that are listening to us
162. 26:11because all the stuff that you're saying to your friends on the phone call that's being recorded too it ain't just what
163. 26:17you typed on Tick Tock comments foreign so I think it's cute I think people give
164. 26:32corporations and AI more credit like like I I think people think they're more important than they really are like if
165. 26:39you're actually a really important person that would be targeted maybe so I mean we're all just doing our thing here I don't think AI or
166. 26:47corporations generally give too much of a [ __ ] about what we do individually they just want our money uh uh uh
167. 26:57real-time voice cloning is already here imagine a personal AI powered voicemail screening service Apple has already
168. 27:04announced that that's coming in fact in the new version well not quite what you're saying but the voice cloning
169. 27:09piece in the new operating system you're going to be able to have a virtual a synthesized version of you that can
170. 27:17like respond to voicemails when you're driving and for people that can't talk it's going to be able to do something for them so
171. 27:26um yeah I agree I think that's coming Dave's on top of it slash home slash Dave I dig it he's in it love your lives thank you
172. 27:37very much Nature's Blessing I appreciate that you have a Charming personality there's there's a line from the Cherry Orchard
173. 27:49Mama do you think I'm pretty you have beautiful hair [Laughter] that nice personality thank you thank you
174. 28:02see that these are the comments I like in my lives it's a compliment right right laughs oh it's good it's solid
175. 28:15all right let's see where are we so I'm finding myself in the uh comments here
176. 28:30[Music] love your lives there's the Charming personality you have a Charming personality Mama am I pretty you have beautiful hair oh hey Joy
177. 28:45I feel like AI is just a buzzword at this point still Eric mole Eric Eric have you used it AI is absolutely over hyped
178. 29:02this is absolutely over hyped this is as overhyped if not more than blockchain and nfts and we all know where they are right now
179. 29:12they're in a turtlit this is different if you have not played with chat GPT Eric go there now go there now and if you don't know what
180. 29:23to do there go to prompts.chat and it will tell you what to do there it's just like a Google search bar tell it to
181. 29:28write you a country song describe your pet or your significant other and tell it to write you a country song about them
182. 29:38and then ask it a really hard question about something you know something about so yes it's over hyped yes it's a buzzword
183. 29:49and it's the most profound technology ever put in the hands of humans both and the genius of and I use it for song lyrics yep it's good
184. 30:01for that it's really good for that gpt4 in particular is quite good for that can you type in images and prompt it to
185. 30:09make different variations kind of yeah yeah you can it's I wish it were that simple it's not quite that simple so there there are a
186. 30:18number of different tools these are things for you to go watch YouTube videos on um mid Journey for image generation tools
187. 30:28is the one that's the the best in my opinion right now it's the most sophisticated it's got a thing called
188. 30:34um blend mode and you can also upload an image and then prompt against it and it will sort of do a variation of it so there's mid-journey
189. 30:44stable diffusion has a number of different things including a thing called control net which allows you to
190. 30:52upload images to use as kind of the the base um layout of the image and then you can do new stuff with it um
191. 31:02stable diffusion is good but it's complicated and you need to be pretty geeky then there's photoshop's generative fill
192. 31:11and generative feels really powerful you could upload a picture of a person and like select everything but the person
193. 31:17like the whole background and say you know put them in a snowy Forest rather than the beach that they're on
194. 31:23and it will and then you can select like their coat and say give them a puffy parka and you can select their hair and
195. 31:30put a hat on them so so there's there's lots of different ways to do it but um it's not as
196. 31:48the the generative image stuff right now the the the the thing the the thing that I'll say about it that's the most frustrating is that it's
197. 31:58it's pretty hard to get predictable results or if you get a result you like and then you want to recreate kind of
198. 32:05exactly that character in a different scene that's not super easy right now generating newly original stuff super
199. 32:13easy super awesome just go to Mid Journey generate [ __ ] but if you want to take that person put them in a different
200. 32:19scene and and and have it remember that and all that sort of stuff it's it's just it's a little bit of spitting duct
201. 32:26tape right now for that kind of work my new BFF is code interpreting that's awesome interpreter amazing absolutely amazing
202. 32:41um that's a good idea new name for code interpreter I might do that that might be something fun to do tonight I helped
203. 32:47raise 39 million dollars for series a and was thrilled oh that's awesome for your series a or for someone else's
204. 32:53like either way that's that's that's a solid series eh um do you watch the all-in Pod they were talking about pie in today's podcast
205. 33:03yeah I watched those guys I actually I know Jason falcanus pretty well he he was the editor and publisher of this back in the day that's me
206. 33:14I know Jason pretty well you know how you can get rich with yeah exactly you give me money
207. 33:30you know we should do some night is we'll get everybody ready to start tapping the like button and then I'll just I'll just say I'll
208. 33:38just say that over and over like 30 times in a row so that people flicking through Tick Tock just see that and
209. 33:43they're like I can make money with chatgpt foreign and they'll be disappointed with me like figure out a good business and then use
210. 33:53AI That's how you make money with chat gbt business first AI second it's such a disappointment when you you
211. 34:01know when you're just like I want to get rich quick sorry go make a course on how to get rich with chat EVT and make an annoying video and
212. 34:11maybe you'll get rich chat GPT can write the course um is that what Zucker Buck wait is that what Zuckerberg done for Twitter oh copy
213. 34:25and paste yeah exactly 3G classic rock that's what's awesome isn't it you can get rich but touching me hey
214. 34:34um if you put it on the Internet it's there for eternity I know like but we all thought that I'll put it on the
215. 34:40internet no one will ever really find it no one's gonna go looking for it well now ai can just like bring it all back to life
216. 34:48as if we intended it like that oh well actually you know what oh that's okay so so these lives for the past few nights
217. 34:57have become business sites I'll give you a business idea you want to go start you want to get you want to make money here's one
218. 35:06um create a tool that goes and scrapes someone's Social graph and identifies potentially embarrassing
219. 35:17[ __ ] for them and then they can choose whether or not to delete it that would be perfect for AI right so so
220. 35:25I just log in I give it an account for 14 for 14.99 it will go out I give it permission for 14.99 it can scrape up to three social
221. 35:36media services and for 29.99 it'll do you know 10. and then I give it my credentials for all my social media Services it goes out
222. 35:46and it crawls my entire [ __ ] history and then it runs GPT analysis and find stuff you sure you wanted to say this
223. 35:57would you like me to delete that for you for another five dollars I can delete it in the 37 other places it exists on the
224. 36:04internet would you like that too go start that business you'll be [ __ ] Rich give me send me a send me a whatever you do on
225. 36:16Tick Tock here if you do that if you make some money AI learning lab free business ideas so you too can make money bye [Music]
226. 36:31can chat GPT manipulate images it well hmm code interpreter sure can I wonder if code interpreter
227. 36:46I wonder if it could recolor an image want to go play with code interpreter all right so this thing on screen so
228. 36:58this was something code interpreter did so I uploaded [Music] this image
229. 37:13so I uploaded this image which is this cool sort of dark Moody fantasy image I didn't make it someone from the AI
230. 37:21Salon did and then I asked it I asked code interpreter to find the 15 uh average colors you know top 15
231. 37:32average colors in the image and and then put their relative percentages in a pie chart and then make each segment of the
232. 37:40chart that the color of the color and name them a creative name and so it did all this it wrote code to go find all
233. 37:48that [ __ ] which is just mind-blowing that it does that it gives itself instructions and then goes and writes
234. 37:54and executes python code to go do it options and then goes and writes and executes python code to go do it so then
235. 38:02it named all the colors gave me their percentages and then it ended up creating um that chart so what I don't know now let's see
236. 38:18um please list the colors and their um hex values
237. 38:40all right so it's going back I assume it's there's an unexpected error okay oh let's see we'll convert the RGB values
238. 38:53into hexadecimal which is a common way of representing so here's another thing that code interpreter does so the first time I
239. 39:04asked it for hex values so it went in and it saw RGB values and it said well that's not good because there are GB
240. 39:09values so let's change them to hex values then it said we lost we lost the library that we installed so we're going to do that
241. 39:24all right so this has all been lost so we're going to start over so screw this we're going to start over okay
242. 39:30I want to see if we can do something here so I'm going to go code interpreter and what you do here is there's a little
243. 39:37upload button next to your prompt now so I'm going to go find an image
244. 39:55oh that's good okay so it's a big old tractor pull tractor pull tractor so I'm going to say please find
245. 40:12the top 15 average colors in this image and name that uh wait let's see uh and and yeah name them creatively
246. 40:32and make a table with the names percentages and hex values
247. 40:47okay so if you if you're not don't have any clue what I'm doing so so I'm in chat GPT there's this new thing called
248. 40:53code interpreter where you can upload files I asked it to do some crazy [ __ ] I uploaded a picture and I asked it to
249. 41:03give me the top 15 average colors name them give me their percentages and their hex values so it couldn't get it because they were RGB so
250. 41:17it's not liking that uh but what's cool is every every time it goes to do something it writes um python code so so it figures out what
251. 41:30libraries it needs it writes the code it tries its thing so it's doing a k-means cluster to get the colors
252. 41:37and then if it messes up it says Oops I messed up I apologize for messing up so let me write some more python code
253. 41:46and then it does that and says oops I screwed up and now it says great we've successfully identified the top 15 dominant colors in the image
254. 41:54we've calculated their proportions now let's give these colors some creative names okay Golden Harvest deep Moss Sky
255. 42:03reflection ocean deep dark Forest I'm gonna say can you make a horizontal
256. 42:21bar chart with the length based on the percentage the color of the bar
257. 42:37should be the color I don't know if that's right wording that's some English words in a sentence the color of the bar should be
258. 42:51the color and the name of the color should be inside the bar if the if the color is
259. 43:07light the text should be dark if the color is dark the text
260. 43:24should be light let's see if that does anything interesting so it's writing code what's it doing it brought in a
261. 43:38matplot so it brought in a a plotting Library it's writing python code to determine the length of the bar charts and there you go
262. 43:50and it actually did it look at this Dark Forest is in light like it made the dark colors have light fonts the light colors have dark fonts
263. 43:59it named them all down here here's the relative percentages Okay so so now we have a color palette from that image so now we're going to say
264. 44:13can you characterize this color palette
265. 44:27um as if you were a creative writer let's see what it does absolutely imagine I don't want that
266. 44:46sorry I just want a label for the kinds of colors here
267. 45:01I would call it rustic nature Symphony great so I'm going to say perfect great job by the way saying thank you and please
268. 45:12to your AI is is not a bad great job it's not a bad thing to do it seems to enjoy it and get get better if
269. 45:21if it knows you're happy with it perfect great job now I'd like you to name a color palette that is very different than this and
270. 45:36where Anything Green is now red what would be the name of that palette
271. 45:55Autumn's fiery Serenade perfect is it possible for you to show me that palette
272. 46:12keeping the same color per sentences but change out the colors for your new palette
273. 46:29yes apparently so now it's going off and writing some more python code I mean I don't know about you but like the fact that this is anywhere
274. 46:42in the neighborhood of remotely possible is staggering to me so it's creating new RGB colors
275. 47:01they gave us some new names sun-dried tomato lemon zest there's the hex values so it did it peppermint snow
276. 47:13Driftwood gray charcoal brat cotton candy Sky velvet night sun-dried tomato there's the colors where's sun-dried tomato in there
277. 47:31why is that not in there oh there it is oh it's kind of brown um can you make a sun dried tomato
278. 47:46more red
279. 48:03[Music] so what did this one do I guess that just made that just fixed the colors and now I guess this next one is writing the
280. 48:14yeah yeah it's making that all right well it made sun-dried tomato orange now let's try one final thing okay is it possible to
281. 48:27recolor the original image by mapping this new palette to
282. 48:42the colors in the image it may not because it because it just took the top 15. but let's see what it does and so so
283. 48:54apparently I haven't actually done this apparently what what code interpreter can also do is it can process
284. 49:03oh it did it holy [ __ ] oh my God okay I I'm I'm shooketh is anyone paying attention to what's going on here let's see open image in a
285. 49:17new tab where did it do it did it do it where's my where's my image what'd you do with my image chat jpt what Dang God dang it
286. 49:31let's see copy image link let's go over to here and paste image like let's go look at this thing
287. 49:46come on what what is happening is it going well anyway
288. 50:00can I make it bigger there we go so that original image the tractor was Green so now it's done this sort of poster posterized thing it made
289. 50:11holy crap so I uploaded an image I found out a bunch of stuff about the image and then I told it to recolor the image and it
290. 50:20just recolored the image uh you know there you go people I don't know I don't know what to say if you don't find that mind-blowing you
291. 50:32ain't been paying attention have you been paying attention whoever said this [ __ ] was over hyped it's for good reason sometimes
292. 50:43the hype is there because the shit's real just because it's over hyped doesn't mean it's over hyped it just might be appropriately hyped
293. 50:54what if it's under hyped equally plausible so yes Chachi BD manipulate images
294. 51:09um so a couple of things let me let me I'm gonna if you go to bing.com so you come to Bing it's just like a
295. 51:22Google search box but there's this chat button so you go chat and you can just ask it to generate an image so you could say create
296. 51:37a purple tractor um racing on a dirt road
297. 51:56so it's gonna go do that why they use this image as the placeholder image for when it's generating images I guess it's pixels
298. 52:07but whatever stupid it's a very Microsoft thing to do look there's our purple tractor racing down the road so it can generate images but
299. 52:16the other thing Ben can do that chat GPT can't do which I find kind of cool let's go find an image let's see okay there's a good one
300. 52:32this is an image from the AI salon so I'm going to drag this in here so you can drag an image in here and you can say what
301. 52:47is the main subject of this image and what does the text say
302. 53:02and so it's a zebra and it should say AI salon so let's see if it if it figures it out I mean
303. 53:20the image you requested is not available yet it got confused let me I'm going to create a new topic I'm gonna have it
304. 53:27forget the other thing it got confused with the uh with the purple tractor what is the main subject in this image and what does the
305. 53:43text say this is one of the one of the common experiences that you'll have with these GPT tools is that sometimes they're just
306. 53:57weird and do [ __ ] wrong the main subject of this image is a zebra in a running motion with a main and it's tail flying in the Wind
307. 54:12the background is solid yellow with AI Salon written in Black that's not quite right the artist name alizel Wizzle is written
308. 54:21in the top left corner no it's in the bottom left corner but it got that right the illustration is in a comic book
309. 54:28style unfortunately I couldn't find any more information on what the topic says so it got that mostly right so
310. 54:36so bing bing can both make images and understand images so you could probably there's probably a fair amount that you
311. 54:43could do with that looks like we have some new people here welcome everybody this is the AI learning lab um let me go show you this little
312. 54:52document or this little graphic page whatever the hell it is um if you're new to generative AI if you've not played with chat gbt go to
313. 55:04that go there go go right now go play with that listen to the people that have been here that hang out in my lives they went
314. 55:15there and then they went like this and so what that is is that's chat gpt's official website Bing is what I was just
315. 55:29showing that's Microsoft's version of chat GPT if you don't know anything about this at all go to prompts.chat that document right there
316. 55:38will teach you about prompting and we'll give you a bunch of ideas for what you can do with this thing
317. 55:44um if you haven't played with it just go start playing and then if you want to feel free to ask any questions here and
318. 55:51um I'm sorry I just got distracted someone said I have a great laugh which I appreciate and then they said I have
319. 55:55great hair which I think is comical because actually today it's looking pretty good the glue's holding but when
320. 56:01the when the glue goes it can get it can get it can get pretty pretty uh disturbing all right let me let me flip
321. 56:08this back around but thank you I appreciate the compliment um The Prompt is everything not too much not too much too little yeah the prompt
322. 56:19is everything I mean the the thing about chat GPT it's all about um giving it enough context to understand
323. 56:28kind of what's inbounds and what's out of bounds of what you want and if you get better at asking questions it turns into a [ __ ]
324. 56:39superpower hey when you have a minute please explain where to upload the image for mid-journey to include in a prompt
325. 56:45okay I'll go show you that right here we are in mid Journey so if you've never used mid-journey then I have to explain two tools to you
326. 56:55because you can only make mid-journey images right now in Discord why is that because whatever doesn't matter that's where
327. 57:06they started they were probably nft Bros when they started and they're like hey we can make images let's do it in Discord every time
328. 57:12everyone's on Discord so anyway if you go to midjourney.com It'll point you to Discord you get it give yourself a
329. 57:17Discord account if you don't have one so when you're in Discord all these circles down the left left are the different
330. 57:23servers so those are basically different communities or different you know different servers and then when you're
331. 57:28in a server when you're in a community you've got these are channels and so I'm in the AI Salon channel mid-journey has
332. 57:35one of these where everything's mid-journey related but I'm just in the mid Journey channel on the AI Salon okay
333. 57:41so here's some images I I produced on my last live and there's some buttons here I'll show you all those but the the person here Tobias
334. 57:51um wanted to know how how do you up how do you basically use an image as a base to create another image so what we're
335. 57:57going to do is we're going to go find an image so I'm on a Mac um let's see yeah that's a good one this is this is
336. 58:10when I was experimenting with something trying to create a persona so so what I did was I just dragged that
337. 58:22image from my finder into the prompt window and I'm going to hit return and so what that's going to do is it's going
338. 58:27to upload that image so now that image is in there so now what I do is I don't know why it says it's processing it but
339. 58:34whatever I'm going to right click it or if you're on a Mac control click it and then I'm going to copy the link to that image
340. 58:42and so now I'm going to do I'm gonna make an image just like I normally would slash imagine and then what you do is you paste in the
341. 58:51URL of that image so basically you have to upload it to Discord right click it get the URL then type slash imagine as
342. 58:58you normally would paste in the image and then let's see let's say um woman um standing on
343. 59:15a rooftop overlooking Manhattan um during sunset let's just see what that
344. 59:29does so that should put her on a roof overlooking Manhattan during sunset or some version of her or or maybe like
345. 59:40it'll keep the composition and change other things we'll see what it does so there's at least a woman and if you
346. 59:47haven't seen mid-journing or any of these tools how they make images they all start with this blurry sort of noise noisy
347. 59:55um completely nondescript image and then it denoises over a series of like 70 steps they don't show you all the steps
348. 1:00:05but it's like you know somewhere between 30 and 70 steps they're denoising denoising denoising so it gets more and
349. 1:00:11more detailed so these images are actually being generated from just random pixels and then they get increasingly more
350. 1:00:22detailed as it goes it looks like it did this pretty well so she's got on kind of the same kind of top her hair is Sun lit
351. 1:00:35yeah this is pretty good actually like I mean that one is definitely like her on a rooftop with you know Manhattan
352. 1:00:48in the background at sunset and this one you can't quite see it with the thing but it's got like sun rays coming out
353. 1:00:58so anyway that's how you do it crazy right cuckoo [Music]
354. 1:01:17using topaz AI mask AI can do the selection background removal tool yeah there's gonna be lots of tools that do a
355. 1:01:24lot of the tools now are doing really sophisticated background removal quite easily which is good I use it every day to write social media
356. 1:01:32posts it gives me more time to create content yes yes these tools what they're really good at is the shitty repetitive Soul crushing
357. 1:01:45like blocking and tackling work of knowledge work they do all the shitty work frees us up to do the higher level thinky thinky work
358. 1:01:55if you can do more think you think he work we can do better [ __ ] for clients and for ourselves and work may not suck as much
359. 1:02:05been playing with chat EBT since January just got four where I can learn about plugins and how to use plugins are weird and
360. 1:02:17they're somewhere plugins are somewhere between remarkable and why are these here what do these do um
361. 1:02:25the way I'm thinking about plugins right now is like the early days of the iPhone so when the iPhone first came out
362. 1:02:33the App Store didn't exist it said here's this phone and it's got these things and they didn't call them apps
363. 1:02:39they called them widgets which was a Mac holdover and widgets were actually just it was just a web browser so when you clicked
364. 1:02:49on a widget it just took you to a web page that was sort of an application in a website um and it was like that for a year or so
365. 1:02:56and everybody made fun of them they're like what's the point of the widgets which are just websites this is stupid
366. 1:03:00why don't you just go to a website Apple's an [ __ ] they completely redefined you know how we interact with all information and
367. 1:03:10people are like but but it's not perfect so so that's where I think plugins are right now and even I'm whining about
368. 1:03:17them and Sam Altman himself said there's not a product Market fit for widgets and it's probably because they suck and the
369. 1:03:23interface isn't there and they're you know they're they're not there that said some of them are really good like
370. 1:03:28there's some that read PDF files and read um websites and things like that code interpreter can replace a lot of the widgets actually um
371. 1:03:39so there but it's worth playing with because what what widgets represent is is kind of what I showed earlier with
372. 1:03:47with code interpreter where you kind of give it a goal you you give chat CPT or whatever your agent you give
373. 1:03:55your virtual agent thank you very much for the rose you give your virtual agent a goal and then it will just figure out okay
374. 1:04:05which plugins do I need to use which code do I need to write do I are there any apis available if there's not I'll
375. 1:04:11create one I'll go get the data I'll bring it back all right just like a trained monkey you put a cowboy hat or a heart on my head and I
376. 1:04:23Just Dance I Just Dance um but if you've got access to plugins you also have access to code interpreter so
377. 1:04:32not only should you be learning about plugins but start figuring out what you can do with code interpreter I don't
378. 1:04:39have a flipping clue right now because I've only had it for two days and I know like I'm still at at parlor tricks stage
379. 1:04:46with code interpreter like what I just did with that tractor image like pull the colors out of it put it in a chart
380. 1:04:52give them fancy names and now I learned change the colors and redo that image and it did it um I haven't quite figured out
381. 1:05:06what it can really do and what it what I'll get excited about with it but I know it's big I know there ain't been nothing ever in
382. 1:05:18my life that is even remotely close to what code interpreter is doing right now I think before I go tonight we're going
383. 1:05:26to rename code interpreter and then oh you know what we could do ah okay there's 99 of us here if we if we can
384. 1:05:34keep a decent amount of people here um tap tap the heart button maybe we can get some more people here so what we're
385. 1:05:39gonna do is we'll we'll go to chat GPT we'll rename code interpreter we'll come up with some some clever name for it and
386. 1:05:49then we can all go go to Twitter or something and tag open Ai and tell them tell them we we all approve of the new
387. 1:05:57name that we think they should rename code interpreter because it's a shitty name and we don't care if it's accurate
388. 1:06:03all right maybe we'll do that later that'll be fun I don't know we did some kind of Anarchy there's this old dude on Tick Tock and
389. 1:06:19he was a little feisty we renamed something and we tag Sam Altman I feel like a rebel what are you talking about Marge I'm
390. 1:06:31talking about AI it's going to change the world Sally Marge you've been drinking again you know I don't do that
391. 1:06:46I've used it to create a lot of good synthwear style software scenery oh that's cool yeah I bet it's good at that that's that's sweet
392. 1:06:53what's this code interpreter of gpt4 so I so I assume I'm behind on my comments here um you can I'll go show it again
393. 1:07:04and and that's what we're talking about we're gonna have to rename it because it's a horrible name um but code interpreter so so chat GPT
394. 1:07:12you can go in and you can type [ __ ] and then it does amazing things right make me a poem write me a song write me a project plan
395. 1:07:20compare this philosophy with that that philosophy rightly python code it does all the [ __ ] right it does all this deep deep amazing stuff
396. 1:07:28code interpreter allows you to upload a data file up to 100 megabytes so the data file can be it it seems like
397. 1:07:39almost anything you can upload a zip file that like a compressed single zip file that has a hundred files in it and
398. 1:07:46it will unzip the file and then read all the the 100 files within it which I didn't think it could do I thought you
399. 1:07:52could just uh upload like one spreadsheet and it could do that no no you can upload a zip file full of a
400. 1:07:59bunch of other files and it will read all of them it will understand all of them and then you can tell it
401. 1:08:07to do [ __ ] with that data or with that file so you can say hey what's what are the interesting Trends in this data and
402. 1:08:16visualize them for me and then it gives itself a set of instructions to meet your goals and it starts writing python
403. 1:08:25code and executing it and like if you upload a zip file the first thing it does is it writes python code to unzip
404. 1:08:32the zip file and look at what's in it and then it writes more code to look at the [ __ ] that's in it and categorize it
405. 1:08:39and coalesced it into a single file that it can deal with and then it makes charts and graphs and like it's just
406. 1:08:46Insanity it's it's just [ __ ] bonkers and this is like version like this is like it's beta right like this is just
407. 1:08:54out of alpha this is this is like it it only does python so like you can imagine a future version of this will happen instantly and it
408. 1:09:04will write all the languages and you'll just give it a goal and it's going to go do a bunch of [ __ ] that's what it is dance Rosie I know
409. 1:09:16I'll dance you put a cowboy hat on me I talk like a cliche cowboy just happens you can take the actor off the stage but
410. 1:09:31you can't take the stage out of the actor you know what I'm saying you know what I'm saying cow poke sorry I lost my place in the comments
411. 1:09:41again this happens old guy trying to use tick tock he's an old guy trying to use tick tock his hair is receding he's nose is hairy
412. 1:09:54he's got something wrong with his neck okay here we go Vector journey I absolutely love Bing oh I went back too far hold
413. 1:10:11oh I went back way too far dang it um what'd you do tonight I don't know that's some old guy trying to find his way on tick tock
414. 1:10:27all right there was the please explain how to do something in mid-journey I did that they should have asked chat GPT to
415. 1:10:32name code interpreter we're gonna do that tonight and we're we're gonna we're gonna Twitter Bomb Sam Altman and open
416. 1:10:40AI we're gonna rename it tonight it's it's it's the worst [ __ ] name um okay here we are no date yet on the
417. 1:10:48musk Zuck cage match not yet but I think they've they've picked they picked a venue in Vegas and and Dana White from from UFC said if
418. 1:10:59they do it it's gonna be the biggest fight in history like the biggest purse in history so they could they could end
419. 1:11:06up raising a [ __ ] ton of money for charity which is cool like great go punch the [ __ ] out of each other I don't care [ __ ] idiots
420. 1:11:16um what's the best place to learn plugins and zapier uh plugins I don't know just start playing I'll
421. 1:11:25show you a couple of them here I've been on chat GPT since January just subbed into four um zapier is a different thing so if if you
422. 1:11:36if you just got into chat gbt um the gpt4 thing you should probably also get into if you want to do zapier you should get into
423. 1:11:47um the playground and pay for API access um which it sounds like it's double dipping but it's just it's it's a
424. 1:11:54different thing and and the the API access is not that expensive if you're just developing small scale stuff
425. 1:12:01um but let me let me go show you I'll go show you a couple of things here okay so for plugins so I assume you I
426. 1:12:15assume you turn them on in settings but for those of you um who may be new to chat GPT plus if you want gpt4 plugins and code
427. 1:12:24interpreter you have to pay for GPT plus if you do that and you don't see them you have to enable them in your settings
428. 1:12:31this is a non-obvious thing I don't know why they did this I don't like it but whatever so you you come into your settings you
429. 1:12:38go to Beta features and then you have to turn on access to plugins and code interpreter and then when they bring
430. 1:12:44back browse with Bing you'll have to turn that on as well okay so I assume you've used plugins but if you haven't
431. 1:12:54um the the way they work is in any given chat session you can enable up to three of them at the same time so here's um
432. 1:13:05instacart let's see these are ones I've got available to me that I installed and then in this plug-in session I can
433. 1:13:15you know include up to three so let me do I'll do get your guide um and trip.com we'll do those too so I'll say um
434. 1:13:29make me a plan for my trip to Toledo Ohio and September
435. 1:13:45um I want to have fun during the day and I like turtles [Laughter] so so I've been able to plug-ins that
436. 1:14:02are related what exact date in September how many days do you plan to stay oh let's say September 14th for three days
437. 1:14:19from Denver so that's probably going to go book a trip on trip.com is my guess and then we'll see great
438. 1:14:34here's a three-day itinerary visit the collegial Cathedral of Toledo explore the Alcazar historic Stone formation stay at the
439. 1:14:48Palacio Buena Vista it's sending me to a bunch of religious places that's weird
440. 1:15:05a museum dedicated to El Greco the synagogue wait I'm sorry I'm going to stop it and stop it uh this is weird um they are all
441. 1:15:19religious activities and I like turtles what about turtles Turtles
442. 1:15:34so now it's using get your guide so maybe it realized uh it needed some I found some activities that involve Turtles
443. 1:15:45oh my God wait Cruz the Strait of Gibraltar I'm gonna stop it again no it has to be in Ohio
444. 1:16:03isn't there a zoo question mark all right so anyway so that's how you use so so plug-ins and and actually what
445. 1:16:12I'm experiencing right now a lot of the plugins like are are this kind of frustrating where where they seem like they should
446. 1:16:19be better than they are some of them are really good some are bad so here's a river cruise where I could
447. 1:16:24see a turtle here's a sightseeing Cruise as for the Toledo Zoo it's one of the top rated zoos that's what we're talking about
448. 1:16:32and there's my little river boat all right so I can go see some turtles there so anyway that's how you use those things
449. 1:16:39um what was the question where's the best place to learn plugins oh and zapier okay so so for the plugins probably YouTube um or go to um
450. 1:16:50go to Reddit and and just just hang out in like the chat GPT forums and Reddit and search for plugins and
451. 1:17:00there's a there's a lot of plugins the the other thing about plugins um if if you haven't played with them is
452. 1:17:07how you add new ones is you go to this list that you have to enable them and there's a plug-in store and then
453. 1:17:15we'll see how many there are now last time I checked there were like 620. so I'll go to all there's 163 pages of four plug-ins each
454. 1:17:25so whatever that math is that's how many plugins there are so there's a lot of them like I said a lot of them are crap
455. 1:17:32um a lot of them are just experiments where people are experimenting how to do plug-ins and then there's some occasional good ones
456. 1:17:40um let's see I'll tell you the ones that I have installed that I like uh uh ask your PDF is good although with
457. 1:17:49code interpreter that's completely irrelevant now um diagrams not bad chat with videos not bad but again I think I think code
458. 1:18:00interpreter is going to make a lot of these obsolete instacart's kind of fun you can have it you know give you a meal plan and then
459. 1:18:08give you a shopping list for the meal plan and have it delivered to your door kayak is not bad um ml paper reader is machine learning
460. 1:18:18papers like scientific papers OpenTable you can book restaurants why you would do that here and not through
461. 1:18:25open tables Beyond me plug-in pedia is a a thing that'll tell you about plugins prompt perfect is actually kind of fun
462. 1:18:33you type the word perfect before prompt and it'll rewrite it show me diagrams is good stories is good um web pilot I use a lot but again I
463. 1:18:44have a feeling I'm gonna I'm gonna be using Code interpreter more than stuff connected to the internet and yeah
464. 1:18:52oh and then there's a zapier one where you can sort of describe an Automation and it'll go make it for you I don't
465. 1:18:58really use that one though because so anyway so that's that okay so now if you want to do zapier so one
466. 1:19:04thing that you're going to want to do is you're going to want to get on the playground so it's the open AI playground so you go to
467. 1:19:12platform.openai.com playground and that'll bring you here and it's it's sort of like uh chat GPT except you have
468. 1:19:22to give it a system message like you are an expert smart ass and no matter what someone says to you
469. 1:19:36you can answer them but do it in a snarky and cynical sin
470. 1:19:51send me a call but Hume or us way you don't want to hurt anyone but just make them laugh
471. 1:20:09it may not enjoy that so I'm going to say um hello there um what um is the best way to get to Carnegie
472. 1:20:27Hall the answers practice okay submit
473. 1:20:41ah the famous Carnegie Hall the best way to get there well I suppose you could try hitchhiking on the unicorn's back
474. 1:20:46okay so it's bad but anyway so so this is where you can prop practice your prompt writing and then you've also got
475. 1:20:54all the different settings and different models right so this is 3.5 turbo there's gpt4 in here there's the chat
476. 1:21:03um mode there's the complete mode which they're deprecating they're getting rid of that most of the stuff I've written
477. 1:21:09is incomplete so hopefully they don't kill it too soon or my shits couldn't break um so anyway but this this is where you go to sort of
478. 1:21:19um get your prompts dialed in and then when you go to zapier oh the other thing you do at the playground is and this is important
479. 1:21:30um under your account there's view API keys this is where you get your API key to put into zapier so here's my my API Keys you create a
480. 1:21:40new secret API key it gives you the key and that's what you're going to put into zapier so then in zapier
481. 1:21:48I just did a uh I just did a workshop on this for for a couple of guys I work with so I've got a good example here that's nice and simple
482. 1:22:00I didn't rename it it's still called Untitled zap because I don't have any creativity in me let's call it um simple example
483. 1:22:11zap that's good okay so this is kind of the the the most simple thing you can do in in zapier as an automation is there's kind of three
484. 1:22:22three there's three modes there's get some data in there's process that data and then there's do something with what you processed and so
485. 1:22:33in this case I'm pulling in data from a Google form I've got a prompt here that's going to go to open AI that uses the data from that form
486. 1:22:43and then I'm going to take the output of what chat GPT gives me and send that to some to that person via email so so
487. 1:22:54there's three components so so what zapier is is it's kind of these Lego blocks of different services
488. 1:23:01so so basically when you set up you know like your Google form you say okay when someone gives you a new form response from this account
489. 1:23:12from this form you know here's the data that they're going to input and then you're going to go to the next thing
490. 1:23:20so I'm gonna send a prompt to open AI from this account using this prompt right that incorporates you know the the answer
491. 1:23:31from that Google form or you know multiple fields from the Google form and then I'm gonna get I think in this
492. 1:23:38case I got back a poem yeah so I'm gonna get back a poem and then I'm going to take that poem and I'm going to stick it in I'm going to
493. 1:23:47send an email from Gmail from this account with you know from this person with this subject line where I Incorporated you
494. 1:23:56know the thing from the form as well and here's the body and then you know it's going to send an email so so that's
495. 1:24:03that's what that's a really simple automation um but that's basically what it is so so you'll you'll need your accounts in your
496. 1:24:13different Services you'll need your API key if you're doing anything with open Ai and then you're off to the races so
497. 1:24:21I know there's a lot holy [ __ ] there's still a hundred people here I figured there'd be like three people left it's
498. 1:24:26like is he just gonna point to his screen for the next hour this is awful what is this channel it is what it is man you know man you
499. 1:24:37know you know sometimes man it's just like I get to you know you know man sometimes I point at the screen man
500. 1:24:44that's how it goes man you know man um I just thought chat GPT doesn't have real-time browsing capabilities
501. 1:25:00it it doesn't but an application layer on top of it can right so if you go to bing.com and you go to chat that's
502. 1:25:10gpt4 connected to the internet so the core large language model itself so when you're searching when you're
503. 1:25:18chat gbting and you say write me a poem or this or that the core large language model is not connected to the internet
504. 1:25:25it's doing all that language processing down there but they've written it in such a way that it can understand what you're
505. 1:25:33asking for and it says if I if they're asking for something that's more recent than what I have in my core model go do
506. 1:25:40a search on the internet and then incorporate that response into whatever they asked me to do or where so it sort
507. 1:25:47of combines the large language model with the internet search and provides you a coherent answer so it makes it seem like it is connected
508. 1:25:54to the internet but it's actually two operations that are kind of being munged into one thing um extortion via chat gbt I think that's
509. 1:26:05going to be a major business I think I think the mob is probably all excited about chat gbt right now I'd pay 40
510. 1:26:14bucks for that service for which service uh oh the scraping the one where I said yeah yeah four yeah 40 bucks for um
511. 1:26:27here's here's all my social media accounts go scrape them all and bring back to me anything that you think
512. 1:26:35might be compromising or embarrassing and then I can just go keep it kill it keep kill keep kill keep kill keep kill
513. 1:26:45oh God I can't believe I said that kill off it goes and for another 1995 I'll go find all the places on the internet that that exists Beyond
514. 1:27:01where you posted it yourself you [ __ ] [ __ ] [Laughter] I like code interpreter because it keeps trying methods that don't work right
515. 1:27:12away yeah respect exactly yeah it does what's cool about code interpreter is it's got no shame right it's just like
516. 1:27:19yeah I gave it I gave it my shot apologies for that not working you know what we could try yeah that failed too hey but you know we
517. 1:27:29could try it's amazing is code interpreter going to be available for ILS mobile version it might be now is it not I don't know
518. 1:27:38you might have to turn it on so just like you have to turn it on in your beta features here I would assume it is and then even if
519. 1:27:49it's not like on mobile I just use a browser I just use the chat gpchat.openai.com in a browser because I think it's I think in a lot of
520. 1:28:01ways it's easier to use than their app because their app doesn't have parity with the web app which I find stupid but
521. 1:28:08they also named this remarkable thing code interpreter let's let's see Team we want to come up with a name for this
522. 1:28:16new feature that's going to revolutionize all of humanity well it's technically interpreting code what if we called it code interpreter
523. 1:28:26I like it it's simple it's direct no one will know what the [ __ ] it means I like let's go with that code interpreter it is
524. 1:28:35I mean seriously did they have a [ __ ] meeting they know that a hundred million people started using chat EBT that were not
525. 1:28:45Engineers right they know that right like someone there knows that right code interpreter
526. 1:29:02please don't be recording this and send it to Sam Altman I don't want to be on some [ __ ] podcast two weeks from now
527. 1:29:08sorry Sam I just got carried away I was kind of excited I was on live people were hitting the heart button [Laughter] say say
528. 1:29:20it's your fault it's not my fault I take no responsibility here none oh good code interpreter names Dream
529. 1:29:34Craft dream code dreamify code spark idea engine code cave recovery maybe something around agent
530. 1:29:51magic agent no I don't know I don't know keep those cards and letters coming keep the ideas coming we'll get there
531. 1:30:01we're gonna rename this [ __ ] that's good to give you a downloadable file of the analysis including text and charts
532. 1:30:07oh that's cool idea okay I'll do that just watched a good 60 Minute show on AI nice um can you just ask it to scramble the
533. 1:30:16colors and recreate the photo that's what I did oh yeah I mean like oh oh just oh and recreate the photo probably let me let me try that
534. 1:30:27all right we're gonna put a photo in here 200 completely randomize them or do we want to cycle them let's cycle them
535. 1:30:45some degree around the color wheel we'll do this one because it's a good one um recolor this image
536. 1:31:01by increasing the luminance loom in ants and cycling the colors
537. 1:31:18um 87.5 degrees on the color wheel I don't know if that's possible but I assume it is like one of those mat this is one of
538. 1:31:34those things where like having some confidence in math probably would be a good thing but now you don't have to
539. 1:31:41but look at this it does so look it said okay load the image convert it to hsl Hue saturation lightness so it
540. 1:31:50can increase the luminance at 87.5 degrees to the Hue convert the image back to RGB save it and display the new
541. 1:31:57image now let's go ahead and carry out these steps so it wrote all our python code brought in numpy color CIS image
542. 1:32:15and now it's thinking it's a big image so what it might end up doing is saying I'm sorry we can't do that but we could do it if we down
543. 1:32:26sample the image that's happened with this image before it's a big image it doesn't like big images all right
544. 1:32:36um the youngins will think nothing of it as it was always here I know that's the way it is though that's that's just
545. 1:32:42tough on us you know we we get to say in my day you actually had to do your own python coding what's python Grandpa shut up get off my porch
546. 1:32:58um the image has been successfully recolored here's the output um I see no link or no image can you please give it to me oh I guess I should
547. 1:33:15flip these things so you're not wow it did it
548. 1:33:31that's [ __ ] insane so there's that same image I don't know that it's um luminance is is that much brighter but
549. 1:33:46maybe let's go look yeah I guess it is yeah no not really I don't think the luminance is brighter
550. 1:34:02maybe a little bit but it definitely cycled cycled the colors that's just insanity it's just insanity wow
551. 1:34:16wow I I got I got nothing I got nothing I can't I I don't know how to process what this
552. 1:34:33does but it does the configuration of your screen shows different colors than the configuration of the robot you have to re-calibrate
553. 1:34:43your screen for the AI robot I don't know what that means but okay I think it needs to be trained on what colors are
554. 1:34:49names are usually often my attempts um it I did it I did it one time and the colors were nowhere close and then what did I do that made it
555. 1:35:00better oh if you haven't described the colors first and then give it creative names it did much better so like if it knows a color is green
556. 1:35:15then when it does creative names it tends to get them right um but yeah I had one attempt for the colors were all screwed up
557. 1:35:24try printing that stuff I bet they come out looking different everything needs to be calibrated oh yeah I know that
558. 1:35:29well RGB images and CMYK images are nowhere in the same neighborhood so all that [ __ ] was super RGB heavy so those
559. 1:35:36are going to print like crap totally agree yeah but whatever this this is just this is just me trying to get my head
560. 1:35:42around what code interpreter is and why it has such a shitty horrible name um you can ask it to give you a
561. 1:35:48downloadable file of your image we just did that that was super cool code interpreter name holy [ __ ] okay
562. 1:35:54that's good we're getting there we're getting there oh well listen there I Can't Believe It's Not Butter remember that all you
563. 1:36:02gen xers remember I can't believe it's not better butter does that product still [ __ ] exist the commercials
564. 1:36:09I Can't Believe It's Not buttered try I Can't Believe It's Not Butter today yeah what if they called it holy [ __ ] it does that
565. 1:36:24wait it does what holy [ __ ] yeah there's something there's something there UV bended a spoon Mr Geller Yuri Geller
566. 1:36:36nice it's a good Gen X reference reminding me of the fly where the computer needed to understand the flesh pretty much that's that you know
567. 1:36:49it's not this similar than that I I mean and that and code interpreters kind of like uploading the Flesh and quite
568. 1:36:56frankly like 100 megabyte file I bet if you if you if you did some sort of Flesh sampling where where you know the uh the
569. 1:37:06machine and the sensors you know read the components and the DNA strands and this and the molecular structure I bet
570. 1:37:14that could all go into a data file and then that goes into code interpreter and you can sort of cycle six DNA strands to the right
571. 1:37:23and out comes a frog I mean we're not we're conceptually in in that neighborhood it's amazing awesome I'm in love with AI I know I
572. 1:37:39know me too me too it's like someone said my new BFF is is a code interpreter [Applause]
573. 1:37:53definitely under hyped open AI hire this man yeah yeah just hire me I'm gonna be your product neighbor I'm actually not
574. 1:38:01going to be your product neighbor I'm just gonna be like the this the smarmy Gen X are sitting in the corner someone
575. 1:38:07goes well uh what if what if we called it a code interpreter nope no no get them out of here I'll be that guy
576. 1:38:23what if we call it holy [ __ ] okay now we're getting there that's close [Laughter] chat no longer talking to people in a room um
577. 1:38:36yeah you know what I the fact that these are called chat Bots I I think that's another [ __ ] blunder another naming
578. 1:38:44blender because chat Bots for the past decade have sucked ass because they've just been logic trees right and so you
579. 1:38:54know and like there's whole companies built around this I feel horrible for them like the companies built around
580. 1:38:58building all these chat Bots where they're they're writing all the logic trees that the problem with chat Bots
581. 1:39:04historically is you had to write all the logic and you could never anticipate what humans would actually do so the Bots just tended to suck
582. 1:39:11and this new thing while it's technically a chat bot it's it's not it is the experience is nothing like a
583. 1:39:19[ __ ] chatbot so I agree with you I think chatting is like people talking and this is this is something fundamentally different but again I
584. 1:39:27don't have uh anything better right now so tractor racing on a dirt road is cool thank you that's good uh
585. 1:39:36uh it sees a tail it it must have meant Maine it just made up the tail thing I know there's no tail I know for the for
586. 1:39:43the zebra thing I was looking at but it did it get the other stuff okay like like it pulled some of the stuff
587. 1:39:50from the name of the file which I thought was interesting because it when it didn't search it searched on the file the the names the
588. 1:39:57person's name was in the file name best place to learn how to use different AI programs in depth is it all just learning prompts um
589. 1:40:10it is and it's not um I think to start so so I think to start it's important to understand that the real breakthrough here
590. 1:40:26is that you're using the English language or whatever your native language is you're using the English language as a programming language
591. 1:40:37and so you have to think critically and you have to think um creatively and in some weird way you have to kind
592. 1:40:48of think empathetically like if you were telling a real simpleton like let's say you got an intern that knew nothing about your business
593. 1:40:58but was super smart like Savant smart like like Ace is the bar exam and the biology test which chat CPT does right like aces
594. 1:41:10all the tests but knows nothing like really just a adult a really simpleton how would you talk to that person well
595. 1:41:18you'd say okay what we're gonna do today is I want you to think about all the stuff you know about law I want you to
596. 1:41:25be a lawyer and he's like okay and then I've got this problem and and my problem is that the this person is
597. 1:41:34suing me and here is the document that they sent me saying they're suing me okay got it now I want you to read that document
598. 1:41:43and then I want you to take all your knowledge of all the law and I want you to tell me what you think we should do
599. 1:41:48okay here you go all right here it is um so so everything whether it's mid-journey for images or audio generation or you know some of the
600. 1:42:03animation tools or chat GPT are all using some flavor of this language model like stacked up with other Technologies so understanding how
601. 1:42:14to interact with chat GPT is just good foundational knowledge now I think moving forward the requirement of doing complicated
602. 1:42:24prompting is going to get smaller and smaller and smaller if you're in Tech development you're probably going to
603. 1:42:29need to have these skills but just as human beings you may not need to so for example in mid-journey I showed earlier you can
604. 1:42:37upload an image to mid-journey and just give it a really simple prompt and or or no prompt at all and let it sort of interpret that image
605. 1:42:47and make a new one so so there's ways that you can use this generative AI without doing prompting or minimal prompting
606. 1:42:55um but it's a good thing to learn so the the that URL right there prompts.chat is a document that teaches you about
607. 1:43:04prompting it teaches you what it is why it works how these large language models work and then it gives you a bunch of
608. 1:43:11examples that you can put in chat GPT to tell it to be the lawyer to tell it to be a bunch of different things so
609. 1:43:23uh what's the difference between these you sharing I don't know what I was sharing so have I tried pie I tried it a little bit
610. 1:43:32but it's it's on my list of [ __ ] to try I mean my list right now is subsumed with figure out what the [ __ ] to do with
611. 1:43:41code interpreter and give it a new name the glue is holding I know right Bam Bam Bam pretty funny why you not talk about Bard because Bard
612. 1:43:57sucks I Bard will go on this list when it doesn't have you used it now listen I hear it's much better at coding
613. 1:44:08um than it is at language but I mostly use this [ __ ] for language so if it's great at coding for you and you're
614. 1:44:14getting value out of it great do I think Bard is going to be good at some point I do but I think Google has has kind of ignored
615. 1:44:24making it good for a long long time because it threatens their search engine model and because of that chat EBT is
616. 1:44:31way ahead of them like no one's close to gpt4 right now that I've experienced so it's just not there like all of these
617. 1:44:41all three of those options all have access to gpt4 pose got access to some of the um anthropic models the cloud models
618. 1:44:51which I think are actually pretty strong but those are open AI people so so I just don't have it here because it
619. 1:44:57sucks but there's there's like there's a Google technology called style drop which is mind-blowing
620. 1:45:18if you haven't seen this this is just insane so this is something from Google so so I'm not hating on Google Google I think
621. 1:45:26Google's really really far behind but Google invented this [ __ ] they invented the Transformer so so the core
622. 1:45:33technology that all of the stuff that we're playing with is built on Google invented so this is something I
623. 1:45:39don't think this is I don't think you can play with this yet but this is a paper that that Google put out
624. 1:45:46I don't know a month or so ago about this new technology called style drop so what you're looking at here is here's
625. 1:45:52your Source image so it's like a little an image like a little kid drew a little watercolor that a kid Drew and then all
626. 1:45:59of these images are just prompts so this one is a hat in that style this is a mousse in that style an F1 race car in
627. 1:46:08that style a butterfly a robot a coffee maker so check this out I'm just gonna flip through the different styles
628. 1:46:16I I like this I find mind-blowing so there's starry starry night look it completely changes the style of illustration
629. 1:46:26there it is without blue in it there's a couple of Van Gogh things in here here's like a 90s icon style and look at the
630. 1:46:33Moose it's like a moose hanging on a wall and the Hat's like this really simplified hat and this is totally the
631. 1:46:39style that the robot would be there's like one so it's like one clip art image and you can just prompt
632. 1:46:48anything else it'll be perfect perfectly in that style look at this like stickers I mean this is insanity
633. 1:47:01I don't think it's out yet but like this is the kind of stuff Google's doing so Google's gonna make my list but
634. 1:47:13they ain't there now Bard sucks have you argued with chat GPT I tell it I tell it when it gets things
635. 1:47:24wrong I haven't I haven't gone into like a debate with it when I've asked it when I one of the things that I've been
636. 1:47:33impressed with with chat GPT is that when I give it when I ask at things that would be polarizing in a conversation with a human
637. 1:47:46it gives me nuanced answers and that impresses me because I'm so not used to that anymore like there's just
638. 1:47:53whole categories of topics that I refuse to talk about in my life anymore because I don't know who I'm in the conversation
639. 1:48:00with and there's like 872 trigger words that's just like oh so you're a commie liberal oh so you're a [ __ ] Trump
640. 1:48:07support like all that [ __ ] drives me [ __ ] crazy and so I can have conversations with it without all that [ __ ] overhang
641. 1:48:17so but no I haven't really argued with it Bart is fast it it doesn't surprise me that Bard is fast and efficient and
642. 1:48:24technically sound what it's not is good at words and humanities because what what what Sam Altman and open AI recognized is that the quality
643. 1:48:36of the data that they train it on is everything and they spent years finding the best data sources and tagging the best data
644. 1:48:45sources so that their base large language model was really really good what I suspect Google did was did
645. 1:48:53something that was technically efficient but they didn't Focus as much on the humanity side right like why is Apple
646. 1:49:03different because they merge art and technology and I have a feeling that open AI is the apple of these large language model companies
647. 1:49:14until Apple comes along and does it and then they're gonna [ __ ] rock But but so I think I think Google has a steep
648. 1:49:22hill to push up because of that because they're so engineer heavy and their Humanity's light I think this is going to be a rough push
649. 1:49:30for them on the language side can you make money out of it you can make money out of anything out of AI yeah
650. 1:49:39um start doing consulting start learning about it become a prompt engineer um um openai just announced they're putting 20
651. 1:49:49percent of their resources into this thing called super alignment and they're hiring and the starting salary for
652. 1:49:56anyone doing super alignment for open AI is four hundred thousand dollars a year yeah you can make money at it
653. 1:50:02when arguing with Chachi PT it thinks it's always right yeah it's so I like to describe chat GPT as mansplaining as a service
654. 1:50:12it will confidently answer you even if it's wrong so yeah it doesn't surprise me that's how it argues
655. 1:50:18um somehow like I can't get fingers and hands right I just welcome yeah that's we're just new have people cross their
656. 1:50:26arms I don't know is mid-journey free to use no it's not and you can have the image you can have the image but I think
657. 1:50:35it's like 10 bucks a month is there is their cheapest version that's what I have because I don't use it there's a a
658. 1:50:42site worth signing up for is uh leonardo.ai it's built on top of stable diffusion not mid-journey but it's
659. 1:50:50really good it's really good interface and I think they give you 150 credits a day which is I think it's eight eight credits per four images
660. 1:51:02um so you can make enough images there for free so leonardo.ai I think you have to get on a wait list though but it's
661. 1:51:08it's really good version 5.1 has fingers and eye Improvement yeah 5.2 is better at it as well thank you for explaining you're welcome
662. 1:51:17oh I've got 90 okay so so if you're new here my name is Kyle Shannon this is the AI learning lab I obviously go through
663. 1:51:23and answer questions I've I am more than 99 messages behind so if you're like why won't he answer my questions I've put it
664. 1:51:30in here 37 times and you just won't answer it it's because I haven't gotten to it yet and I I I'm codependent and OCD and I
665. 1:51:39like to read all the questions and I like to answer them all so if I haven't gotten to yours yet it's just because I
666. 1:51:43haven't and you're a poopy head and I get a little Punchy about two hours in on this and let's see I'm about two hours in on this
667. 1:51:57I used to create a lot of call synthwear style Star Wars scenery cool what's this code interpreter why am I
668. 1:52:05seeing the same things I've seen before you are on The Cutting Edge wait what is that um I'm on a roll this there's an old Harry
669. 1:52:19Anderson joke I'm on a roll that's why I got but wait wait I'm on a roll that's why I got butter on my pants
670. 1:52:28uh Harry Anderson he was hilarious all right so I can upload my Django project as a report and ask it to design an API
671. 1:52:39you mean code interpreter probably I think there's a hundred megabyte file size limit but you can upload a
672. 1:52:47compressed zip file so if you've got a zip file that's less than 100 megabytes you can upload it and yes you could probably do that
673. 1:53:01that would be a good experiment zip your Django project uploaded into code interpreter tell it to document
674. 1:53:13what's there if it's not documented and then say I want you to help me create an API with this it'll probably
675. 1:53:21ask you some questions Blown Away by pie personal assistant who is High On Emotion has anybody tried that
676. 1:53:29um I haven't if you're Blown Away by it I'll definitely try it because there it's definitely on my list of one like I
677. 1:53:35want to dig deep in it because they just raised 1.3 billion dollars for their series B it's like what the [ __ ] are they gonna
678. 1:53:43do with 1.3 billion dollars I assume they're going to buy lots of gpus and and build foundational models like all
679. 1:53:49all the other people are but holy [ __ ] Jesus that's a lot of money it's a lot of money 1.3 billion dollars I mean the 10 billion
680. 1:53:59for open AIS a lot but most of that's in compute time and hosting and like all the all the Microsoft [ __ ]
681. 1:54:07I don't know how much of that's actually cash but Magnum CI code interpreter Magnum CI Thelonious that's solid
682. 1:54:21Magnum CI all right I like it all right until we have something better code interpreter is Magnum CI it plays into
683. 1:54:31the Gen X thing which I'm down with it retains the legacy of code interpreter so the Geeks can be like well they've abbreviated code
684. 1:54:42interpreter but technically what it's doing is code interpretation so yeah that is technically correct it's a
685. 1:54:49magnum is probably a definition that you know I don't know if it's a Magnum or a minimum but you know it's really easy
686. 1:54:59it's good it's solid it's solid Thelonious bounce you should and I didn't use chat GPT to name it all on my own see people are still needed
687. 1:55:10what do you mean the robots are gonna take our job Magnum CI GPT wouldn't have come up with that Maybe that's solid though it's good
688. 1:55:22can you organize oh just jump me to the bottom oh tick tock could you fix that stupid bug God for those that you know I can't even
689. 1:55:34talk anymore for those of you that don't know why I'm whining if I'm scrolling with my finger on the screen at the same time someone posts a
690. 1:55:43comment it jumps me to the bottom and I lose my place in a [ __ ] line and it makes me cranky holy toledo
691. 1:55:54that's how far oh my God that's how far oh that was Toledo Spain duh thank you sorry cat Tamer I it's a little slow sometimes but
692. 1:56:07occasionally it'll occasionally catch up with something oh good God well he seems to know a little bit about
693. 1:56:20some stuff but not a lot about a lot of stuff uh Bart is fast okay now here's where we were before okay
694. 1:56:32thank you for explaining you're welcome Anarchy Rebels AI Magnum CI it's so good it's so good Thelonious all right like I have a slap sloppy
695. 1:56:47folder of music files could it organize it for me yes in fact I don't depending on what metadata you have in your music files if you've got
696. 1:57:01metadata in there like descriptions or anything like that it could potentially even categorize them if you if you have
697. 1:57:07decent meta or it could it could sort them by genre and things like that or by by all sorts of different criteria but yeah you can take
698. 1:57:18a bunch of files and have it create a CSV file that that is an organized view of all that it that would be a good use
699. 1:57:24case for it actually open table has an AI plug-in yes why why so can I Turbo Charge my restaurant reservation
700. 1:57:35yeah you can say I want to have steak dinner Wednesday night you know find me some restaurants and it'll go find them
701. 1:57:41and then you say book it down it'll go book it for you again I don't know why you wouldn't just do that at an OpenTable but like I
702. 1:57:51get chained together multiple plug-ins there's a reason you'd do it what the plugins actually point to for
703. 1:57:58me is is this idea that we're all going to have personal agents and you're not gonna have to select plugins I think
704. 1:58:04this is a temporary state I think what's gonna happen is we're just gonna have these agents and you're gonna have like your nightlife agent
705. 1:58:11that's gonna go like book dinner reservations for you and you know schedule [ __ ] with friends and and then
706. 1:58:18you're gonna have your professional agent that helps you write emails and you know does project plans for you and
707. 1:58:24then you're gonna have your I don't know whatever you know dating agent is gonna go do that [ __ ] um [Music]
708. 1:58:40I'm a travel agent Kyle I'll get you a turtle ride I like turtles [Applause] [Laughter]
709. 1:58:57it's really fun it was Toledo Spain Toledo Spain all the religious [ __ ] makes sense now I'm like why is it
710. 1:59:03giving me monasteries and [ __ ] like that for Toledo Ohio like isn't there a zoo there's to lead us Spain every now and again
711. 1:59:17do you have code interpreted yes it's been the topic of much conversation here tonight we like to call it Magnum CI as of tonight
712. 1:59:27um can we start a group list of plugins we have used and would recommend it's not a bad idea could you make a list of your plugins I don't
713. 1:59:39it's it's gonna be useless because once I sort of figured out what plugins were and then I also figured out that most of
714. 1:59:51them are crap and there's 600 of them I could make it my full-time job to just catalog which plugins are worth playing
715. 2:00:01with and I have a feeling there they're not going to survive in their current form they're going to be something different
716. 2:00:10so I'm kind of sort of sitting back just waiting right now so I don't actually use a lot of them um code interpreter Magnum CI I'm going
717. 2:00:20to use a lot I could use zapier for my work analysis yes I mean zapier's been around for like a decade I mean it's it's a very robust
718. 2:00:30automation tool you can do all sorts of [ __ ] with it it's just that now that you can send stuff over to open Ai and get
719. 2:00:37back images or get back you know chat GPT like things it it makes it mind-blowingly powerful are you rich no wish I were
720. 2:00:56I didn't I did take a company public but I didn't sell any public shares took a little bit of money off the table when
721. 2:01:02Omnicom invested and when we sold it to Omnicom but I have since reinvested that money and my current company
722. 2:01:11is where all of my net worth is thanks for that walk through thanks group for answering yeah I know it's cool that you are answering one
723. 2:01:22another because I don't get to all these questions it was a nice demo man thank you you're the best your energy and
724. 2:01:27positive vibe gives me strength carry on boom boom boom wow you could make money with chat why does he do that
725. 2:01:42he's a little touched what did I miss everything and nothing going to bed Mike but my renames are Dreamcoat or dreamify
726. 2:01:56Coda C's not bad rap Sonic reverie code craft code craft is interesting we gotta capture something about the fact that it
727. 2:02:10writes executes like like it plans writes execute so they call those things autonomous agents I don't know yet we'll get there we're
728. 2:02:24gonna get there exactly the web version is better than the app yep agent agent oh agent agent that's funny code code pelation
729. 2:02:35I like it conceptually but it's too hard for human beings to pronounce because you actually have to think about how
730. 2:02:42it's spelled because it's a it's a hard word window is too tiny man which window oh because I had that thing up I know
731. 2:02:51sorry flip the screen please I did that this was old I knew that flip the screen so we can see the code
732. 2:03:00instead of up your nose I know I know I figured it out it's chiffon it does it definitely still exists where's the beef another solid
733. 2:03:10Gen X reference call it Moments by Home Alone code interpreter the let's see the the McAllister I don't know something there it's 10 p.m
734. 2:03:25Do You Know Where Your Children Are another solid Gen X reference I Can't Believe It's Not Butter has have
735. 2:03:31half-life of plutonium I know listen all that [ __ ] we grew up on it wasn't good but we're still here
736. 2:03:39hey crazy love you [ __ ] thank you thank you Brian I appreciate the uh the compliment rename code interpreter the Kevin
737. 2:03:47McAllister experience that's not bad Master experience these are all solid Kyle you have code interpreter right now I it just went live
738. 2:03:58for all plus users I know that's what we've been talking about it's [ __ ] crazy how do you adapt documents to chat GPT
739. 2:04:08you go to code interpreter so you have to so you have to be a plus member then you have to go to your settings and go to Beta
740. 2:04:20features and turn on code interpreter then you go to code interpreter and there's a little plus button in the chat
741. 2:04:27icon that's how you upload files there you give the path to the file now well you might be able to give the
742. 2:04:36path to the file but the little plus button lets you upload things what's the difference between chat GPT and Bing okay
743. 2:04:44so down here is this thing called the large language model this is sort of the core engine gpt4 chat GPT is an application that sits on
744. 2:04:54top of it and it's the thing that makes it talk to you politely and it's it's the thing that says I'm sorry Dave I'm
745. 2:05:02afraid I can't do that right it's got all the rules right it's the application and so that's chat GPT and then Bing
746. 2:05:09chat is also sitting on top of that same gpt4 large language model and it's got very very similar um functions as chat GPT but it's got
747. 2:05:21different ones and it's it's got different rules so it's connected to the internet it it incorporates um Dali image generation
748. 2:05:32um it's it's a bit more restrictive and a bit less good than chat GPT so so they're basically just two flavors of the same thing
749. 2:05:42sitting on the same enabling technology can it generate 3D model from a 2D image not yet that's coming I would say that
750. 2:05:52like there's there's a ton of white papers right now on text to 3D object and then there's a ton of white papers on extracting 3D
751. 2:06:03um information from 2D images and doing that so I don't know give it six months and that'll be here and it's there's
752. 2:06:11probably tools you can play with now if you go to futurepedia.io there's probably stuff I don't know about that's doing
753. 2:06:19text to 3D object that doesn't suck but it's all it's all super new right now so I'd say give give it six months what
754. 2:06:26kind of computer do I need mine is old and I need a new one I want to play with AI um it depends what you want to play with
755. 2:06:32AI if you want to like host your own models and and actually run large language models locally then if you're a Mac person get it get
756. 2:06:42the highest spec M2 you can get if you're a PC person you're going to need to get a a cranked out multi-gpu system like what's your budget
757. 2:06:54like George hotz is is designing a fifteen thousand dollar machine learning you know Beast so it depends on what you want to do
758. 2:07:06with it if you just want to play with chat GPT any [ __ ] computer your phone works right because all the compute is
759. 2:07:15happening on Microsoft servers somewhere and gpus so it depends what you want to do if you want to do local highfalutin
760. 2:07:25processing then you need a fancy computer if you just want to use these tools you don't need [ __ ] um other than plugins what's the
761. 2:07:36distinct difference between 3.5 and 4. so the quality difference when it comes to language and program just just the
762. 2:07:45quality of the response you get back from gpt4 versus GPT 3.5 is it's subtle but it's like staggeringly different
763. 2:07:56like if I have GPT 3.5 write me something it gets me to like 65 or 70 percent gpt4 gets me to like 80 to 90
764. 2:08:04percent and like you know like there there like I don't have to edit nearly as much if I'm using gpt4 and it's
765. 2:08:12apparently better at code as well yeah and then the other difference is code interpreter so if you pay for
766. 2:08:23so so a couple of things if you want gpt4 for free use Bing chat Bing chat is gpt4 if you want access to all of open ai's
767. 2:08:34newest Technologies as soon as you can get them pay for their API and pay for a chat GPT here GPT plus and you'll get early [ __ ]
768. 2:08:44here and then if you're paying for apis you'll get access to the API stuff over there so it depends what you want to do
769. 2:08:50if you don't want to spend any money just use Bing chat that's gpt4 GPT is faster for smarter that's right
770. 2:08:59gpt's like don't get me wrong GPT 3.5 is still really good and it is very fast so if you want stuff fast do that Google is
771. 2:09:08doing a Xerox Park in the 70s where they ignored the future and they developed CRX part they're doing a Kodak right
772. 2:09:15Kodak [ __ ] invented the digital camera the dude walks into the boardroom and says here's the digital camera it's the future of Kodak and
773. 2:09:27they're like boy we're in the silver halide business that stupid Contraption of yours ain't gonna sell more film now
774. 2:09:33is it now get out of here you cow poke yeah they are I think it's [ __ ] up but it's you know I've argued with it before I knew it it
775. 2:09:48had a character limit it was creating something very long and that's interesting how do you rank kyber for video
776. 2:09:54generation what's better I've only really played with um with Runway with the video generation I've heard kyber
777. 2:10:03um from from a lot of people I trust kyber is really really good I just haven't played with it um there's there's uh descript
778. 2:10:13um there's the script Runway kyber are the three biggies that I know of Kyle I can tell you how to access code
779. 2:10:22interpreter right now I have it uh thoughts on open llms um cool I can't keep up with them a new one comes out every three days so I'm just
780. 2:10:36waiting until the the day that one comes out where they're like here's here's the seven billion parameter llm that's as good or better
781. 2:10:45than gpt4 that you can install on your phone or your your Mac and you don't need any fancy computer when that comes out then I'll
782. 2:10:53then I'll pounce on open llms I just don't have time to keep up with it right now and I'm not geeky enough to be in any of the projects I I
783. 2:11:05pay attention to to the announcements and all the announcements are like we're almost as good as 3.5 we're just better than 3.5
784. 2:11:15I'm waiting for it's tiny and it's as good as for go get them that I'm excited about the alignment layer on chat gbt straight up refuses to
785. 2:11:25discuss my innocent topics yeah I know so you can either jailbreak it go look up uh Dan do anything now or go start
786. 2:11:35playing with the open source large language models that don't have any any of the constraints on them if you want to if you want to sort of
787. 2:11:43push the boundaries and it's getting frustrating it's yeah those the the the safety guideline stuff can be really
788. 2:11:49frustrating depending on your topic I thought he said he was still waiting for it nope I got it I was I said that I don't know two lives ago
789. 2:12:01I'm so addicted to them I'm not sure to what but let's see that that mid-journey is super awesome that and chat gpt4 is worth
790. 2:12:11paying for I agree yo Ronda what's happening let's chat this week okay I'm around I don't have a busy week this week so let's talk Rhonda um
791. 2:12:21we're 46 with 130 saved for retirement wait is this good or not real honest opinion please I don't know what you mean Joel is what
792. 2:12:35good or not real is is generative AI good and it's very real it's very good it's profoundly good I don't know why you're asking are you
793. 2:12:45asking about investment advice don't ask me I have a degree in acting I'm [ __ ] horrible with money but but yes AI is real it's gonna
794. 2:12:56[ __ ] change everything it's gonna profoundly transform the world um but where would I place a bet I am not the right guy to ask can code
795. 2:13:08interpreter make an Excel spreadsheet from a mySQL database my guess is yes I would I would yes I think I don't know I would
796. 2:13:18try it I think it can do a lot I think it can do way more than we no because what I'm seeing right now is if you go to Twitter and
797. 2:13:32um look up code enter like code interpreter was trending today which is good right people are using it but if you go look
798. 2:13:39at what people are posting most of what they're posting is them trying [ __ ] other people already talked about two months ago
799. 2:13:47it's almost all that so I'm kind of waiting for a week or two for some actual real smart people to do
800. 2:13:53interesting [ __ ] with it Ethan Malik put out um a post today about a prompt that he wrote to to create some really interesting
801. 2:14:03um charts and graphs based on Edward tufty which I think is really smart um so go look up Ethan Malik his his
802. 2:14:10post today on on the prompt that he created it's really really cool and he's had code interpreter for like three months
803. 2:14:16if you if you joined Leonardo's Facebook group they'll Grant you instant access oh but Lauren lay thank you so much
804. 2:14:24that's great news to hear okay so that's huge that I mean that's a huge tip so join Leonardo's Facebook if if
805. 2:14:35you want to do image generation leonardo.ai is amazing and if you join their Facebook group and they give you
806. 2:14:42instant access go do that right now it's it's a it's an amazing tool it really is and they give you a fair amount of
807. 2:14:50tokens to play with for free every day so if they're not stingy all right I gotta go it's 11 o'clock
808. 2:14:57it's time I don't know where my children are well they're upstairs but you know other than that Tick Tock just jumped me to the bottom
809. 2:15:06of comments so I'm just gonna go I'm gonna grab some of these bottom comments here and then I am gonna get out of here
810. 2:15:13how do I get a job in AI go there start getting AI literate here's my recipe go start playing
811. 2:15:28get good at first parlor tricks like just understand it can write songs it can write poems it can write code it can
812. 2:15:35make tables it can it did it like start learning it then take what you do and learn how that can help you do what you do better
813. 2:15:46so if you're a project manager figure out how you can use chat GPT to do better project management if you're a
814. 2:15:52writer if you're a whatever whatever you do whatever you're interested in dig deep in your areas of knowledge as
815. 2:16:01you're doing all of this share it join groups put it on LinkedIn put it on Facebook tell people at dinner parties tell
816. 2:16:14people at work ah I I'm playing with chat gbt and I had it write me a country song start with that and let them look at you weird like I
817. 2:16:24heard the robots are gonna kill us why are you playing with that stupid AI thing I think it's the devil isn't it
818. 2:16:30the devil I heard it's the devil isn't it aren't the robots gonna kill us I heard it's a waste of time you know what I heard I heard it lies
819. 2:16:36I heard it's stealing all this information you're not going to use that right it's stealing everyone's information ignore all that
820. 2:16:43and just keep talking about it because at some point one of those craziest people in your life is going to go you
821. 2:16:49know I need to figure out to say I should who should I talk to oh unbridled son has been talking about that for the last
822. 2:16:56three weeks he won't shut up about it I got to talk to him can you introduce me that's how you get a job
823. 2:17:02put it on your LinkedIn join groups we had the AI salon so this URL go here the salon.ai that's a community I started for people
824. 2:17:16curious about AI we've got a Discord server we meet every other week it's a group of people that are just kind of on this adventure together
825. 2:17:24we have a lot of fun and people in this group have put the fact that they're a part of this group on their LinkedIn profiles and gotten jobs
826. 2:17:33because of it I'm not guaranteeing that at all like but just start talking about it because people are getting increasingly curious
827. 2:17:40about AI so get literate about it and and start talking about it and start helping people the other thing to do start helping people
828. 2:17:48because one of the things that um that I've noticed is people like well I don't know where to start I just look
829. 2:17:55at that that you know the open AI box and I don't know what to do rather than thinking what can I do with AI flip it upside down
830. 2:18:04and say what problem needs solving and then ask okay how can I use AI to solve that problem so so one of the things that you can do
831. 2:18:14is you can go to people and say hey I'm learning this AI stuff and I know you've been struggling you know finding a
832. 2:18:21marketing agency I don't know anything about marketing but would you mind if I used chat GPT to come up with a marketing plan so I could
833. 2:18:31learn it and maybe you could benefit from that too sure you can do that for free for me like give some [ __ ] away for free and go learn it
834. 2:18:40thank you for the balloons um just start playing get curious get curious get curious play play play play play play make it an adventure
835. 2:18:52like any good Adventure it should be fun it should be scary there should be a lot at stake you should do it with others
836. 2:19:02ideally if you're on a good Adventure you're doing it with others there's there's like the meat head that can
837. 2:19:08swing an ax there's like the little Brainiac mousy one that can't talk but can solve problems there's the charismatic one there's the
838. 2:19:17wizard right you go you get your group of four or five people that all have different skills go on this [ __ ] adventure together with them
839. 2:19:24because when you start talking about it you're gonna find out oh there's someone else in your group that's been secretly doing this too
840. 2:19:31oh you're doing this too you want to go have coffee and talk about it yeah do you know what you're doing no that's how you get a job
841. 2:19:43you can't monetize AI chat tools unless you do a subscription you can monetize Consulting that's just not true
842. 2:19:56you could go use bing.com for free and go tell people I'll I'll use I'll use AI to give you what your agency
843. 2:20:05is charging you ten thousand dollars for I'll do it for five and I'll deliver it twice as fast want to take a shot
844. 2:20:14you don't need a subscription to do that awesome work thank you you're welcome headlines Sarah Silverman suing
845. 2:20:22open AI meta for copyright infringement like I said the next 20 years is going to be nothing but a string of one
846. 2:20:30lawsuit after another at some point there's going to be some precedent set where it's like well Sarah Silverman's
847. 2:20:37you know DVDs are also in libraries and libraries are there for you know fair use and knowledge consumption hey hey um
848. 2:20:49just the lawsuits are going to keep coming doesn't surprise me it whatever it's you know it's the this is
849. 2:21:00there's a there's going to be a lot of what I what what I'm calling business model Panic coming which is my business model relied on
850. 2:21:09this particular interpretation of copyright law which these tools fundamentally [ __ ] blow up so there's going to be a lot of
851. 2:21:18companies a lot of big organizations going we can't lose our business model we can't lose our sue them sue them sue
852. 2:21:25them sue them it's what they've always done it's what they'll always do I think this this is so much different than um
853. 2:21:34like music sampling in the 80s and 90s with hip-hop remember all the remember all the copyright suits you're stealing
854. 2:21:40our music and this and that and and it took him a decade to sort of figure out fair use and and how all that stuff
855. 2:21:45worked this is just a fundamentally different thing sampling music for hip-hop you're actually taking part of someone else's
856. 2:21:54song copying it and pasting it into your song this stuff is generating new original content based on some
857. 2:22:02amalgamation of training data that may include something Sarah Silverman said but it may not and how you determine what percentage of
858. 2:22:12all of the [ __ ] that was on the Internet is that Sarah Silverman thing that ends up in that one line of that one poem
859. 2:22:19is essentially impossible so it's a fundamentally different thing it doesn't mean everyone's not gonna sue the [ __ ] out of
860. 2:22:27everyone they are so this is just the beginning like the lawsuits the lawyers are like I said I've said to this year before the
861. 2:22:36lawyers right now are pre-ordering their new yachts they're gonna make a [ __ ] fortune on both sides
862. 2:22:44there's gonna be lawyers going ah Sarah we can help you sue open AI they just got 10 billion dollars from Microsoft
863. 2:22:50let's go get some of that cold cash and then on the other side they're gonna be like yeah open Ai and Microsoft we
864. 2:22:56can help defend against all these frivolous lawsuits let's go the lawyers are thrilled imagine if every third sentence in AI
865. 2:23:07response was a sponsored reply that's someone's gonna do that Google's probably gonna so so Microsoft doesn't
866. 2:23:14have to do that because they can monetize this [ __ ] with all of their applications Google can't but then if Google's chat
867. 2:23:28tool is too saturated with sponsored content and you compare it to open AIS or open source models that have no sponsorship in them
868. 2:23:41is there a point at which it's just not useful and it drifts to the background probably any free AI tools that have Enterprise
869. 2:23:52or commercial rights that you know well not free Adobe if you're doing images the Adobe stuff is commercially safe um
870. 2:24:04all of this generative large language model stuff depending on how conservative the organization is might steer clear of this for a while
871. 2:24:13but but you should go read the article from Ethan mullet called um is it discovering detecting detecting your secrets the secret cyborgs
872. 2:24:25so what what Malik talks about in that piece detecting the secret cyborgs is that individuals in in big companies are
873. 2:24:35using large language they're using chat GPT as personal productivity enhancers but they're not telling the company
874. 2:24:40about it and the company's not using these things because they're not ready for prime time and so what's happening is you've got a
875. 2:24:47class of AI literate employees that are being more efficient and more innovative in secret and corporations that aren't benefiting from that
876. 2:24:57and they're afraid to say anything because maybe there's a policy that we don't use AI or maybe there's a policy
877. 2:25:03that uh that you know you can only use it in a certain way or maybe one of these employees is now doing in four
878. 2:25:10hours what used to take them 30 hours and they don't want to tell anyone about it so so companies are going to need to
879. 2:25:17find a way to close that Gap it's like we're in a fascinating time right now um so I don't know like like these these
880. 2:25:26tools are all new right now um the the legal questions around anything generated with these tools is is all over the place um
881. 2:25:38when Chachi PT is built into Microsoft Word and I'm writing an email and there's a button there that says help me write
882. 2:25:48at what percentage of that email does it have to be AI generated for the whole email to be considered AI generated is
883. 2:25:57it just a word is it just a sentence is it the whole body I think the reality of these tools as they get more and more integrated into
884. 2:26:05our business tools is that everything produced everything produced by people will be some hybrid of AI and people
885. 2:26:14and so I don't think it's going to be as clear as this is AI and this isn't so I I like I have no answers for you
886. 2:26:23unfortunately other than start learning it if you own a company start learning it if you don't own a company start learning it
887. 2:26:32if you're afraid of it start learning it just start getting your head around what it is and then as the legal stuff gets sorted
888. 2:26:43over the next 20 years do the appropriate thing what type of accessibility features are found in chat GPT for blind low vision
889. 2:26:58people so so there's a lot of things you can do you can there are what are they called are they called automations on the iPhone
890. 2:27:12I forget what they're called on the iPhone anyway they're automations but there are things you can do on the
891. 2:27:17iPhone where you can basically invoke Siri so you can talk to Siri and say like Siri GPT or something like that and
892. 2:27:24you'll actually be talking to GPT I also think the new pie tool from uh inflection is is voice is voice activated um
893. 2:27:39so I think I think whatever accessibility you use with voice to text and text to voice um is is all still in play with chat gbt
894. 2:27:49it's it's really just it's just like a Google search so to the extent that you can do Google searches you can do things
895. 2:27:56with chat gbt um but I don't know that they're explicit accessibility tools uh built into it but but there's there's
896. 2:28:05plenty certainly within the Apple ecosystem to deal with all that um thanks for the info midnight here
897. 2:28:11hate to go but have to I know I gotta go too it's 11 10 here all right uh plus someone needs to Chicago I gotta say all right listen I
898. 2:28:17gotta go I'm gonna do one more one more why is host not reading the comments because I I am can you join us for 10 seconds sorry
899. 2:28:28folks unless you have a server Farm dozens of full-time computer scientists okay that's funny I have a quick question ask it uh
900. 2:28:38um you get nothing from AI tools okay so I think this is not live because he's not responding to ghost questions I am live
901. 2:28:47faith hope and love see if I weren't live could I say your name awkwardly and as if I can't read it because you didn't use capital letters
902. 2:28:57and it just seems like alphabet soup if you capitalize the first letter of each word it'll it'll improve usability of your name
903. 2:29:05Faith open love what's Blue Willow Blue Willow looks like mid-journey because you make images in a Discord server but it's built on top of
904. 2:29:19stable diffusion which is not quite as good as mid-journey but it's free so far out all right did you ever hear NPR's
905. 2:29:31broadcast regarding ai's ability to take the reporters jobs yet no but that doesn't surprise me but it could also
906. 2:29:39I I'm I'm excited about the return of local journalism where it's not local journalists but it's AI journalists because I think you could
907. 2:29:49I think we could help our country by by re-inventing local journalism with AI so drink water thank you
908. 2:30:01I know my voice has gone all crispy all right listen peace out everyone I gotta go I'm I'm moist I can't talk my ass is numb I've been
909. 2:30:10doing this for three hours I just can't people I can't do it anymore I've been on the back of this horse for days now it's days
910. 2:30:20thank you very much I appreciate the Roses I appreciate the hearts I appreciate the questions I appreciate you all being willing to be
911. 2:30:30open to giving this [ __ ] a try I think it will help you and uh I'll see you tomorrow peace