aiLL 7/7/23 - 1 - The Impact of Code Interpreter on AI's Potential

Transcript

1. 0:03foreign
2. 0:27welcome welcome Happy Friday night just kind of went out to uh First Fridays here in Denver saw a bunch of art that was nice hello ice cream
3. 0:39welcome to the AI learning lab the nightly live it's been nightly for like two weeks now nuts love these lives Kyle thank you very
4. 0:49much Jamie or James I appreciate that the the hair is special tonight that that that looks like I tried that that
5. 0:59is not that was not on purpose [Laughter] oh man uh welcome to the AI learning lab AI learning lab we'll get it going here
6. 1:13if you have questions about generative AI or want to I don't know look at something or goodness goodness goodness
7. 1:22take a look at code interpreter from chatgpt which I got access to today holy crap uh it's making me feel very stupid
8. 1:31because I don't know enough about data analysis to use it right now but I'm figuring it out and it looks remarkable
9. 1:37and then there's prompts.chat that thing is if you haven't played with chat GPT and you don't really know what prompting
10. 1:44is and you hear people talking about it that document is a good primer for how to do prompting and then it gives you a
11. 1:52lot of examples of personas that you can tell chat GPT to act like some sort of expert act like an expert on marketing
12. 2:01or on sales or on writing or whatever it might be and it dramatically improves the content so anyway welcome here if
13. 2:10you have questions pop them in the comments below I do my best to answer everything happy Friday night everybody
14. 2:15let's get this show rolling rolling rolling Friday Friday Friday all right can AI elaborate upon huge heat wave hitting the Southwest
15. 2:29U.S in a few days possibly that might be something for being let's go let's go I don't know what you mean
16. 2:42elaborate but let's go see what Bing has to say about it my shirt is apparently my my massive muscles are too big for my
17. 2:51little short sleeves here I can't move my arms up um all right all right see uh let's do more precise and so but it
18. 3:02looks like you know like a Google kind of search engine then if you click the chat button that brings you here and this is Bing
19. 3:10chat it's it's basically chat gpt4 and let's see um can you tell me what is uh what
20. 3:26is happening with the heat wave in the U.S Southwest
21. 3:41Heatwave us Southwest yes there is a heat wave according to NASA some of the high pressure settled over
22. 3:50Southwest in early July causing ground surface temperatures to rise the heat wave is expected to fix affect much of
23. 3:57Arizona New Mexico build into California the heat wave could rank among its worst in history in terms of intensity or
24. 4:05longevity holy crap well here in Denver it's it's been raining like it threatened to rain really bad tonight
25. 4:12it's been raining and hailing and storming we've gotten more rain for the first six months of 2023 than Seattle Denver
26. 4:20which is a high desert got more rain in the first six months of the year than Seattle that's never happened in history
27. 4:26before so there's some [ __ ] going down um I do know that there's an El Nino that we exited in La Nina and we entered
28. 4:35an El Nino so I know that's going on I don't know if that affects it but also July 3rd was the hottest recorded day in
29. 4:43the history of the Earth so we're getting better right all right it's okay the robots are gonna kill us anyway can AI generate images
30. 4:57such as logos and how it can it's not super great it doesn't do text very well at all but if you want to do like logo
31. 5:05ideas I'll do I'll do some here at Bing so let's just say um let's see graphic graphic logo logo um flat illustration
32. 5:20let's see um of a oh I don't know um of a flower shop um with
33. 5:36a rose as the main element and Geo match Rick shapes behind it
34. 5:50let's see what this does so so Bing does images now now it's it's sitting on top of Dolly which is the weakest of all the image models so my
35. 6:03guess is that this is gonna suck but I don't know we'll see um it's not as horrible as this way better than I thought it would be
36. 6:15it's not really a logo um those are kind of more like Design Concepts for for you know maybe some printed materials things like that
37. 6:26that's not horrible let's let's flip over I'll flip over to uh to Discord we'll do something in uh mid Journey
38. 6:34that was a piece I did last night on the live uh first for mid-journey someone's picking on me from for making
39. 6:45art that looks like theirs that's pretty funny um let's see so if if you haven't used mid journey by the way how you use it this is we're in
40. 6:54Discord so in Discord you've got you know all these servers so these are all different communities and then I'm in
41. 7:00one uh for the AI Salon which is a community that I started and I've got a mid-journey channel in here so let's
42. 7:06just hop in here and we'll go so how you do this you go slash imagine and then it gives you a prompt oh I'll
43. 7:12do another thing before we do this we'll we'll go make the prompt even better so I'm gonna go over to so I just showed
44. 7:20you Bing that's an image that was generated in bing now we're going to go to go to Poe po.com I assume that means as in Edgar Allan
45. 7:30Poe so when you're at Poe it's a normal like you know chat bot search engine and you've got six different models you can
46. 7:36play with but you also have these Bots you can make a bot and you can sorry trying to get the Focus to focus there we go um
47. 7:47you can look at other people's spots but there's there's a mid-journey bot that's actually pretty good so if you go to
48. 7:54poe.com mid-journey it takes you to the spot that specializes in fixing your shitty prompts so we'll say logo of a flower shop flat
49. 8:10illustration let's see graphic I don't know on white let's say and then it's going to rewrite that so color logo of a flower shop flat
50. 8:22illustration graphic on white featuring a blooming flower with a with delicate petals simple and elegant minimalist
51. 8:29design Adobe Illustrator software RGB color model vector format Mary Blair Charlie Harper Saul bass so let's copy
52. 8:39that that's our prompt we'll hop over to um to Discord and mid journey and I'm just going to paste it in and hit return
53. 8:48so so we're starting to get to get where you know the AIS are starting to prompt the other AIS and you can you can use
54. 8:54them in conjunction with one another and you can figure out well this tool might be better for this project and this
55. 9:00might be better for that project and in general with images mid-journey is better than anything else it's just the
56. 9:06way it is I don't make the rules I just enforce them [Laughter] uh actually some of these are looking kind of logo like what you'll notice
57. 9:18here is um the the way these diffusion models work they start out as kind of these blurry you know highly noisy images and then
58. 9:28they seem to get a little more clear over time that's exactly how they work they kind of are averaging lots and lots
59. 9:34of random noise generated images that were based on you know real images but just random noise they kind of average
60. 9:41them together and it starts as this random noise amalgamation and then it denoises them over like 30 or 40 or 70 steps
61. 9:51and boom there's these logos so this one isn't exactly a rose on the lower right but that's pretty good logo actually
62. 10:00it's pretty interesting it's at least good um it's at least good uh whatever you call it inspiration don't you think
63. 10:09I mean that's pretty dang nice it tried to do text but you could obviously wipe that text out of there
64. 10:18and you know that one's all any of these for flower shops I mean actually this this upper right one um it's not really a logo for a store
65. 10:27but it certainly looks like it could be on a sign for a store and they put it in a store window which is kind of cool so
66. 10:36yeah it can do that that's that's a couple of ways you can you can do it um and uh let's see
67. 10:51what else do we have here I like net nip like a net beans for dummy dummies leave that chat bot alone they just got
68. 10:59sued for copyright infringement I assume you mean mid-journey everyone's getting sued right now everyone's going to be
69. 11:05sued for the next long long time um you know yeah I I understand you your your resistance to progress uh but
70. 11:15you know it's this stuff isn't going away so um yeah collar shirt tonight formal night well it's not really a formal
71. 11:24night but uh but I I did I am wearing a shirt dark enough to read text in front of so so I don't have to change it uh
72. 11:33uh this is a great way to learn about the great AI thank you uh who actually knows the good news for
73. 11:42the Colorado River with all that rain oh I well Utah had like what did they have I know some 40 or 50 feet of snow I mean
74. 11:52they had a ridiculous amount of snow so that all contributed to that water basin and then yeah it is it's been it's
75. 11:59honest to God like living in well no it's not like living in Florida at all but but like afternoon thunder thunder
76. 12:06thundershowers like just it's been weird it's been not like Colorado so yeah I think that is good for The Californians
77. 12:14who are like where's our [ __ ] water our Lakes are drying up please help so their water should be okay government
78. 12:25weather experiments yeah Chemtrails and all that stuff huh are we hacking into NASA not yet although uh code
79. 12:32interpreters doing some pretty insane stuff I'll show you that in a second thank you for sharing you're welcome
80. 12:37I'll be getting three arms and three legs on some of my mid-jury journey true to life rendering how do I fix Ed
81. 12:45um Google or like go into Reddit into the mid-journey things you can put in some negative prompts in mid-journey to
82. 12:51have it not do certain things like you know don't give me weird hands don't give me too many arms things like that
83. 12:56so you can give it some negative prompts um you can also do things like remixing where you upload a photo and sort of
84. 13:04remix it into something else it'll keep the composition of the first thing but change everything out that's a way that you could handle that
85. 13:13um just keep experimenting just you know just I there the weird thing about these models just the nature of how they
86. 13:21produce the images they produce they don't really know what they're producing it it is literally sort of
87. 13:27this Amalgamated set of pixels that gets denoised into a real image so I I suppose if you know certain parts got
88. 13:36overlaid over certain parts or however the hell it's doing what it's doing um it just does that so you just got to
89. 13:42keep working on your prompting and uh yeah you'll get there you will get there can AI validate the theory to increase
90. 13:53monsoonal moisture for the Colorado Basin by evaporation of the ocean uh I don't know I don't think so you
91. 14:01might play around with uh with code interpreter um go get some go get some weather data it's possible sounds like we got a
92. 14:09little conspiracy theory action going on but you know whatever uh I'm in Los Angeles for the week and it's definitely
93. 14:16dry I'm supposed to be on vacation but I can't stay away from the AI news I know the AI news is just crazy so check check this out so
94. 14:26let me flip up got um yeah it's not back for me so um browse with Bing is still missing but look at this code interpreter is here
95. 14:39and it's Live code interpreter is live so this is something I've been waiting for for a long long time so the way this
96. 14:47works you go to the new code interpreter feature and notice you get a little you can't really see it you get a little plus button here
97. 14:56and you can upload a data file and it can you can even upload so so I went to go get a bunch of data I got some from
98. 15:06LinkedIn I got some from some different data sources and one of the data sets I got was all these individual CSV files right they
99. 15:14like you can you don't have multiple worksheets you can upload a zip file that has multiple csvs in it and it will
100. 15:22read them all it reads inside the the zip files it's pretty amazing so let me see there was one here this one so this was a bunch of Chess
101. 15:33data so I'm uploading a spreadsheet with a bunch of Chess data and then I'm going to say um can you let's see visualize this data
102. 15:50in in a visually interesting way and describe what it's all about
103. 16:08and sure let's start by unzipping the file so so it's just like a plug-in it's just the way plugins work it just pops open
104. 16:16that thing it says finished working the zip file contained a single CSV file let's load this file and look at it and then it says it's got
105. 16:27ID rated created at last move at turns Victory status so it's all this data about Chess to provide visual representation of the
106. 16:37data we can create plots for various aspects let's proceed with this visualization okay and so now it's firing off now let me go
107. 16:47show you what's going on behind the scenes here so every time it fires one of these off it's actually writing python
108. 16:53so so it wrote The Python code and installed the libraries to unzip the file and then the next one it it imported
109. 17:03Panda pandas and then went and looked at the data and so so it analyzed the data and there it made [ __ ] charts
110. 17:14of the data it made [ __ ] charts of the data distribution of game results the majority of games are won by either
111. 17:23white or black with white having a slight Edge there are very few games that end in the draw distribution of
112. 17:28number of turns the number of turns in the game followed roughly a normal normal distribution with most games
113. 17:33having around 50 turns there are also some games with a very high number of turns indicating particularly long games
114. 17:40distribution of player ratings 10 commonly used openings the the most commonly used opening data opening in this data set is Van uh
115. 17:52cruises opening followed by the Sicilian Defense and the Scandinavian defense misis cotric variation I don't I don't
116. 18:01play chess I don't know all these the rest of the openings in the top 10 are used less frequently I mean holy [ __ ]
117. 18:10um let me show you another thing it can do let me do I'll do a new chat um I'll get a code interpreter I'm going to upload an image
118. 18:18and then I am going to go grab is this it yeah okay so I've got this image so this was an image that was
119. 18:27created in the AI Salon it was really cool she was playing around with extending the image and doing panning
120. 18:33okay so I'm going to upload this image it's not doing image recognition like um like gpt4 is supposed to where you
121. 18:41can see what's in the image but I can do this I can say um uh let's see find me the 15
122. 18:57um like I I most common colors in this image and create a graph of the percentage they are used in the
123. 19:16image make the bars of the graph match the color it represents and label with
124. 19:31um hex codes and even name the colors like they are crayons okay
125. 19:46so all right to find the most 15 most common colors in the image I'll follow these steps I'm going to load the image
126. 19:53calculate a histogram of the image which will give us the frequency of each color identify the 15 most common colors
127. 19:59convert those colors to hex representations so it's it's figuring out its own logic it's then writing python code
128. 20:09to execute that logic and if it gets it wrong it'll rewrite the code and re-execute it I obtained the 15 most common colors in the image
129. 20:19however I noticed that the color values are in a Range zero to two let's see finished working I apologize
130. 20:29for the oversight it it seems that the environment does not have the web colors Library installed however we can still
131. 20:36convert them to hexadecimal so it made some mistakes next I'll create a bar chart to visualize the percentage of the colors used
132. 20:46these colors are not right it didn't get the colors right here and it didn't name them there's a horizontal bar chart um let's see
133. 21:00um can you get uh let's see can you average um let's see can you get the 15 colors that are
134. 21:17an average rather than the most pixels I don't even know if that's the correct question but anyway so this is what this thing does you can
135. 21:31you can upload spreadsheets you can upload PDFs it will write the python code to pull the text out of a PDF
136. 21:39you could use this the file that you upload here can be a hundred megabytes so so right now chatgpt has this 4 000
137. 21:50this 4 000 token context window like 3 000 words where it kind of doesn't really have its memory um you could you could put in a hundred
138. 22:01megabyte prompt basically and and have it start executing against that prompt I mean I I just got this today so like
139. 22:11I've only played with it for about an hour and I like it this is making me feel incredibly stupid because I don't even know what to ask it
140. 22:18but holy crap holy crap is it amazing so anyway what's it doing it let's see what's it done here has it found anything
141. 22:29average color hexes so it wrote code to get the average colors kit K means cluster centers as type integer
142. 22:40average color hexes RGB to hex color apologies the algorithm's taking too long I'll use a method called mini batch K means
143. 22:55so it didn't it didn't like that so so it's like it tried some code it took too long it said let me try another approach it took too long
144. 23:03now it's trying another approach now we've got some real colors so it looks like it might have worked next I'll create a bar chart to
145. 23:15visualize it I I mean this is nuts there look at that now it didn't get the percentages right but it got the colors
146. 23:26so now so the so the horizontal bar chart is the 15 colors that are in the image and if we go back to the image let's see um
147. 23:37if we go back to the image you know that's the image those are the colors in it right so see those colors and then we go back
148. 23:45to what it actually created those are the colors so so it pulled the palette colors out of that image so now I can say um
149. 23:56please name the colors creative names based on their hue
150. 24:10um saturation and bright Ness in a way that makes them memorable
151. 24:32artistic interpretation but it's going to go ahead and do it anyway [Laughter] I converted the colors to HSB
152. 24:44all right now it's explaining what HSB is I didn't want it to do that I just wanted it to name the colors but whatever
153. 24:50we'll see what it does [Music] welcome everybody my name is Kyle
154. 25:05Shannon what's this Channel about watching an old guy try to use a computer it's Elder talk welcome to old guys try to compute uh
155. 25:18okay what did this name these these name this red dark cyan okay let's see um make the names more creative and put them on a
156. 25:35chart with with the name next or the name below it's colored Square
157. 25:50we'll see if it can do that see if we can get it to write some python to lay them out I see You'd like the names that
158. 25:57are a bit more imaginative Emerald Knight for dark muted green Sunrise gold for light orange midnight shadow for dark cyan so that's better
159. 26:05now let's create a chart with the color square and its corresponding name so it is off writing some python code oh my God it did it
160. 26:16it's just this is insanity this is just Insanity look at that there's all the colors now it like the formatting is a
161. 26:25little messed up and I can't let's see can I copy this image copy image let me go over to here we'll do new the like
162. 26:33the um the formatting's a little screwed up like it it overset the names but like hang on so like Emerald Knight is green oh wait
163. 26:46where are we here Emerald Knight is green Sunrise gold midnight shadow Ruby blast ocean wave I mean the ah I can't I can't
164. 27:06you know you know I say on the channel a lot you know dick around with this Chachi BT thing until you have your
165. 27:11Kevin McAllister moment you know the guy from Home Alone this is that for me I mean I don't know anyone paying attention to this good
166. 27:20Lord good Lord what's happening uh this thing is insane it's insane it's insanity so I don't know much more like like
167. 27:31that's about as much as I know right now um I haven't messed with it but things like you know someone was asking about the
168. 27:38weather data you could go to the NOAA site and get weather data for the past however many five years 10 years 20
169. 27:47years for a certain region of the country pop it in here and have this thing start analyzing it this is full on data analysis in a box
170. 27:56it's it's doing full-on data analysis with coding um oh thank you for thank you for subscribing that's amazing wow I know
171. 28:07I love you Kyle thank you did you look at the storyline alternative website I did I did and it's it's it's not bad
172. 28:14it's it's in the neighborhood it's in the neighborhood uh but we are we are moving on we're rolling some AI into storyline it's going to get
173. 28:25slick pretty quick I hope I'm excited all right I'm gonna try and have try to do that with stock analysis yeah that's the thing oh so
174. 28:33here's the deal with with code interpreter so if you pay for chat affiliated with any of these um companies or sites in any way I don't
175. 28:43make any money off this so these are just here because these are the sites that I think are the the best if you want to get started
176. 28:51in figuring out what the hell this AI stuff's all about these are the best places to start if you subscribe to
177. 28:58um chat gbt they're they're plus um thanks for the panda if for their their uh Plus subscription um you'll get access to code interpreter
178. 29:07that thing I was just playing with you'll get access to it they're rolling it out wide this week um at some point they're going to give
179. 29:14us back access to the it connected to the internet I don't know why it's taken so long it seemed like it was a
180. 29:20relatively straightforward thing to fix and being bing is still connected to the Internet so I don't quite know what's
181. 29:26going on there I assume it's got to do with lawyers and some copyright and some trademark and things like that um but yeah I think
182. 29:35that would be really interesting to try with uh with financial data will chat GPT plugins be available with free chat
183. 29:43GPT they probably will over time the reason you know if you want them now you have to pay for the Plus subscription
184. 29:53um I would say that plugins are maybe not worth paying for if they were on their own but code interpreter which
185. 30:01they're making available wide this week to everyone who pays um that seems well very very very well worth it um it it
186. 30:15I can't I couldn't get my head around what chat GPT on its own made possible the fact that it's now got this this code executing
187. 30:28thing that just I don't I if you've been to my channel you know that I'm not speechless very often I'm
188. 30:36I'm a little I'm a little shook I mean I think it's a good thing I think it's a good thing but holy crap
189. 30:43all right my light was a little bright it was making my eyes hurt and I don't like it when my eyes hurt it's not fair
190. 30:50how do we get it it's not my gbt dashboard so here we are at Ye Old chat GPT so we've got our plus so if you have plus there
191. 31:00if you've paid your 20 a month you'll have GPT 3.5 and gpt4 what I have access to is default code interpreter and plugins now if you just
192. 31:11have default what you need to do is you go to your settings so I'm going to go to Kyle Shannon hey Kyle Shannon what
193. 31:19are you doing boy all right and then I'm going to go to Beta features and then you have to turn them on
194. 31:25so you have to turn on plugins and turn it turn on code interpreter turn those on and then you should have this now if
195. 31:34when you go to settings um and you go to Beta features you just have plugins but you don't have code interpreter it just means they haven't
196. 31:42gotten to you yet I assume they're rolling this out to like you know waves of people that were on the wait list and things like that so
197. 31:49um so that's that should get you there and you know like I've been waiting for code interpreter for like two months
198. 31:57like so the fact that it just showed up I'm like it's like it's like there goes the weekend AI programs like chat GPT have learned
199. 32:08how to send Bitcoin payments from one machine paying another interesting great prompt though great I don't know
200. 32:15which one I did that's hysterical how it apologizes along the way I know isn't that funny it's like ah apologies but uh
201. 32:22I didn't bother to load that Library that's required let me go ahead and load it and try oh apologies it took a little
202. 32:28too too long to do that like to get those colors it did three different approaches to to getting average colors
203. 32:36and then it just did it amazing I know it's like I mean if you think about that if you think about I mean
204. 32:45why you would want to get the 15 you know you know uh most uh prevalent colors in that image whatever like like
205. 32:54but but that's that's kind of irrelevant the fact that you can just say you know here's this file I want you to perform
206. 33:02some tasks on it and it goes and figures out exactly what it needs to do like like like when you upload a zip file
207. 33:10and you say what's what is this it says well it's a zip file would you like me to open it or or you know it's a zip
208. 33:18file and there's probably something inside it so let me go ahead and open it and then it writes the code to open the zip file
209. 33:25and then it reads the the stuff inside and then it tells you what it is and then it goes and makes charts and graphs and
210. 33:34wild who's it apologizing to you or itself or the world well retropunk the the um like it's it's basically just the way
211. 33:44they program the application layer so you've got the large language model down here and you've got Chachi BT and the
212. 33:50fine tunings is this application layer sitting on top of the base model so the base model doesn't do any of that [ __ ]
213. 33:57the way they wrote the chat bot they want it to be friendly and to give you what you want and to accomplish your goals and so I
214. 34:07think programmed into it is you know be a polite person and if you make mistakes apologize and you know whatever whatever
215. 34:14they're doing in the programming that's why it does that and I think that's also why because it does that if I Met Myself
216. 34:22saying please and thank you to it um it seems to get better if you if you treat it if you don't treat it like an
217. 34:30[ __ ] which you know there's anthropomorphizing stuff and then there's just talking to a thing that is it's essentially the equivalent here
218. 34:40this is the way I think about chat EBT you have access to an idiot savant intern so they're completely clueless
219. 34:53you have to tell them everything but they know everything right so they know all the facts they know all the knowledge
220. 35:01but they don't but they're clueless so you have to say Okay idiot savant intern here you know here's who I want you to act like oh okay
221. 35:11here's what I want you to do ah okay and then I want you to you know go analyze all the crap in that image okay
222. 35:19let me write the code for that it's like I don't know it's just crazy this is crazy I noticed you said please I know it like
223. 35:30it's it's habit it's it really is like working with a person and it you know what's funny is like I I
224. 35:38feel like covid because I was a I was a massive non-believer in work from home I I just and I I still like like the the amount of um
225. 35:53like the emotional connection to my company feels like it's been bled out of the company like work from home all the tasks are getting done
226. 36:03but the connection to like the feeling of being a part of something is has just been bled out so I still am not a a huge
227. 36:12believer in work from home I think it's we lose the humanity piece um and uh but but there's there's something about
228. 36:22the way this thing I I feel like kovid kind of trained us how to work remotely and if you don't talk to someone for a
229. 36:31month you you just interact with them on slack and on this and on that and you know they occasionally turn on their
230. 36:37camera so so we're used to interacting with people remotely now and chat is just kind of like I have another employee and they're smarter
231. 36:47than any employee by far I've ever had and they're smarter than me by far but they're complete idiot and it'll lie and it'll screw things up
232. 36:57and then just go sorry boss I'll try again hey the the names of those colors are really boring sorry about that apologies
233. 37:05let me go ahead and make those more interesting how's this amazing amazing amazing Max Amaze anyway welcome everybody if you're
234. 37:16new here my name is Kyle Shannon this is the AI learning lab I am an entrepreneur I'm the founder and CEO of a company called storyvine
235. 37:25uh I am if you've been here for the last 10 minutes I I am I am an Enthusiast for this generative AI stuff but I'm also
236. 37:35um I've kind of been on been in this Rodeo a few times I was I was there for the early days of the world wide web and
237. 37:41started one of the early digital agencies company called agency.com and the parallels between the early days of
238. 37:47the web and the early days of this generative AI stuff are insane except this generative AI stuff way more profound and it's coming at us
239. 37:58way faster so um so that's who I am that's what the channel is about if you have questions or comments or ideas or things you're
240. 38:06trying to do pop them into the comments I'll do my best to answer them Emilio's wife good evening Mr Shannon good
241. 38:11evening Emilio's wife hope you're well all right let's see here where was I what template are you using to go live
242. 38:22to use both cameras it's called dual camera and it's just when you go live I have this dual icon and it they took it away
243. 38:32for like a month because I don't know maybe that was a bad boy were you bad I might have been bad and they they took
244. 38:40it away from me and said you've been naughty all right let's see I noticed you said please it's just great what template are you
245. 38:53using we just did that one read New York Times article today saying 10 drop in users was the end of the AI Revolution
246. 39:01that's [ __ ] hilarious well that was like when the.com bust happened right you know like like you know
247. 39:09to be fair like way too many companies were valued way too high but but when they predicted the end of the internet
248. 39:16when the.com bust happened like we're all sitting there going like uh have you used this [ __ ] no the internet's not going away
249. 39:26um and it didn't and you know it [ __ ] changed everything same same [ __ ] here I mean to put in context the the
250. 39:32difference in speed of adoption the World Wide Web took six years to get from you know when it was created to 100
251. 39:40million users chat GPT took six weeks six [ __ ] weeks 200 million users fastest adoption of consumer technology
252. 39:53and history now Zuckerberg is going to try to claim threads is getting there faster which it is but they're basically
253. 40:00just doing that on the back of Instagram so it's not fair if you if if you have you know whatever a 300 million you know
254. 40:07user base to build on it's not that's not not a fair thing so I'm gonna I'm sticking with chat GPT um as the fastest but but that adoption
255. 40:16of of kind of zero to 100 million people in six weeks on a janky ass piece of software because chat GPT is is still it
256. 40:24like it is it's a very thin piece of software it does not have a lot of features it's got a lot to be desired but
257. 40:33it's a simple interface to remarkably powerful technology and they did a smart thing and paired up with Microsoft so
258. 40:41they have enough Cloud compute to actually run the [ __ ] and pay for the [ __ ] you know 10 billion dollar
259. 40:46investment I bet they've burned through a lot of that Sam Altman has said he thinks they need a hundred billion
260. 40:52dollars to get to AGI artificial general intelligence and I think it's his goal is three years maybe three years from now
261. 40:59I have a feeling they're probably closer than than they let on but we'll see um anyway wild yeah the the end of AI because of a 10 drop um
262. 41:11which platform are you talking about I don't know what I was talking about so if you remind me what I was saying I'll
263. 41:18maybe do that just join saw you had an image up what did you do with it which image this image I don't know which image it was digital
264. 41:31product coach uh if you tell me what image it was I'll pull it back up um but if it was this image this is this
265. 41:39is my little cheat sheet it's not even a cheat sheet this is if you haven't played with chat GPT the top URL there
266. 41:47is the official chat GPT website it if if you haven't played with it you just you just need to go play with it
267. 41:54you need to go play with it you need to go play with it I used to be polite about it now I'm just like no like it
268. 41:59like it's [ __ ] changing everything so so just do that um bing.com is also chat GPT but it's this is Microsoft's implementation of it
269. 42:11um the nice thing about Bing is that it's gpt4 for free you have to pay for gpt4 up here this is for free you can make images there
270. 42:22um it's connected to the internet so bing.com has a lot going for it and then poe.com gives you access to six
271. 42:28different models plus a whole bunch of bots like chat Bots that are really specialized like a mid-journey maker and
272. 42:34a HTML coding bot and uh I made a Shakespeare bot and a Yoda bot and a sling blade blot bot so so you can have
273. 42:42some fun there and then prompts.chat will teach you about prompting how how to interact with these tools and future
274. 42:49PD is just a big old directory of AI tools there's like probably close to 4 000 of them now um so I don't know if this is the image
275. 42:58you're looking for but if you're looking for the other images let me know I'm happy to put them there
276. 43:03all right how can we explain the vibe on your LinkedIn live today um it was It was a held it was wild it was that that one dude I
277. 43:18forget his name I don't know if it was it might have been you half-assedly I don't think it was um kind of the intent I mean it's called AI
278. 43:26office hours so my LinkedIn live is just an audio thing I call AI office hours and I do it every Friday and and the
279. 43:32basic intent was it was you know if people are trying to figure out this AI [ __ ] I would just answer questions
280. 43:39um and today I brought some people up and it was I don't know it was going good and then the one dude came up and
281. 43:43he's like you know they've started World War Three and ai's like like why are you trying to figure out how to how to do
282. 43:50copywriting with AI when it's going to destroy the planet so it was strange but whatever that's the deal with going live with
283. 43:58this sort of stuff how do you download for Windows download what chat GPT you don't you just go there with your
284. 44:03browser you just go there with your browser and you just do it in your browser you don't need to download the download it's a web app
285. 44:13unless you're talking about something else and if you're talking about mid-journey you have to do that in
286. 44:18Discord so you would download Discord for Windows but you can also do that in a web browser all right did you know
287. 44:26when you switch between your cameras you go offline for five or so seconds I did know that and I'm sorry I got I I talk fast and I forget
288. 44:36so I will try to remember to count to three every time I switch the camera so it's just I you know listen I am here to
289. 44:46disappoint I and I and I it seems like I've succeeded in that so look I mean look at the [ __ ] hair what's going on with the hair today
290. 44:54you know this is the eyebrows are starting to take on the life of their own again it's just just things are going south here we're going
291. 45:03into the weekend strong code interpreter can explain what it is code interpreter can you explain what it is and uses
292. 45:15um I'll do I'll go back to that um because it's I don't so I just got access to it today I don't really know how to use it but I'll show
293. 45:22you some couple of little things that I've I've learned retro is correct okay okay I'll be better when I switch my cameras stop
294. 45:31yelling at me will the lawsuit have any impact yes no I don't know maybe I don't know which lawsuit I assume you
295. 45:42mean the lawsuit of the book publisher saying hey open AI you trained on our [ __ ] you shouldn't do that
296. 45:49here's here's how I think all these lawsuits are gonna go um 10 billion dollar investment from Microsoft Microsoft isn't going to want
297. 46:02to piss off their advertisers so I'm sure Microsoft lawyers are sitting in open ai's offices going okay here's a list of the
298. 46:13some amount of number 247 companies that you can't be on the wrong side of we will defend you but we want you to cut deals with bertelsman
299. 46:23and CNN and fox and that whoever the big advertisers are for Microsoft I would assume that they're saying to open AI if
300. 46:33any of these people are upset with you you gotta you gotta you gotta find some way to make it right and and that's probably why they took
301. 46:41browse with being the connected to the internet version of chat GPT that's probably why they took it down they're
302. 46:47probably you know appeasing someone cutting deals whatever the hell they're doing until we cut a deal you know you
303. 46:53can't have that tool out there so either go fix it or take it down um and then I think everything below that is just going to be noise and
304. 47:01they're not going to deal with it so so I think though there there will be though there's going to be lots of
305. 47:06lawsuits I just just like my sense of capitalism is it will only be the big lawsuits that will matter and most of
306. 47:15those big lawsuits won't even get to the lawsuit stage it's probably a call to the CEO of Microsoft going uh you're
307. 47:23gonna make this right right so um I think mid-journey which is a small company with a cowboy CEO that you know kind of
308. 47:37flaunts yeah I don't give a [ __ ] about that it's everyone's images are free um I think they have a real shot at being
309. 47:47sued into Oblivion and I think that stable diffusion stability AI potentially does as well they don't they
310. 47:55don't have a big backer like Microsoft my instinct is that the reason stability AI is open sourcing all of their
311. 48:04technology is as a hedge against getting sued into Oblivion because all their Technologies open source so if
312. 48:13they're like well okay stability AI we won't do stable diffusion anymore we won't do that but the model's already
313. 48:20been trained and it's already open sourced it's already out there right so I that's my sense is that that's a hedge
314. 48:26against that sort of [ __ ] I but it's going to be fascinating like I think the next 20 years are just one lawsuit after
315. 48:33another for for two decades I am so kind to chat GPT I'm thinking Karma I think that's a good idea yeah
316. 48:41like don't be an [ __ ] to the AI bot because if it decides that it doesn't like you and now that it's got code
317. 48:48interpreter and can write all this code and execute it you know it it might come after you so yeah be polite to your
318. 48:56little robot Overlord can confirm when I get mad at GPT it performs three to times four times worse that's really
319. 49:03interesting wow that's cool some comments in this live were filtered to protect the community's experience don't
320. 49:09be [ __ ] and comments please if you're gonna troll be creative a good troll is hard to distinguish whether it was a compliment or an insult
321. 49:20so do that be smart enough to do that and don't be [ __ ] to each other please like we got enough of that in our daily life
322. 49:27you know what I'm saying all right um what format do you think code interpreter would want to analyze
323. 49:36financial data it doesn't [ __ ] matter like it's you can upload anything you can upload unstructured data you can upload
324. 49:45you can upload a scan of an invoice and it will scan it OCR it pull the data out of it and put it into
325. 49:56a spreadsheet or you can upload just a bunch of crappy broken data and it will figure it out already too late I don't know what that
326. 50:06was in reference to but maybe I'm on here too late what time is it it's 10 o'clock Do You Know Where Your Children Are oh I do I do do you
327. 50:18hello Bob how have you been hmm I've been looking into generative AI have you heard of it I heard it's a waste of time
328. 50:27I think the robots are going to kill us I'm trying to find my place in the comments but I can't because I'm old and
329. 50:41tired and my eyes don't work and my hair looks crazy new development's coming what what are the new developments about a victory
330. 50:51mistake are we all learn mistake or we all learn mistake yep I don't know what you're talking about but
331. 51:00if you put it in some form of English it just seems like random words together 100 agree with you on the work from home
332. 51:08thing after working in an office for 20 years before that I know like you know what it is I kind of feel like what
333. 51:14kovid did is it turned my company into a set of to-do lists and everyone does their to-do list and we check in and we check in and do our
334. 51:25to-do list and we'll occasionally do like you know the haha it's good to see you good to see you oh you're digging
335. 51:31out from snow oh it's really hot Where You Are but what's missing is that unspoken palpable sense of how we doing
336. 51:43you know when you like work in an office with people and you can just feel like what's going on and you can sense is
337. 51:50there enthusiasm are people pissed off is it just everyone's busy and they're keeping their head down and doing the work
338. 51:57that's gone it's gone and so it sucks it sucks it's my company it's an 11 year old company I [ __ ] love what we do I love the people in it
339. 52:07and I can't stand how [ __ ] disassociated it feels it's oh and and you know and listen I get the I get the thing like don't force us to
340. 52:20come back to work man we like working from home don't be the [ __ ] man but something's been lost and it and it hurts the people that
341. 52:33don't want to come into the office as much as it hurts the people that do want to come into the office
342. 52:37they don't know it because they're like I get to work from home [ __ ] it hey man price is right's on yeah I'll do that
343. 52:43I'll get to my work I got my to-do list I'll get it done go [ __ ] yourself it's hurting them too I'm new here welcome all right
344. 52:55good evening loved your feed oh love your feet thank you met a victory I appreciate that hey Grandpa hey little Whipper Snappers get off my
345. 53:05[ __ ] lawn goddamn kids these days [Laughter] you're probably one of those ones like Chachi BT's a waste of time man don't
346. 53:15get heavy on my scene I'm just sitting here working on my copywriting for my social media blog come on Grandpa pick it up there Skippy
347. 53:25you have to mind your p's and q's Kyle I know I do what AI can I use to create website and mirror other websites okay I'm gonna
348. 53:33switch my camera now and I'm going to pause so I don't lose you for some moment I want to talk okay um I don't know I haven't um
349. 53:45kept up to speed on all the website building [ __ ] that's happening there's a lot of it I know that so go to
350. 53:53futurepedia.io that's a that's a list of almost 4 000 AI generated tools there are probably a lot of you know
351. 54:05talk your website into existence tools down there most of them are gonna suck I mean the thing that you need to know about
352. 54:12futurepedia.io and and actually the phase that we're entering because of this AI [ __ ] is there are going to be an infinite number of very
353. 54:25slick looking websites and very slick looking applications that claim to do miraculous [ __ ] that are all just spit and duct tape there's
354. 54:37like an incomprehensible amount of them are coming and are already here so of the 37 or 3 800 tools at Future pedia
355. 54:49we're probably lucky if 10 of them are are solid it's like it I would my guess is it's probably 95 pieces of [ __ ]
356. 55:00um that won't survive will take your money and waste your time but how do you find the five percent that
357. 55:08are really good I don't know you got to go dig through them so go to Future pedia go to the web you know I'm sure
358. 55:14there's a web design or a web building part of the part of the directory and just go start playing but I I don't have a good answer for you
359. 55:22there sorry can I talk now um thank you very much thank you thank you welcome we've got a lot of people
360. 55:36here if you're new here my name is Kyle Shannon this is the AI learning lab uh it's I give advice on hairstyles
361. 55:43and oh no uh sorry that was the other Tick Tock channel that one failed miserably see uh I have totals
362. 55:56uh I like talking about AI stuff and you know kind of in and I like turtles kind of way I did that on slack one day someone
363. 56:10posted something and I just typed I like turtles it just it caused a kerfuffle people didn't people were seriously concerned
364. 56:21for my well-being they didn't know the meme so if you don't know the meme I like turtles go find it on YouTube it is [ __ ] hilarious is
365. 56:28your name Pfizer because you made my heart stop oh Bilbo Baggins you're the best Oh see now that's a solid troll right
366. 56:41that's somewhere between a covid Joe like a like a covid burn conspiracy theory complement and a troll that is
367. 56:51that's a solid comment for for the AI learning lab that you get can I can I do the heart button let me see no anyway give yourself a
368. 57:02heart Bilbo that's a solid one just got code interpreter far out that's AI tools Hub Matt uh so so do me a favor Matt
369. 57:13um uh start you know start playing with it while you hear me rambling and and you know I'll jump back here to it here
370. 57:19in a second but um but start playing with it if you discover anything cool let me know that's super exciting Bilbo ball Baggins
371. 57:29is there any way to transplant open AI to my brain well that's what Elon musk's trying to do with neural link um I mean
372. 57:43you know kind of joking aside there sort of is right you can get there there's um automations what are they called on
373. 57:51on iOS they're not called automations they're called the things that you play anyway I think are they called automations whatever the
374. 58:00[ __ ] they're called there's things you can do on um on the iPhone where you can basically just speak you can call Siri and just
375. 58:10speak to it and then it'll send it to GPT and then it'll talk back to you so you know starting to get there with that also
376. 58:17that the the pie there's a new company called I think it's called inflection and their product is called pi and
377. 58:24that's a that's a a GPT like tool that you can talk to and I think it talks back to you um they just raised 1.3 billion dollars
378. 58:36for the series B that's a lot Instagram has a billion users I thought they might have a billion users yeah so so threads
379. 58:44threads getting to 100 million users faster than chat GPT is [ __ ] because everyone just went add on my Instagram people
380. 58:53that's how they got to 70 million people it's still pissing off Elon Musk like it's fun watching those two those two
381. 58:59getting ready for their cage match did you do you see the thing Zuckerberg did where he made a comment about uh about
382. 59:07Elon sleeping with Sergey brin's wife it's like holy [ __ ] boys it's like it's getting personal we'll never Facebook ever again I know
383. 59:19Facebook's just a giant steaming turd but but you know the uh the battle between him and Elon Musk is pretty funny uh never wait
384. 59:29never iki Kami Zuck have a buck I don't know what you're trying to say there met a victory but I like the intent of it I
385. 59:37think I see where it's going uh it looked like an mid-journey image then you did something in code interpreter
386. 59:45okay so I was in mid-journey so this this image uh I'm gonna switch cameras this image was um someone had asked me earlier to how
387. 1:00:02would I make a logo for a business and I showed them uh something in bing and then I showed them this in mid-journey and
388. 1:00:12they're not they're not great logos but they're but these they're certainly good Inspirations like that that that could
389. 1:00:18certain I could certainly see that being the basis to do a design off of um The Prompt that I used I I went to
390. 1:00:29poe.com and had it write a prompt for me I I put in a crappy prompt it rewrote it color logo of a flower shop flat
391. 1:00:39illustration graphic on white featuring a blooming flower with the delicate petals blah blah blah um so that was that image and then yeah
392. 1:00:48then I went over to code interpreter which is just another mode of chat GPT now so I'll go ahead and you know I'll show
393. 1:00:59you something that I did so okay this is old but I should be able to see what I did before right or not no I guess not
394. 1:01:08all right we'll go we'll go do a new thing so I'm gonna go new chat I'm gonna go code interpreter so so what
395. 1:01:17code interpreter is like I think code interpreter is the the worst name of a product that that like like if they tried to come up with
396. 1:01:28a worse name for what this thing does I don't think they could have I I feel like chat GPT helped them write the
397. 1:01:34worst possible name for a product it this is a really bad name this is sort of like um I I it's like um like um
398. 1:01:47file inspector or file magic or I'll figure [ __ ] out for you I don't I don't know what the I don't know what
399. 1:01:55you would name of a code interpreter's bad name so anyway what it does is when you've got code interpreter when you're
400. 1:02:01in code interpreter mode there's a plus button next to your prompt so what you can do is you can upload a
401. 1:02:08file and so I uploaded two different files one was um one was um some data about Chess matches over the past 10 years or something like that
402. 1:02:19and one was a was a an image a mid-journey image and whatever you upload Chachi BT looks inside it understands what that file is
403. 1:02:30and then you can start querying against not even querying against it you can start asking it to do things with the data in that file
404. 1:02:39and it writes its own python code and just figures [ __ ] out so so what I just uploaded is is a spreadsheet of this chess data that I
405. 1:02:51got off of what was the site that I got it off of begins with a K it's a site where you can get a bunch of data anyway whatever
406. 1:03:01doesn't matter so I got I got some data it's it's chess data so I'm going to say okay I'll say please just just to piss people off please
407. 1:03:11um visualize um um what is in this spreadsheet
408. 1:03:25in a way that is visually compelling and brings the data to life
409. 1:03:40okay so it's a zip file right now so so basically I don't know what's in this file I haven't opened this file it's a zip file
410. 1:03:48and so I say please visualize this it says sure first let's extract the contents of this ZIP file to see what's
411. 1:03:55inside so it understood it was a zip file now it says the zip file contains a CSV named games.csv let's load the file
412. 1:04:03and inspect the data it does that now it's telling us here's the the fields that are in the data and it explains what those fields are
413. 1:04:12and then says given these columns we can create a variety of visualizations for instance we could look at the
414. 1:04:17distribution of game lengths compare the ratings blah blah blah blah could you please specify the aspects of data
415. 1:04:23you're most interested in visualizing how about um uh victories um ratings let's see victories chess
416. 1:04:39ratings and game lengths sure let's create some visualization based on these parameters so it describes what it's going to do
417. 1:04:52and what it's doing when it pops up these little um uh crap we lost our focus focus so so there's there's a
418. 1:05:0349.9 games won when white started 45.4 when black started 4.7 percent were draws here's the distribution of white player
419. 1:05:14rankings so about 1500 chess rating um there's black player rankings which is a little higher and then here's a distribution of game
420. 1:05:27length number of turns so it looks like most most games end around 40 35 or 40. and it Trails off and you know there's a
421. 1:05:36handful of games that go to 200 moves so so and if you go look at I know you can't read it the focus is all [ __ ] up
422. 1:05:43but if you go look at the what it's doing in these little boxes it's writing python code it's writing and executing python code
423. 1:05:53to parse the data in that file so that's what code interpreter is it's like my face is melted and I knew it was coming I've been
424. 1:06:06waiting for it for a while now one two three four five okay hopefully I'm back um anyway hope that helps you digital
425. 1:06:19product coach I only use two oh only use tick tock what is AI um Nikola Tesla FX I I if you're serious I'm happy to
426. 1:06:31answer that artificial intelligence all the crap we're playing with is really machine learning more than artificial intelligence
427. 1:06:39I missed LinkedIn live um maybe if you're on LinkedIn I do it's every Friday at 2 p.m Mountain 4 P.M Eastern it's just an audio it's kind of
428. 1:06:52like um Twitter spaces but it's linkedin's got this live audio thing which is okay but the problem with it is like they
429. 1:07:01don't promote it very well and it's just Audio Only so it's I can't show anything so I don't know I'm gonna keep doing it
430. 1:07:07for a little bit because I feel like linkedin's a different audience so it might be worth doing I don't know
431. 1:07:14insta has tracking across platforms yet do you like David whitcock's material or Wilcox material uh I am conspiracy theory adjacent I
432. 1:07:29don't you know I I think all those guys go a little far most of the times but some of the stuff that they he talks about is interesting
433. 1:07:39um it's just like when it I don't know it just like it it just goes into play I we're in such a [ __ ] polarized world
434. 1:07:47right now I have backed off of anything I I try to figure out you know what are the trigger words for both sides and I
435. 1:07:56just don't have any [ __ ] interest and so conspiracy theories live on both sides of the of the the nut house right now and so I
436. 1:08:05just you know I just try to keep it down the middle and um it's kind of interesting what's happening with uh
437. 1:08:12with um what's his name Stephen Greer um it looks like his stuff's getting a bit more like mainstream acceptance with
438. 1:08:21the UFO stuff like the UFO stuff's really weird right now I have a feeling a lot of that is is [ __ ] that we've
439. 1:08:28built that we just don't want anyone to know about but I don't know I don't know had not covered God bless you what with
440. 1:08:35hair like that I know that should have a hat on it shouldn't it or that like something should be done with that but whatever
441. 1:08:41I still have hair I mean at my age that's not bad hair perfect yeah hey now hey better Victory not a disappointment I just
442. 1:08:51didn't want to miss anything oh that's cool uh yeah the switching thing now I know I'm just I I like making fun of
443. 1:08:58myself because if I don't you know then someone else will run greetings from Israel yeah I noticed Ron I I saw on
444. 1:09:06Facebook or LinkedIn or somewhere you're you're in Israel so I hope you're having a good time there unless you're from
445. 1:09:13there in which case I still hope you're having a good time there but I assume you're traveling um I keep seeing llm on AI articles what
446. 1:09:22does that stand for please okay so llm stands for large language model so Chachi PT I'll give you a couple of things here
447. 1:09:31chat GPT the GPT in GPT stands for generative pre-trained Transformer you don't need to remember any of this
448. 1:09:40but know that it stands for something generative basically means it's generating original content pre-trained means
449. 1:09:48you don't have to bring your own data historically with machine learning you would have to bring like your Radiology
450. 1:09:53data and you would specially train a model to do this one specific thing and everyone kind of had all these different
451. 1:09:59methodologies and things like that the T in in GPT stands for Transformer so the Transformer is technology that
452. 1:10:08Google actually invented and they put out this paper in 2017 called attention is all you need they invent this thing
453. 1:10:14called the Transformer what the Transformer does is it allows these machine learning models to be trained on
454. 1:10:21massive data sets before they had to be relatively small and you could throw massive amounts of compute power so gpus
455. 1:10:29graphic processing units that happen to be perfect for this kind of processing and so what openai did what what chat GPT did
456. 1:10:40um is they said hey if we can train these these models on massive data sets why don't we go find all of the possible
457. 1:10:49data we can on the public internet every let's let's scour every corner of it we can get to so whether it's Wikipedia or
458. 1:10:56blogs or websites or research papers or all that sort of [ __ ] and so they trained it so so when you're
459. 1:11:05interacting with with chat GPT the the pre-trained part is it's been pre-trained on all of the internet so it
460. 1:11:12kind of knows everything that's been put on the internet for the past 35 or 40 or 50 years like a remarkable amount of data
461. 1:11:20um the large language model is sort of the base layer that they trained on all of this stuff and then there are
462. 1:11:27different large language models like meta released one called llama and there's different sizes of them and then
463. 1:11:34chat GPT actually sits on top of that large language model and it's actually an application it's like an app that is
464. 1:11:41calling that large language model so when you put a prompt into chat GPT it goes thank you very much sends it into
465. 1:11:48the large language model pulls out the relevant data and spits that back in your face so so that's what a large language model is the the
466. 1:11:57why they keep mentioning llms is this approach of being able to use the English language or whatever your native language is really
467. 1:12:07um but the ability to use the English language as a programming language is new and the fact that you can ask it to
468. 1:12:16do programming and ask it to do poem writing and marketing writing and problem solving and all sorts of stuff
469. 1:12:25like that is is new and we're going to be able to take these large language models and place them on top of really
470. 1:12:33complicated systems like imagine imagine so you so you know how health care sucks and is really kind of criminal and we're
471. 1:12:45all angry about it and companies like United Healthcare who's my health care provider that I'm not happy with and really raging all the
472. 1:12:55time about you know how they write these policies that are like 460 pages that you can't comprehend at all
473. 1:13:04well imagine having a large language model that you could sit on top of that and just read that [ __ ] and actually
474. 1:13:12come up with strategies for how to get around their [ __ ] that they they assume you're not smart enough to get
475. 1:13:18around because no one reads those rules well that exists and that code interpreter thing that I just showed you
476. 1:13:24you could take the PDF of all of your United Healthcare policy [ __ ] that they don't want you to read upload it to
477. 1:13:31that thing and have chat GPT give you strategies for how to work around all their [ __ ] so that's why large language models are
478. 1:13:39a big deal is because they're going to put this this translation layer on top of all of these remarkably complicated systems
479. 1:13:47that we have to deal with so that's one of the reasons I'm really excited about it just don't cover your head oh that's so lovely hmm
480. 1:14:00it's like I got a [ __ ] trampoline on my head tonight how much does chat GPT pay for computer cloud storage I don't know but it's a
481. 1:14:08[ __ ] lot Sam Altman said when when so so when they launched chat GPT they got to a million users in five days and
482. 1:14:16there was an interview with him like like 10 days after they launched so they were probably up to like 30 or 50
483. 1:14:22million people at that point and he said um they were interviewing him and someone said like you know how much does
484. 1:14:28it cost to run these things and his response was the cost is eye watering um so you'll notice when when Microsoft
485. 1:14:38announced their investment in open AI they said okay we're going to give them a billion dollars then they launched
486. 1:14:45chat GPT and you know it got to 100 million people in six weeks I don't think it was much more than a month or
487. 1:14:51two after they announced a billion dollar investment that Microsoft said okay now it's a 10 billion dollar investment
488. 1:14:58um I think most of that 10 billion dollars is in it's it's not just cloud storage but it's also so Cloud compute
489. 1:15:05right so it's it's the gpus so every time you do a chat GPT inquiry it it costs a lot of money there's like
490. 1:15:14massive massive super computers that are you know processing all this stuff it's amazing permanent band will ensue if your video
491. 1:15:28does not end within five minutes that's weird maybe I need to go I'll come back