aiLL July 8th, 2023 - Exploring Generative AI Trends, Ethics, and Possibilities

Transcript

1. 0:21welcome welcome welcome to the AI learning lab happy Saturday night I was just watching a little Tom Segura his new special
2. 0:35what's it called I forget he's a funny funny mofo I like Tom Segura uh but I'm Pom-Pom thanks for your time
3. 0:45on these lives or for these lives you're very very welcome I appreciate the kind words good day good day good day
4. 0:55welcome everybody we'll wait for some other folks to come on in feel free to share the live greeting Mr Vasquez
5. 1:05greetings greetings greetings yeah I think hitting the hitting the heart button lets people know we're live too so
6. 1:17if you haven't talked about it recently would you mind talking about API keys and their use I can talk about that sure
7. 1:27I assume you mean do you mean just generally or or using open ai's API keys let me know let me know because it's a
8. 1:36it's a broad topic and I don't go very deep in it but I can give you some I can give you some of the basics
9. 1:43will they ever bring back Bing on chat GPT I don't know apparently when a lawyer gets out of their ass oh man there's a lot of people here
10. 1:52welcome everybody my name is Kyle Shannon this is the AI learning lab a little bit about me I am an entrepreneur
11. 1:58I've started I don't know 12 or 15 companies in my day I should probably get that number that's something I
12. 2:04should probably know uh my current gig is I am the co-founder and CEO of a company called storyvine which is an
13. 2:10automated video storytelling platform and this channel is all about exploring generative Ai and dealing with uh what
14. 2:18it is and what it's going to do and and sort of talking about you know now that we've hit this inflection point where
15. 2:24chat GPT has happened hey Emilio's wife class is in session yeah class with the disturbing Professor
16. 2:36um uh you know now now that now that you know the the cat is out of the proverbial bag with with all these large
17. 2:43language models and chat gbt um what the hell do we do with this and how do we make sense of it how do we
18. 2:48learn about it and what do we do and is it going to take your job and is it gonna you know are the robots gonna kill
19. 2:54us all that sort of stuff um I tend to lean on the more optimistic side of the the conversation I tend to
20. 3:01tend on the less technical side of the conversation I'm more about what do we do with it now that it's here rather
21. 3:07than you know how do we build the models and do the engineering so that's why I have a few questions about AI pop them
22. 3:14in the comments below I'll do my best to answer them and uh you know other things like I I swear sometimes so if you don't
23. 3:22like that I would encourage you to find another AI learning lab that has better values and then uh what else uh I'm kind of a smart ass if
24. 3:34you're gonna troll be creative about it don't just be a jerk and don't be a jerk to other people in here please and don't
25. 3:39bring in politics and let's not have Division and yeah yeah [ __ ] like that all right let's go answer some questions
26. 3:46he's legit adorable no he's so cute what are we gonna do oh do I learn about it yeah all right I don't know why I did that
27. 3:59because I'm an idiot [Laughter] all right let's see old as in 1994 well I don't know what that's in reference GTS wait GT's G
28. 4:15as old as in 1994. I assume you mean you've been here before and you've heard me talk about uh starting uh a company in 1994 which was
29. 4:26one of the early web companies which is one of the reasons I think this this uh what's it called generative AI stuff is
30. 4:33such a big deal as I was there for the early days of the web and started one of the early digital agencies and this is
31. 4:39very very similar to that so anyway if you haven't played uh is that top one there chat.openai.com that is
32. 4:49um that's the official chat EPT website so if you haven't gone if you haven't played with it just go there right now
33. 4:54and play with it and if you don't know what to do go to that document prompts.chat and that'll teach you what
34. 5:00prompting is and how to talk to these things and what they really are and it'll give you a bunch of ideas about
35. 5:04sort of personas that you can give chat GPT to act like like act like an accountant act like a strategist act like a marketer act like
36. 5:14a screenwriter whatever it is and and it's it's a really good document and then Bing and Poe are also chat GPT
37. 5:21variants that you can play with being in particular is gpt4 connected to the internet and you can make images there
38. 5:29for free you just type in make me an image of a I don't know 1969 GTO Judge in a field and it will it won't look
39. 5:37like a GTO judge but it'll be close it'll look like a 60s 70s muscle car and then pro.com gives you access to a
40. 5:44bunch of different models and then future pedia the one down below is uh just a big directory of AI sites and
41. 5:51services and [ __ ] like that most of them are crap but there's some good stuff there all right let's get into the questions okay
42. 6:06one to add to this oh want to add this competency on my resume what's a good way to learn what I need to claim confidently
43. 6:14um can you spell AI you got to get those letters right like if you're dyslexic it's a [ __ ] nightmare so I'm an expert at IA
44. 6:28um let's see well I I you know I I would I would say you know the reason I have these URLs on screen for
45. 6:39these lives is that's where I think you should start um just start to understand what you can do with chat GPT
46. 6:49start solving problems with it um I don't know quite what you are if you're a project manager or if you're an account person or if you're a
47. 6:57salesperson start to understand what you can do with these tools for your job just start to get AI literate about what's possible
48. 7:06here and then what you can do for what you already do there and that's a place to start and then the the thing that I would add to it is
49. 7:14start talking about it like from day one like tonight go there if you haven't and just start [ __ ] around have it
50. 7:23write a song have it write a poem have it write a project plan have it write a recipe have it write a meal plan like
51. 7:29doesn't [ __ ] matter just go there and start playing and then share that share it on Facebook share it in your community group share
52. 7:37it with your family even if they think you're [ __ ] weird and I I heard ai's gonna kill us all why are you doing that
53. 7:43that's stupid you're dumb dumb Jenna you know Jenna I don't know why you do all this computer stuff it just takes you away from your family
54. 7:51just ignore all that [ __ ] I assume your family doesn't talk like that but if they do I'm psychic uh um and start telling people about it
55. 8:04because um that may lead to a job for one thing or a different job or a different opportunity in your current job but
56. 8:11what's gonna start to happen and it's already starting to happen is that um people are increasingly going to want
57. 8:17to figure this out and they're going to start asking who do I know that knows something about this and initially you might be crazy Jenna
58. 8:24who just talks about AI um but at some point they're gonna go oh I think I gotta deal with this I just
59. 8:31got laid off because of this stuff what is this stuff can you help me um be that person be that educated person um
60. 8:40my hair looks good so do you thank you I don't know that hair looks good that hair looks like it needs an intervention ah
61. 8:55all right how do you get rid of business when it's Tanked how do you get rid of business when it's tanked like if you started a business
62. 9:06I don't know you gracefully shut it down you thank people for their hard work you thank your customers and you do what you
63. 9:12can to get out of contracts I don't quite know what you mean there um if you need like legal advice I'd get a lawyer
64. 9:24but I don't you can clarify that can AI give lawyer advice it can but you need to check it and double check it if
65. 9:33you're talking lost stuff but you can have it like what it can do really well is it can read um it can read documents and and there's
66. 9:40this new thing called code interpreter which now is available to all chat GPT Plus members where you can upload like it seems like
67. 9:49any kind of file and it it sort of figures [ __ ] out about them um but you could upload a big lawyerly doc document
68. 9:59and have it analyze it and help you understand it and potentially even come up with strategies for how to respond to it um
69. 10:09and then you know unless it's kind of a low stakes thing like like um a few months back we started we started
70. 10:17this newsletter called everyday Ai and then some partner from Andresen Horowitz also started a thing called everyday AI
71. 10:25um and that was a case where like we didn't want them to do it but it didn't really matter so we're actually thinking about you know
72. 10:32having an AI lawyer maybe we could get a like a a reporter to follow the case where we were just going to use a
73. 10:39full-on AI lawyer it ended up not going anywhere we ended up dealing with them you know without needing to do law stuff
74. 10:46but if it's something that is actually really important then you should get a lawyer but you can absolutely use chat
75. 10:52CPT to understand documents to come up with ideas for strategies things like that and then again um go to prompts.chat and you know look
76. 11:03at the the structure of those prompts where it tells the the um chat GPT to act like a certain Persona and have yours act like a lawyer
77. 11:16bow all right let's see okay this is this one's actually important best singer Mariah Celine or Whitney Celine bugs me
78. 11:31I think you gotta go with me I think you gotta go Whitney I mean Mariah's remarkable celine's remarkable
79. 11:39but there's just something about her that I don't know and and you know Whitney's just you know she's just on a different level it's
80. 11:49Whitney it's it's Whitney uh there's a fun there's a fun Christmas we had one year where my wife's family she's she's from the Bronx and
81. 12:00they they came up we had a country place in Pennsylvania and I heard all this screaming like after dinner like we
82. 12:07finished dinner and I went up to watch TV with some of the guys and and there's just you know loud loud
83. 12:13chatter and it's going on for like an hour and it's and it's getting escalated and it's turning into a fight
84. 12:20and so I went downstairs and I said what the [ __ ] is going on like what what do you and they were they had it was like
85. 12:27an a 90-minute full-blown argument about who was more famous Barbara Streisand or Michael Jackson those are the best arguments
86. 12:40what operating system are you I I'm an apple Fanboy so I am OS X through and through don't give me that Microsoft [ __ ]
87. 12:49except in the last six months I've become a Microsoft fan non-swearing classes down the hall exactly teach me something okay I'm six two um
88. 13:02taller people sometimes have to duck their heads through doorways hope that's new information for you
89. 13:11um are you a robot no but I am chat GPT 5. don't I look realistic yellow little smart ass eat tonight I
90. 13:26like it I like it I haven't I have a sense there's some there's some wine drinking going on there might be a whiskey or two
91. 13:35you know it might be might be some uh some people playing the ganja what company is doing the most with AI well
92. 13:43Microsoft is [ __ ] with the most competitors and and doing it joyfully open AI is the thorn in the side of the most companies
93. 13:54and those two are like you know they're little pals right now they're little corporate Buddies [ __ ] with the world um
94. 14:01don't discount Google they invented the Transformer which is the T and GPT that's Transformer that Google invented
95. 14:09now their Bard model sucks but it'll get better so they're doing a lot in AI um but they're definitely playing catch-up
96. 14:21um meta is trying desperately to detangle itself from its name but it's not a meta's not really about the metaverse per se I mean you know
97. 14:34meta has an A in it and then AI like if we just added a an i at the med AI we could we're an AI company right so but
98. 14:43they're actually doing some interesting [ __ ] where they're they're releasing um some of their technology into the
99. 14:50into the wild as open source which which is interesting um so they're doing interesting stuff mid-journey
100. 14:58um is for the image generation tools in my opinion by far the the uh the best of the image generation tools but uh
101. 15:08stability AI that does stable diffusion they keep um releasing their um all of their technology as open source which I think
102. 15:16is another interesting move um then there's anthropic and then there's um inflection which just got funded for 1.3 billion dollars
103. 15:33um but that's the dude that was the founder of um Deep Mind so he's he's a literal [ __ ] Brainiac um so there's some some good pedigree
104. 15:44there um those are the biggies Nvidia seems to have a a relative Monopoly on gpus for for AI processing which I find baffling
105. 15:55but it seems to be the case AMD is trying to come up and do something with them but like the the like all the
106. 16:02software stuff is so entrenched with Nvidia that I don't know it's that that's going to be a tough one to topple um apple is
107. 16:13Apple to me right now feels like a shark circling the boat [Laughter] because they've got their own silicon that processes
108. 16:25um machine learning locally and they've got a real good reputation for privacy and they're [ __ ] Apple when they when
109. 16:31they get into this stuff it's gonna be really good and they're gonna like they're gonna kind of reinvent it like right now ai is
110. 16:37really geeky and you gotta go to chat GPT and you gotta learn about prompt engineering and now there's this thing
111. 16:43called code interpreter and you got to figure out how to what it is and how to upload a file and how to query a file
112. 16:48and then you watch it do its thing and it doesn't do it and you kind of it's like everything right now is just this
113. 16:54janky spit and duct tape don't get me wrong it's remarkable profound gonna change the world forever technology but
114. 17:01right now it's janky and duct tapey um those are the biggies for me those are the ones I'm paying attention to
115. 17:09I'm looking for a good free AI to help with coding well go there go there go there go there
116. 17:24that's why I have it here and when there's a better one I'll put a better one here so so if I I don't know if
117. 17:33um if co-pilot on GitHub is free but co-pilot on GitHub is a a an a GPT coding assistant so you do pair
118. 17:44programming where it does kind of code completion for you but gpt4 does code really well and they just launched code
119. 17:51interpreter which I'll show you here in a second it's mind-blowing it you upload a file and then you start
120. 17:59asking about it and it writes its own python code and executes it what it essentially does I'm starting to understand it now
121. 18:07when you when you launch an instance of code interpreter like a code interpreter chat what it's essentially doing is spinning up like a
122. 18:17virtual computer and then you ask it to do something and it's like sure and then it writes python code it installs libraries if it needs
123. 18:27to it runs the code if stuff times out it's like oh well let me try another thing it it basically keeps running code
124. 18:36until it succeeds and then it gives you what you want and it's doing this all in this little kind of virtual computer
125. 18:42it's like you have this little this little coding assistant this little technology data analyst assistant just sitting there waiting
126. 18:51um so if you haven't played with code interpreter you should it's not free but it's 20 bucks a month and for what it
127. 18:58does it's [ __ ] remarkable but if you just want it to do code like GPT 3.5 does code really well um bing.com is gpt4 so I don't know how
128. 19:10well that version of gpt4 is for code but it's probably pretty good because they're the ones that put together um copilot for GitHub
129. 19:19and then there's a bunch of if you go to the Bots at poe.com they're all free and there's a bunch of coding Bots if you go
130. 19:25into the technology into the explore button at poe.com here I'll I'll show you a couple of those [Laughter] um po
131. 19:41so when you go to po.com it's just a normal chat chat bot kind of Chat gbt thing there's all these different models
132. 19:48you can play with but this explore button here you can go explore Bots that other people have made and then click on the programming one
133. 19:57and these are all programming bots so python machine run python code HTML machine JS machine Pie golf Geo Json it
134. 20:11was just a ton of them and you know you put in some stuff and it'll write code for you so lots and lots of things to
135. 20:20play with but but honestly the just chat GPT is is the deal so so here's check this thing out so what you see on screen
136. 20:28right there this is code interpreter this is just [ __ ] insanity so what you're looking at is you're looking at a pie chart of different
137. 20:37colors I know that's obvious it's like duh Kyle we can see that uh but let me let me read some of the color names
138. 20:43midnight mystery Chestnut chestnut horse muted mustard Rusty Nail creamy Pearl Golden Dawn Autumn Blaze cocoa Darkness
139. 20:53Lagoon Ripple these are all really good color names and then there's these different colored slices so what this is is this is code
140. 21:02interpreter so what I did was I uploaded this image which was an image from the AI Salon this is one of the women in the
141. 21:17AI Salon made this which is super cool so I uploaded that image and and then I said to chat GPT or I said
142. 21:25Dakota interpreter whatever the [ __ ] it is chat GPT I said find the top 15 average colors in this image plot their
143. 21:33relative percentages on a pie chart make each segment the color represented and label each color with a creative Crown
144. 21:40color that is memorable and brings that color to life and and so then what it does is it came up with its own instructions number one
145. 21:49load the image and flatten it down to a list of RGB colors apply clustering using the k-means algorithm for instance
146. 21:56to find the 15 most common colors calculate the relative percentages generate creative and memorable names
147. 22:03for each color plot a pie chart so it came up with its own instructions I mean I I did it in English but it sort of
148. 22:08wrote them so it could understand it and then it said Okay first uh loading the image and applying k-means clusters
149. 22:15blah blah blah blah and and that one failed and then it says I'm sorry it took too long because that image was too
150. 22:21big but what I can do is I can down sample that image and I'll down sample it now what's Behind These Little Boxes if you open these up
151. 22:30it's python code it's loading libraries it's writing code it's running code and then if it times out it gives itself new
152. 22:40instructions and it does the next thing down so this one down sampled the image there it found the color clusters
153. 22:48um now it says now let's generate creative and memorable names and so it came up with the names and the
154. 22:55relative percentages and then I guess on the back end it's got the the actual RGB values or the hex values
155. 23:02then I asked it to create the chart and it screwed up once and I asked it to create it again and it screwed up again
156. 23:07and I asked it to create a third time and it did it and so so that's it with an image you can also just upload a zip file full of csvs full
157. 23:19of data and it just knows what they are um and and can analyze it and write charts about it and so anyway that's that's code interpreter um
158. 23:32it's just it's just mad nuts it is happy Saturday everybody hope you're all doing well Silver Fox I asked chat the exact
159. 23:47question and it gave me a learning path and job subscriptions or job suggestions which which question has chat the exact question
160. 23:57I don't know which question you're talking about we'll get back to it I have a prompt engineering wait I have
161. 24:10prompt engineering on mine oh on your LinkedIn got a job data automation Tech at twenty dollars an hour
162. 24:17keep talking about it keep putting it out there twenty dollars an hour seems low to me um just but keep going that's good like
163. 24:25yes far out good for you keep going and uh keep knocking on doors because as more and more companies start to hit
164. 24:36the panic button and realize oh [ __ ] we gotta figure this stuff out who knows about this you know about that I'll pay you double
165. 24:43that's gonna start happening that happened in the in in the early days of the web I had when when it when it got
166. 24:49really frothy like like a year or two before the the.com bubble burst like this like 90s late 97 early 98. I had
167. 25:00three different employees walk into my office and say hey I love working here but um I just I I was out with drinks
168. 25:09with some people I didn't know and I I talked about some hat idea I had about the web and they funded me for 20
169. 25:15million dollars I had three people in one month quit because they got funded for like 10 or 20 million dollars
170. 25:22um so so keep talking about it keep putting it out there and keep exploring like even if you're even if you got the job
171. 25:30um doing data automation Tech or whatever that is um just keep learning keep exploring keep playing on the weekends just keep
172. 25:37keep expanding your mind learn what this code interpreter can do I I can't get my head around it yet it's
173. 25:45it's too insane what it's doing like it's prompting itself writing code executing code correcting it if it gets it wrong and
174. 25:57giving you a result and then if the result's not right you're like no I don't like it it's like okay here I'll
175. 26:01go again it's like it's like it's like it's like a golden retriever it's like a coating golden retriever
176. 26:09thank you for answering my question I'm a junior Cloud engineer been trying to use it for all I can oh that's awesome
177. 26:15that's awesome Jenna yeah just I mean keep going and the other thing I would encourage anyone if if you're sort of
178. 26:21digging into it use it in your specialty use it in the stuff you're doing but also start to explore the other tools
179. 26:28right if you haven't done image generation go play with mid-journey um if you're geeky install um stable diffusion and automatic 11 11.
180. 26:38um if you're a level geek geekier than that start playing with Lang chain and start doing your own embeddings and play
181. 26:45with auto GPT and baby AGI and some of these autonomous agents things because again the more the more you can get your
182. 26:52head conceptually around all of what's happening I think the better off all of us are going to be so it's a good time
183. 27:00to just geek out just get you know OCD and ADHD tunnel vision and whatever the [ __ ] you do if you're really
184. 27:10um if you have executive function unlike me make a list check it twice and just you know go down through the list and
185. 27:16teach yourself a bunch of [ __ ] if you're more on the add side of the thing just [ __ ] dick around
186. 27:23dick around until you find a rabbit hole that you want to run down for three days and then do that can I marry AI what AI stocks you like I
187. 27:32just talked about a bunch of companies that I liked so I I would never um take stock advice from me I have a
188. 27:39degree in acting and I'm shitty with money so don't take stock advice from me I um all the all the stock [ __ ] I do I
189. 27:46do it completely emotionally like data I mean now that I have chat GPT and and code interpreter in particular now what
190. 27:55I might do is get emotionally hooked on some company and then I'll actually go run the analysis have it run the
191. 28:01analysis and just tell me you know what I should know about it uh but yeah you can probably marry AI I
192. 28:08would be surprised if we don't see some sort of headline in the next two years where you know someone applies for marriage Hood to a
193. 28:18to a bot that's happening because it quite frankly it probably goes someone makes the news because they did
194. 28:28it first right and then a bunch of people do it and then it becomes a reality show right isn't that America isn't that the way we go
195. 28:36like the first person does it like you're a [ __ ] freak you can't marry a robot dumb dumb and then two years later
196. 28:42we're like watching you know bot love the bot Bachelor oh I keep freezing ah [ __ ] should I um restart I might need to restart my
197. 28:58router um let me jump jump to the bottom here fear and greed control the market AI is still young has a long way to go very young
198. 29:06are you concerned with the safety of your data with when you download from these AIS yes am I still freezing a lot
199. 29:17fine now okay not freezing for me great Amelia's wife sounds like it you must have Comcast Xfinity too it's the worst
200. 29:28but at least it's expensive okay so I won't restart anything all right let me go back and find where I was in the
201. 29:33questions if you haven't been here before I'd like to answer all the questions so I get a little OCD about that but you know
202. 29:41um API key someone asked about API Keys okay so so wait a minute so let me I'll start at a I'll start at the I'm going to try to
203. 29:56keep this as as non-geeky as possible um at the the base definition of an API it's called it's called the app uh
204. 30:07application programming interface so an API is basically a set of instructions that says if you want to send something
205. 30:14to my computer or want to pull something from my computer he here's the API and what the API is um you can ask for
206. 30:24um let's say let's say that the computer holds movie reviews right and so the API might say you can ask for movie title
207. 30:35um number of reviews you know average ratings and the text of the reviews or whatever it might be and then you know
208. 30:43that that's if you're asking things here and then there might be another API that says you can submit a review and you
209. 30:50could submit a movie title you could submit a review you could submit you know the rating whatever it might be so
210. 30:56the API is just the set of instructions to to say if you want computer a to talk to computer B that uses an API the API key is is is
211. 31:08basically your magic token and I'll show you one in a second well actually I won't because if I show it to you let's
212. 31:16see do I have any old ones well I can you know oh you know what I'll do here we'll go to we'll go to chat gbt hang on hang on one second
213. 31:33all right um um create 10 fake API keys so I can see what they look
214. 31:48like okay so API Keys tend to look like you know how sometimes when you when you go to someone asks for a password and then
215. 32:00and then the your password software goes do you want a strong password and it makes some kind of random ass thing if
216. 32:06they're kind of like that so so it's it's generally a hexadecimal string of numbers and sometimes it's got a little
217. 32:13letter and dash at the beginning of it like uh like uh open AIS I think have SK or something like that and then it's
218. 32:20just a big long list of numbers this big long list of numbers is essentially the the key that Associates you your account so so if let's say
219. 32:34you want to do API calls to the open AI thing you want to write in you want to write a chat bot and you want that chat
220. 32:40bot to send [ __ ] to open Ai and get back a response well you put in your API key and what the API key does it says securely okay
221. 32:50this person has permission to use the open AI API and then when one of their customers uses their application we're going to
222. 33:00pass that key along and then we'll know that we're going to charge your account you know 0.2 cents every for every
223. 33:06thousand tokens that that customer uses so it's literally your key to be able to use the API so how you get it for for open AI
224. 33:18is you go to this thing called The Playground so you go to platform.openai.com playground and this is where you do your prompt
225. 33:28engineering you can you can choose different models like you can choose I now have access to gpt4 woohoo so does
226. 33:35everyone I'm not special I've got GPD 3.5 turbo 16k and 3.5 turbo they all have different pricing and things like
227. 33:43that you can read it in the documentation but then you go to your account and you say view API keys and it
228. 33:51automatically doesn't display the full thing here but basically you say create a new key and then it gives you the key and you
229. 34:00copy and paste that and then for example where you might use that is let's go to zapier oops
230. 34:17it's like I'm going to zapier but not come on Internet is slow tonight so so what zapier is is you've you've
231. 34:30got these little automations that you make out of these kind of Lego blocks and I'll show you what I did
232. 34:36this is what I did the other day as an example but so um here's a Google form so basically when someone inputs things into a Google
233. 34:46form it's going to take the input from that Google form and then I'm going to send it out to open AI and I've got a prompt in here if I look
234. 34:55at the thing here's here's my prompt right so you know I want you to use the following variables and then I got a variable
235. 35:02there and then I've got a a prompt write a poem about a funny colored animal in the style of Edgar Allan Poe so this is
236. 35:10going to get sent to open AI when I when I first set this up with my account it asks me for my open AI key or the API key
237. 35:22so now that I've established it I don't have to put it in anymore but but when you first do this and connect your
238. 35:27account to it it's going to say where's your API key and then you will have copied that to your clipboard and then
239. 35:34you paste that into your thing and then you're off to the races so hope that helps I mean here's the good news
240. 35:50the good news is we're early and you know if you want to do things like automations with zapier you kind of need
241. 35:58to know what an API key is but you don't need to get much more geeky than that code interpreter I feel pretty [ __ ] stupid right now
242. 36:06with code interpreter because I don't know [ __ ] about data analytics and it's really good for that but there's going
243. 36:13to be all sorts of things it's good for like like people are just now starting to talk about you know examples other
244. 36:20than the examples we've heard about for the last two months like all the examples have been the same worst I'm
245. 36:25starting to see some new ones crop up but I I feel like in the next month or two as I get my head around it like I'll be
246. 36:33able to share more about it because because it yeah it's so it's so powerful like I like it there's probably so much it can do I can't even
247. 36:49comprehend right now I mean I know that I know it is I know it is so anyway I don't know that's what's going on so I'm
248. 36:55feeling so so here's the thing the thing about these Technologies is even though chat GPT is this simple Google like thing um
249. 37:07where it's like well it's actually really simple well it's actually not chat.openingi.com chatgpt is a simple
250. 37:14interface sitting on top of Decades of you know deep computer science and deep engineering and all of this stuff is is
251. 37:26um evolving in real time and and it's incredibly powerful stuff so in in the current state of the generative AI world
252. 37:36if you're not coming at it from the technology side if you're coming at it from like the end user side you know
253. 37:41your project manager your you know an art director you're a copywriter you're a pro whatever all that sort of stuff
254. 37:48you're gonna need a little get a little geeky right now to get good with it like if you really want to start to do things
255. 37:54like automations and things like that um but if you wait for a year um It's Gonna Get Easier and easier and
256. 38:02easier right these things are ultimately going to turn into our agents they're going to be like little personal
257. 38:06assistants we have we're just like hey Sally why don't you go you know fire up you know you know these emails and then go order
258. 38:16my groceries and come up with a recipe for tonight and it'll run off and do that [ __ ] um but right now we kind of have to all
259. 38:26build that stuff ourselves the Bills should use AI could they win Super Bowl that's funny what do you feel are the
260. 38:40most impactful AI startups talked about that um Celine bugs me too Whitney gets my vote as well yes Whitney's it all right
261. 38:48how to build an app no code experience so I just kind of showed that um the way the way zapier works is there's
262. 38:57three things you need to know with zapier there's an input and then there's you sort of do something with the input and then
263. 39:05there's an output it can get way more sophisticated than that but that's the basic concept and they call the input a trigger and so
264. 39:13the trigger is something like a new row gets added to a spreadsheet or a new um document is uploaded into a folder or
265. 39:24somebody fills out a Google form that you created so the the content generation engines that I've been doing
266. 39:31are I'm using Google forms so I'll create a Google form that'll have a couple of inputs I then go into zapier and I say okay I'm going
267. 39:42to connect it to that form and when someone puts in new data in that form I want you I want that to act as a trigger
268. 39:48you're going to take those inputs from that form I incorporate them into a prompt that gets sent with the open AI little
269. 39:58Lego block from zapier that gets sent to open AI I get back to response and then I put it in an email or I send it to a
270. 40:04slack Channel or something like that so just at its most basic there's an input you process it and in this case send
271. 40:11that stuff off to open AI you can do lots of things in there but that's what I've been doing lately and then you do
272. 40:17something with it you save it to a drive you add a row to a spreadsheet you put it in an email something like that
273. 40:26um there's lots of no code tools here here's the deal with no code tools while there are no code they're not um
274. 40:36no complexity they tend to be you know they're they're conceptually as complicated as coding is because you still need to think about
275. 40:48the logic okay I need these inputs do I have someone fill in a spreadsheet do I have them post something to a slack
276. 40:55channel do I create a form is it is it going to be an input on my website is it something coming out of
277. 41:02Salesforce it could be all those right so you gotta you gotta think logically and critically and yeah and then you
278. 41:08think okay now that I've got the the starting point what's the prompt that I'm going to write and how how am I
279. 41:13going to incorporate that data into that prompt that's a whole nother sort of logical [ __ ] crazy straw you got to
280. 41:20go through and then like okay now that I got open AI responding with some sort of response and it does it I got the
281. 41:26response I want cool awesome um now what am I doing with it how do I get it back to the person or do I get it
282. 41:33back to the person all that sort of stuff if you look at what code interpreters doing this new thing it's doing a lot of
283. 41:41that logical thinking you're just kind of giving it a goal and then it goes and does some [ __ ] now zapier just added a
284. 41:52large language model input where you can you can just describe the automation you want and it will attempt to build you
285. 41:58that little Lego block thing I find that I find that harder than just doing it but that's probably just because I started doing it um
286. 42:09the the tools that I know of that are worth looking at zapier's been around for I think 10 years now they're pretty
287. 42:15good but it's it's it's janky um there's one called make m-a-k-e that's like zapier but it's kind of got
288. 42:23a Slicker interface and it seems like it might be a bit more modern in terms of its structure but I I don't know
289. 42:30and then if you want to do apps there's one called Bubble and then there's a couple of others people have mentioned in the live here
290. 42:37um but but I would just you know go uh go to Reddit or go to Twitter and look up no code tools and just find one that
291. 42:45doesn't suck and doesn't make your head explode definitely not a robot I'm here for the wine exactly I although
292. 42:54I'm I should go get a actually there's there's a bottle of wine out my kids had some wine tonight maybe I'll get a wine why not
293. 43:03why not get it wine why not all right it was like a dad joke but I have enough
294. 43:18self-awareness to know that it was a stupid dad joke so I made fun of myself making a dad joke get it what AI can summarize audio files uh
295. 43:32well uh summarize so whisper so if you go I was showing before let me switch my um good playground so if you're at open
296. 43:45ai's playground so again the the URL for this is platform.openai.com playground and if you come here and you go to the complete mode
297. 44:01in the upper right hand corner of this window is a little microphone and if you click on that that's whisper
298. 44:10which is open ai's audio transcription you can upload a video or an MP3 or a web you know web audio and
299. 44:20um or you can just record stuff and it will take that audio and transcribe it for you you can also do that via the API
300. 44:30so you could you could probably do audio transcription through zapier um there are also lots and lots of tools
301. 44:40um where they put it in a in a sort of easier front end if you go to futurepedia.io just look up transcription or audio or
302. 44:49things like that there are probably I I would be surprised if there were less than a hundred Tools in futurepedia.io
303. 44:58that do that I tried uploading audio file and I couldn't it couldn't summarize it if it's too big you may
304. 45:05need to break it into chunks that's again where I'd go to futurepedia.io and and look for ones that say you can
305. 45:12upload up to two hours of a file right um so so there's lots of things out there that you can use the the open ai's
306. 45:21Whispers really really good but even in the playground it's limited to a certain length so you may need to chunk it up but that should help
307. 45:33all right I don't know I tried to use it for C sharp and it failed a lot then it comes to wait file to file reference
308. 45:46not quite sure what you're talking about there um yeah depending on what you're trying to do um GPT can be a mess with programming
309. 45:56you know I would say if you're trying to do serious coding stuff I'd go to GitHub and use copilot um because it's more it's it's specially
310. 46:04designed for coding um so is code interpreter available through the open AI API I don't know where's the bottom I I
311. 46:16don't think so but I don't know like I I didn't dig into that today that's that's a question that I have as well um
312. 46:24I I assume it will be but I don't know so it should be are all plug-ins accessible through the API I don't think so I don't think they
313. 46:36are right now the plugins right now are a little little janky you can make your own plug-in so wait if you can make your own plug-in
314. 46:46you're interacting with the API so they probably are I don't know I don't [ __ ] know but but you know how you now have
315. 46:53function calling within the API right within the openai API so you can kind of write your own plug-in function calls
316. 47:02plug-in like function calls within the code um I haven't done I haven't I haven't needed to and so because of that I
317. 47:09haven't dug into that part of the the world um but if you find out let me know that would be good to know I don't have code
318. 47:16interpreter yet lasai oh that sucks um I just tell you um did you just cabrono I assume um I assume you've gone to your settings to
319. 47:30just to to your uh beta features and and enabled it if you haven't um go do that how I could work in AI um
320. 47:43I don't know what you mean silver fox how can you get a job I don't know start learning it start talking about it
321. 47:51would there be some time when AI helped network security yeah and and there's also a time when AI gonna attack network security
322. 48:03there's going to be all sorts of new you know white hat and black hat battles going on where where people are using AI
323. 48:09increasingly to exploit vulnerabilities and then other people are going to be using AI to catch the bad guys and
324. 48:16they're going to be able to catch the bad guys in increasingly sophisticated ways so yeah that's coming
325. 48:23um do me a favor we're kind of dwindling down here if you would just hit the uh I know it's been a little boring here for
326. 48:28a section I've been geeking out a little bit but um just hit the like tap the screen get some likes going and if you
327. 48:33would share the live so we can just let let some people know we're here that would be great uh um are you concerned with safety of your
328. 48:41data with the download of these AIS um yes so so I wouldn't put anything super critical in there um I I'm not super worried about it you
329. 48:54know for me personally but I do a lot my company does a lot of work with Pharma clients and they're you know
330. 49:00appropriately anal about their safety and privacy and and things like that so anything we're doing for clients
331. 49:08um we're going to be creating you know Standalone servers and firewalls and all that sort of [ __ ] so um so if if you need to do
332. 49:19proprietary kind of stuff and it's actually super critical then you need to look into that you know go go look at
333. 49:26Microsoft Azure like how they've Incorporated open AI calls and the options they have for for like like
334. 49:33stability or I mean open AI announced three months ago that they've got these business servers that you can buy that I
335. 49:42think they start at like 25 Grand and go up to a million where you can have your own Standalone instances of open AI servers um so
336. 49:51thank you for all the hearts that's appreciate that yeah AI is still young we have a long way to go totally Ryan it
337. 49:57is we are we are in such early days if you're feeling at all like I'm behind I can't all these people know more than I
338. 50:04do my sister's I can't I can't right now because I don't know anything nobody [ __ ] knows anything if you know the engineers the people on
339. 50:14the science side of this there's lots of people that know [ __ ] there but on the on the post chat GPT side of this where
340. 50:20people are now incorporating this into their businesses this is brand new for [ __ ] everyone it's like like even the
341. 50:28even the uh the people in the data science side of this um I I was watching more of the interview with George hotz who's the
342. 50:35who's the comma AI guy and you know super super smart programmer he said he said the the the difference
343. 50:43between how he was using AI five years ago and what's happened in the past six months he said it's like it's not
344. 50:49anywhere in the same ballpark of of how he thinks about Computing anymore um it's a completely different thing so
345. 50:59all this generative AI stuff the Chachi PT stuff and the the large language models this idea of you know all the
346. 51:06stuff the the the image generation things and people reading brain scans and music being written and and songs
347. 51:13that sound like this artist and that all this [ __ ] is new all of it's new and and it's you know the the the the popular press has heard
348. 51:24about it has been talking about it for the last six months since November 30th but it's really only been this stuff's
349. 51:30been kind of in The Ether for maybe a year and a half maybe two at tops we're very very new and this stuff is transformative
350. 51:40fear and greed control the market I would agree with that but uh this channel exists to inject a little
351. 51:47optimism in the middle of that fear and greed so [ __ ] those [ __ ] don't be scared AI can be your friend got a bot and his name is Charlie
352. 52:01Charlie makes me nice stories in pretty pictures um about is probably more emotionally intelligent than all the broken toys on
353. 52:10hand that's [ __ ] genius I'm telling you man the the what's cool about the Bots is that the the Chachi BT things is um
354. 52:22you can ask them questions about really sensitive topics you know especially in today's environment and it doesn't get triggered it just
355. 52:33responds with like a balanced answer like just with with like nuance and like well you know here's an opportunity and
356. 52:41or here's an option and here's an option and and it's just way more nuanced and it is it's way more like like
357. 52:48these language models talk with just a [ __ ] ton more empathetic response because they don't have empathy but they respond more
358. 52:56empathetically than almost all humans so the botular that's good um not freezing for me fine now okay great
359. 53:04um not freezing for me this is the freezing stuff trying to find my thing where was I oh there's the fear and great fear and greed
360. 53:19Fear and Loathing in Las Vegas did y'all see the the [ __ ] sphere in Las Vegas that that new venue with with like the
361. 53:28LED outside of it holy [ __ ] is that thing amazing it's crazy like I saw one picture of someone driving down the Las Vegas Strip
362. 53:38and and the sphere was like at the like at the end of the road it looked like the Moon Rising but it was like an eye blinking I'm just like
363. 53:46[ __ ] trippy man um and tokens are okay tokens good question so
364. 54:01this is painful it's like why do they call it code interpreter because that's the technical name of what it's doing is code interpreting
365. 54:10it's a shitty marketing name and so are tokens okay so tokens are thank you for that how do I describe it okay when you type
366. 54:22a prompt that prompt gets turned into numbers and and then those numbers are sort of put into this big fancy s
367. 54:30mathematical formula that then determines the next most probable number called a vector and then that Vector is
368. 54:35associated with a word so all of the [ __ ] going on behind the scenes with um with chat GPT are Vector addresses they're basically these
369. 54:48long strings of numbers and so tokens are you know those those Vector calls basically and I think spaces and words or parts of words all
370. 55:00end up being tokens so how they charge someone before was asking about apis the the open AI apis how they charge for the API is for every
371. 55:11thousand tokens you get charged two pennies or a tenth of a penny or whatever the hell it is um that's just their definition of the
372. 55:21chunk that they're sending but it actually translates into like for us as humans if you've got a thousand tokens that are
373. 55:32being sent that actually translates it's about three quarters the amount of words so 75 or 750 words is about a thousand
374. 55:40tokens so it's not a direct thing and why don't they don't just call it words well because they're engineers and
375. 55:47scientists and they're called tokens and the token is the mathematical thing right so so that's what so tokens are
376. 55:54basically just kind of your word count and so chat gbt right now has a um a token limit of 4 000 tokens or about three thousand words
377. 56:05three quarters of the amount so what that means is you've kind of got this window of 3000 words that as you're typing into chat GPT
378. 56:14that's its context window so that amount is is essentially what's being used to to determine the answer and that's why
379. 56:23sometimes if you have a long conversation with chat GPT it seems like it like you you started out you told it
380. 56:29about your lovely wedding and then it seems you know 10 minutes into the conversation it doesn't seem to have any
381. 56:35[ __ ] clue about the wedding anymore that's because the Wedding part like drifted out of the context window
382. 56:41so that context window is growing um the the some of the API um uh things for for openai right now are sixteen thousand tokens which is
383. 56:52twelve thousand words so that's starting to be better and then ultimately gpt4 can do 32 000 tokens which is about 25
384. 57:00000 words and there's if you go to poe.com the that one right there um there's one of the Claude models that
385. 57:10you can actually use has a hundred thousand token limit and that's 75 000 words and it's it's the it's 75 000
386. 57:18words is the combination of your prompt and the response it's got that it's got a big token window so you could you could literally
387. 57:26paste a whole screenplay or a whole novel in there and start interacting with it right but you can't with chat
388. 57:33GPT because it's got a much smaller token context window so that's what tokens are tokens are what you're charged yes
389. 57:42tokens are what you're charged and they are Loosely related to the amount of words you can use it's complicated as like like so so
390. 57:52here's the deal all of this stuff we will not need to learn about um two years from now this stuff's gonna
391. 58:01all drift into the background the reason it's all in the foreground right now is because we're early right and so if you
392. 58:07want to play with this stuff you kind of got to get a little geeky you gotta understand what tokens are
393. 58:12you think when Apple launches their AI agents they're going to mention anything remotely like [ __ ] tokens nope the
394. 58:18developers might need to deal with terms like that but none of the tools will will be like this for long
395. 58:26um so no need for court reporters I I Court I I there no need for them or like like I look at I think one of the one of the areas um
396. 58:38that where I think America has been hurt the most in the past I guess it's been happening for about a decade
397. 58:46but the lack of local news local reporters because all these National conglomerates bottle bought up all the
398. 58:54local stations and basically just gutted the newsrooms um you don't have local reporting anymore a lot of what the local reporting was was
399. 59:04reporting on public documents and you know their you know permits and you know the building they're going to build and
400. 59:12the bridge they're going to do and are they going to fix this and are they going to fix that and how they're
401. 59:16changing streets and when are they doing the water mains um so a lot of that reporting is not getting done but a lot of that reporting
402. 59:24could be automated with with these tools um because if you got access to the public records you could write prompts that parse those
403. 59:35public records and write stories and get the word out right so so I think it's a double-edged sword you might not need
404. 59:42court reporters but you might be able to dramatically increase Court Reporting and local reporting and things like that
405. 59:50all right C sharp like using program code to run from other file the main file and using other file for info how um
406. 1:00:02I don't quite understand your question I'm also not an engineer so I'm not the right person to ask the the right person to ask
407. 1:00:09if you don't have a engineer buddy is right there so instead of just having chat GPT try to do the coding
408. 1:00:17have chat GPT tell chat GPD to act like a senior c-sharp engineer and and see if you can get it to tell you how to use it to do that
409. 1:00:30because I'm sure it knows it might just this might just be a sequencing thing if you can get it to tell you how it would
410. 1:00:37do something in a step-by-step fashion and then have it start to do that for you that might help it might need that
411. 1:00:44context first it's one of the things about these tools is they don't do well with sparse instructions if if you just
412. 1:00:51say you know write me a poem it's going to write a really generic poem If you tell it you want it to you know act like Emily Dickinson and
413. 1:01:00you know it's a master of her works and her tone and her this and her that and then you ask it to write a poem it's
414. 1:01:08going to be you know much more what you're looking for so it's no different than the programming stuff
415. 1:01:13I got it yesterday are there are there rumors that being AI will have plug-ins this would be a game changer against
416. 1:01:20open AI it's not going to be against openai because um so so Bing is from Microsoft they have already announced it's not a rumor
417. 1:01:29they've already announced they're going to support the openai plug-in architecture that means chat GPT plugins
418. 1:01:35will work in bing and they're going to work in all of Microsoft's office Windows 11. um Dynamics 365 uh
419. 1:01:47GitHub co-pilot all of the Azure cloud services um Microsoft invested 10 billion dollars in open AI so even though they may have
420. 1:01:58competitive products like there's there's some stuff that openai's looking to do that may be competitive with some
421. 1:02:04of the applications that Microsoft is doing they're they're um they they call it in the business World Co-op Edition
422. 1:02:13um there might be some areas where they compete and there's going to be other areas where they're deeply deeply
423. 1:02:17cooperating so that's where they are right now so yes Bing will support plugins and it is going to support open
424. 1:02:25AI plugins they're trying to make it a standard I assume and and the way if you look at the the nature of how the plugins are created
425. 1:02:33they're very very um they're very very it's a very very simple protocol so it would be a good standard for lots of people to adopt
426. 1:02:44they're very easy to write they're very easy to call you're basically just describing here's what this plugin does and here's
427. 1:02:51when you would use it and then that's what chat GPT uses to determine you know should I use a plug-in here or can I just answer this
428. 1:03:00um so yeah how do you sum object or function in a file um again I don't know I don't know the function calling stuff I haven't gotten
429. 1:03:11into that yet and I probably won't it's that's a level deeper than my pay grade and my knowledge so so just using Code interpreter right
430. 1:03:21now is is sort of bouncing up the the upper limits of my intellect it's making me feel like a [ __ ] idiot it was
431. 1:03:28funny I saw that on Reddit today so someone someone put on Reddit um does anyone else feel you know like immediately stupid again yes
432. 1:03:40I was answering something from a long time ago okay Silver Fox cool um y'all hiring nope sorry I wish we were
433. 1:03:48we're we're stable as they say which Stables better than laying people off right where's the bottom I got a bottom for you right here
434. 1:04:05name's Buck that's my bottom a bot is oh okay hang on hang on I lost my place again old man trying to read little text on
435. 1:04:21tick tock it's the Tickety Tock it's the tickety tog I'm playing on a Tickety Toc oh wait now I'm I'm way far behind or like back
436. 1:04:32where I was whatever I'm still lost hang on I saw them building it in February the biggest fear eyeball that thing's cool I
437. 1:04:41mean I was I was thinking about that like like like the uh AI like degenerative AI stuff and the generative video stuff you're gonna
438. 1:04:55they're gonna be able to come up with a generative AI thing to just dynamically generate stuff for that that sphere like
439. 1:05:01it's just gonna be insane it's so cool and apparently the inside of it's going to be even better I think a word is a token but so is each
440. 1:05:12punctuation mark yeah it's something like that but I think I think longer words they even break up into multiple
441. 1:05:18tokens it's just it's whatever like it doesn't really [ __ ] matter for for real people like it matters for for
442. 1:05:24engineers right because because you know they need to like for them a token is a token and you know they're doing whatever they're doing
443. 1:05:32with the token you know certain tokens are getting thrown out other tokens are creating probabilities of you know
444. 1:05:39semantic waitings or whatever I don't [ __ ] know but but it's basically three quarters however many tokens it is three quarters of that is
445. 1:05:48about how many words it is but it but it ain't exact the cost depends on the model you're using yeah the like the the GPT 3.5
446. 1:05:56models now are super super cheap gpt4 is more embedding is more but that also came down in price Kyle's saying words though I am saying
447. 1:06:09words what'd you do Saturday night I watched some old guy say words what did he say I don't really know
448. 1:06:18I don't know he seem to enjoy himself Anna I don't what am I doing with my life what what am I doing I need a drink
449. 1:06:29all right should I go get a wine I think I should go get a wine yes it's so true I was a small town newspaper reporter for six years right
450. 1:06:37like I mean holy [ __ ] like like all these National [ __ ] private Equity [ __ ] that came in and just [ __ ] decimated um
451. 1:06:48local reporting and local news and local journalism it's like [ __ ] um you know it just feels like there should
452. 1:07:02be some boundaries in [ __ ] society where you like you know that's not for profit you can make a profit but it's not its sole purpose
453. 1:07:15there should be laws but we would need to have a functioning legislature to have those wouldn't we bitter you all right
454. 1:07:26automating public interest stories could be awesome yeah I know it's I I like that is I feel like that's a
455. 1:07:33that's a business that's business you want to go start a business I've been I've been giving out free business launching advice that's one
456. 1:07:40go start the local news bot business business [Music] again I've never stopped feeling stupid I know me too I've like I feel like I
457. 1:07:57feel like oh I know everything and then I go to try to do something and I fail miserably like I don't even know where
458. 1:08:03to start what do I do [ __ ] it's just it's been like that for for I don't know it's like a year now it's
459. 1:08:12like I even feel like on these on these lives sometimes I'm like like am I actually is anything progressing or am I
460. 1:08:19just saying the same [ __ ] over and over again and then I realized like well some people are still just starting but also
461. 1:08:27like the fact that code interpreters here now like that's going to completely change how I use chat gbt I don't know how yet but it is
462. 1:08:35and and so and and then things like they take away they took away browse with Bing and then I saw on Twitter today someone else said
463. 1:08:44oh they also took away plugins but they didn't for me so so this stuff is just changing so much that I think it's just I I think the
464. 1:08:55the thing for all of us to do if you're willing to do it is just think of it as an adventure like we're on a [ __ ]
465. 1:09:01adventure and and what are Adventures well they're super exciting what else are Adventures they're super [ __ ]
466. 1:09:07challenging okay what else are Adventures sometimes they're really scary right sometimes they're really fun
467. 1:09:14but like but you're but you're committed to it right I'm gonna get the Grail we're gonna storm the castle right
468. 1:09:23whatever the [ __ ] Adventure is we're gonna go kill the dragon I think that's what I think that's the
469. 1:09:29mentality we have to be on for at least two years because it's just the stuff's moving too fast and it's it's like the it's just the
470. 1:09:41changes are going to be so profound that that I don't know like like you know the the people that were asking earlier about
471. 1:09:49well what job can I get how do I get a job and where do I get a job I don't know I don't know like I feel like the thing
472. 1:09:55that's going to be super in demand in the next six months is gonna be either non-existent or just a completely
473. 1:10:02different set of requirements six months out it's crazy it's crazy I tell you my head hurts welcome welcome to the welcome to
474. 1:10:11the adventure I know my head's my head has been like like I like I'm just bursting at the seams educationally I
475. 1:10:24feel stupid every day I I feel like I'm in a constant state of it's not quite anxiety and it's not quite shame but it's just that like that
476. 1:10:33thing where you know you're not good at something but you really want to be good at it and with this stuff it's like I feel not
477. 1:10:42good at it I really want to be good at it I'm getting better but I'm not really good like I just it I don't know it's hard right now it's
478. 1:10:53hard Kyle's saying words versus characters vis-a-vis tokens yes correct Thelonious Thelonious Monk you are correct
479. 1:11:03you are a call to action field trip buddy indeed this is your call to action people get on your Steed and head into the deep
480. 1:11:13deep dark Forest the one that shall never be spake laughs take your [ __ ] swords go to chatgpt get code interpreter or go make a bonnet
481. 1:11:29Poe or go I don't [ __ ] it doesn't matter it doesn't matter it really doesn't matter I mean it matters deeply and profoundly
482. 1:11:37but what you do doesn't matter and you know we were talking about this with the AI Salon I've got this thing called the
483. 1:11:44AI Salon that if you want to hang out with a bunch of adventurers um come join us go go to that URL the salon.ai and
484. 1:11:55we were talking about what is it about this group that is that that we like we've been around for six months it's it's a it's a really
485. 1:12:06cool it's it people are generous they're curious um they share what they know um they want to keep learning what is it
486. 1:12:15about them and and what we came up with was this idea that we're all on an adventure and like in a good Adventure
487. 1:12:20you have people with all different skills right like you have the the big dumb [ __ ] muscle guy who turns out to
488. 1:12:29have a heart of gold and you've got the the really mousy person that is not you know doesn't seem to have any skills and
489. 1:12:37then they get in into like an escape room and she's brilliant at cracking the code right like you have all these archetypes right
490. 1:12:45and I feel like that's where we that's that's where we need to be right now I was like the more we can connect and find
491. 1:12:53you know our little group of adventurers to kind of go out and do this with so rather than me having to become an
492. 1:13:01engineer like why don't I pile up with someone who's really good at engineering and they can do some of that [ __ ] and do
493. 1:13:08it really well and then I can do the [ __ ] I do really well and someone else could do this stuff they do really well
494. 1:13:13and and and you know together you can kind of learn this stuff and at least learn what's possible with it that's
495. 1:13:20that's where my head is lately find your adventurers get on the adventure get on the horse go go go go go go go go go go go go go go
496. 1:13:31is it scary yes is it worth killing the dragon well [ __ ] yes because if we done with the dragon's gonna kill the village
497. 1:13:40right someone's got to kill the dragon but that seems scary yeah is it scarier than the dragon coming and
498. 1:13:54eating the village I don't know why I went to you know fantasy metaphor but I think that I think that's where we are [Music] um
499. 1:14:09speaking of the legislature wait where'd that go oh here we go what do you anticipate will be the first
500. 1:14:20proposed legislation around AI oh [ __ ] I don't know if they do as I if if they do as my cynical self expects them to they'll ban
501. 1:14:36it but I think I I think they uh you know I I think the the Biden Administration is actually taking it seriously um
502. 1:14:50I think it'll probably be something around um companies like open Ai and Google like like some sort of you know legislation
503. 1:15:04around regulations around um what they're allowed to do what they're what some boundaries around what's
504. 1:15:14called alignment which is how aligned are these powerful models with the goals of the human race so I think they'll probably do some
505. 1:15:23regulations like if you're a company with as much resources as Google or open AI or Microsoft or meta
506. 1:15:31you can't make the evil bot that's going to kill kill us all right like it it'll probably be something around that um
507. 1:15:42I don't know I don't know I just I you know our government right now they don't understand what [ __ ] Wi-Fi is and they're they're you know
508. 1:15:51they're just not functioning as a government right now they're just they're just functioning as a money
509. 1:15:57raising machine that's that's you know polarizing all of us so we don't pay attention to the grifting that's going on um
510. 1:16:07and I think that this AI stuff actually requires good bipartisan functional government to like learn about it learn what it does well
511. 1:16:21learn how it can be a competitive Advantage for our country and for the world um and get their [ __ ] heads around it
512. 1:16:31get their heads out of the the [ __ ] petty [ __ ] that's going on do I have much hope that that's going to happen no I don't [Music] um
513. 1:16:43do I have deep hopes that something will shift yeah I do um and it's also you know I think it's why I'm doing these lives
514. 1:17:02because I don't think we can wait for government I what can they [ __ ] do like like the the open source large
515. 1:17:11language models are have no safety restrictions on them and are already almost as powerful as the stuff that
516. 1:17:18open Ai and Google have produced so that [ __ ] ain't gonna be legislated and then Europe's gonna do their AI legislation
517. 1:17:29and America is going to do their AI legislation but everyone else in the world is just going to keep rolling ahead forward so
518. 1:17:39if either the EU or or the U.S make it too restrictive they're going to absolutely kill Innovation and the rest
519. 1:17:47of the world is going to kind of subsume that like all the models right and and and you know the momentum
520. 1:17:55um and they're gonna do whatever they're gonna [ __ ] do so it's like I think that it's it's really important for us as
521. 1:18:04individuals to start using these tools in ways that bend it toward the light like it's the only way I can think about
522. 1:18:12it right like go do the adventure we gotta go kill the dragon but I don't know how to kill dragons it doesn't
523. 1:18:18[ __ ] if we don't kill the dragon The Regulators or someone or China or whoever something's gonna happen right so
524. 1:18:26it's on all of us to start learning these tools get AI literate and and start putting [ __ ] in the world that is bending it toward the good
525. 1:18:36I think that's gonna have I think it's gonna make a difference it's it at least lets me sleep at night right because I
526. 1:18:43understand it as much as I can and I'm doing what I can to get the word out about it and that's why this is here so
527. 1:18:51I know it's very philosophical tonight dragons and so many business ideas there are business ideas in August I'm going to
528. 1:19:00see my 87 year old mom going to introduce to Chachi PT and watch her head explode that's [ __ ] awesome I
529. 1:19:07know I know it's so it's one of my great joys in life is showing people that either have no clue what it is or or or
530. 1:19:15sort of like cocky but dismissive yeah yeah I've played with chat GPT and then you actually show them what it can do and they're like
531. 1:19:22I can do that really could it do my [ __ ] uh-huh I'm gonna see not really here look it's [ __ ] the best
532. 1:19:38because it's like you can't go like once you once you have that Epiphany about this [ __ ] like I don't think you can't go back you
533. 1:19:45can't unlearn what it does what it makes possible but that's going to be awesome that have fun with that who has time for a job
534. 1:19:53we've got three three hours of the AI learning lab every night exactly I know I feel the same way this is my job I looked at my I figured
535. 1:20:03out where to go look at my uh my my balance for all my earnings I have earned um 24.32 so far doing these lives I
536. 1:20:12can't take it out until it hits a hundred but you know so so this this is my job if I on an hourly basis of making I think I'm
537. 1:20:21making like 3.8 cents Thelonious you're evil can you make a video you know showing this stuff to your mom it would actually make a good
538. 1:20:30tick tock learning never ends exactly my mom is going to sus site she keeps hitting links the shuts her PC down yeah
539. 1:20:38that's rough that sounds that's what she calls exactly control alt delete mom exactly oh that's hilarious
540. 1:20:55[Music] okay the salon.ai yes that is the correct URL just it should go to a link tree um and then maybe just do HTTP instead
541. 1:21:11of https I probably set up the site wrong or something like that if it's not going there but that's the correct that's the correct URL
542. 1:21:19um and we've got a there's a meet up there so we meet every other week um on a Google meet and in person here
543. 1:21:26in Denver and then we've got a Discord where you can play with mid-journey and ask questions and hang out and then
544. 1:21:32we've got a LinkedIn Group which isn't as active but it's just a different kind of group do you think we can use our GPU to
545. 1:21:39provide computing power for AI just like Bitcoin mining yeah sure if you've got like a 30 90 if you're on a PC I'm a Mac
546. 1:21:46guy so if you've got a decent GPU yeah that can be used for this kind of stuff and it's you know um if you want to use anything big like a
547. 1:21:5665 billion parameter model you need Big Iron like George hotz the guy from Kama he's he started a company called tiny Corp
548. 1:22:08and they're creating a computer called Tiny Box which is going to be a fifteen thousand dollar machine learning
549. 1:22:16um computer that's got like six gpus in it and like just massive throughput and processing power and I don't know some
550. 1:22:25stupid amount of gigabytes of you know GPU RAM and [ __ ] like that um and that'll be able to run big models
551. 1:22:33but if you've got a single one you can like go to go to Reddit and Twitter and just start geeking out on open source llms
552. 1:22:43like the seven billion uh parameter models and they've got a bunch of them that have come out it's it's like every
553. 1:22:50three days they've got a new model I don't keep up with them anymore but a lot of those can be installed on local
554. 1:22:55machines like I know they've got some models now you can install on M1 Max and then M2 Max are apparently doing really
555. 1:23:02good with this stuff and then there's other ones that you can install on PCS um the the CEO of stability AI says he
556. 1:23:11thinks that by the end of next year so by the end of 2024 will have GPT 4 level large language models that you can
557. 1:23:23install on your phone that don't need an internet connection so these these things are like very powerful tools
558. 1:23:30you're going to be able to install locally so yeah if you've got a GPU you can do that and you can start playing
559. 1:23:35around with embeddings and all sorts of crazy [ __ ] if you want to geek out I've been playing with code interpreter
560. 1:23:40today for my restaurant and it's brilliant oh Silver Fox tell me tell me more like what what um what are some of
561. 1:23:46the things you've done that sounds really cool um then there'll be those of us still secretly using it yes well that's the
562. 1:23:56that that goes back to that Ethan Malik thing the secret cyborgs I think there's gonna be a lot of people that use these
563. 1:24:01tools certainly short term for like the next two years where you know your friends just seem to get a little bit more productive the big
564. 1:24:10secret is are you using chat GPT for all this no how dare how dare are you hahaha uh can't legislate AI anymore than you
565. 1:24:23can encrypt with encryption I know exactly it yeah yeah exactly like that's what that that's where I come back to like it it comes back to
566. 1:24:34it comes the more the more people that get AI literate the more it's going to help us stabilize its direction
567. 1:24:47um I don't I if we wait for government to do it they're just going to [ __ ] it up and they're going to take too long
568. 1:24:52this stuff's moving too fast so I think it's on us get on your Steed strap on the sword head to the forest
569. 1:25:03oh you're an Archer fine bring your bow I like hammers you too Thor you can bring your hammer um
570. 1:25:23all right Silver Fox it looks like everyone wants to know what's going on with your restaurant how are you using
571. 1:25:28it for your restaurant we want to know curious minds want to know what's the best AI bot you suggest for
572. 1:25:33trading stocks the ones that I have on screen for the entire three hours a night and if you're asking me is is open AI paying you to do this
573. 1:25:44no but they should be [Laughter] um go there like well okay two things if you want to find a trading specific AI
574. 1:25:55tool there's probably a bunch of them at futurepedia.io almost all of them are probably pieces of steaming fresh [ __ ]
575. 1:26:06that look pretty because they used mid-journey to make the images um there are probably some there but uh chat EPT especially with code
576. 1:26:21interpreter now you can upload piles of financial data and have it help you with the trading stuff teach you but like I don't know
577. 1:26:30where you are in your education if you're clueless it can teach you if you're really sophisticated it can help
578. 1:26:36you write Bots and could do all sorts of crazy [ __ ] so um for me I would be learning how to do it myself at chat.openai.com rather than
579. 1:26:47paying someone 20 or 30 or 50 bucks a month for their glorified prompt sitting on top of chat GPT because that's all
580. 1:26:54those things are going to be with some minor exceptions they're so let's say there's 4 000 tools at Future pedia
581. 1:27:01there's going to be 400 that are mind-blowing real companies with really cool technology built on top of this
582. 1:27:09[ __ ] and or or have their own technology and then there's going to be what's left 3 600 that are just steaming piles of
583. 1:27:17[ __ ] it's probably going to be 10 that's probably being generous we have to identify the dragons first
584. 1:27:24don't we yeah we do I mean well or we just need to know that um we keep hearing Rumblings in the darkness of the woods and we are pretty
585. 1:27:33sure those are dragons we don't even know if they're dragons or what they are yet I I agree with that the guardrails
586. 1:27:39scare me information control well Beer Hunter I know like that that's that's a good conspiracy theory you know skeptical place to start
587. 1:27:56um if you listen to Sam Altman talk he's very thoughtful about it um I would get thoughtful about this stuff because
588. 1:28:07it's easy to demonize a single company and say wow Sam Altman this and Sam Altman that and Elon Musk this and Elon
589. 1:28:14Musk that and Mark Zuckerberg this and Mark Zuckerberg that there's a whole open source Community out there
590. 1:28:20building these models with no safety guard rails on them and it wouldn't surprise me at all that within the next year
591. 1:28:30one of those non-guard railed models someone's going to use it to do some awful [ __ ] um so it's not
592. 1:28:42it's not like we don't we we need to be talking about and thinking about safety and alignment and and when we're talking
593. 1:28:50about alignment with these large language models it's basically as as these tools get more capable and they exceed our intellect
594. 1:29:05are they aligned with human values that's what when they talk about alignment that's what they're talking about
595. 1:29:12so the guard rails are really just these big organizations that have the money to do incredibly powerful tools you know looking at
596. 1:29:23how do we keep these things in line with human values because if they get too far apart that's when you get into the Sci-Fi
597. 1:29:32dystopian scenarios so it's fine to be cynical that oh you know the man is gonna is gonna suppress all of us um and
598. 1:29:48we need to be thinking about these issues three cents an hour I know I got the 3.8 in my head so I guess I'm okay at math all right
599. 1:30:02[Music] he's approaching the end again people ask more questions you're asking that's it if I get to the end it's over you
600. 1:30:12know that right so like like we got a nice tight crew in here uh normally we have a few more so I'm normally way
601. 1:30:20behind on the questions but I am getting toward the end I love that you guys are talking to each other
602. 1:30:37[Music] PC versus MAC for using chat GPT well for Chachi BT it doesn't [ __ ] matter because you're using Microsoft servers
603. 1:30:47you're using Microsoft Azure every time you go to um chatgpt you're not it doesn't you could run Chachi BT you can run it on
604. 1:30:56your phone it runs great on your phone because all of the computes happening out there if you're talking about if I'm going to
605. 1:31:03install a large language model locally what's better a PC or a Mac I don't know I think that the uh it looks like the M2
606. 1:31:09Max are they're getting some pretty impressive performance out of it and if you paid attention to the Vision Pro
607. 1:31:16announcement from Apple the R1 chip so so you have two chips over your eyes and those goggles one is an M2 Chip which is
608. 1:31:24an incredibly capable processor the other one's called an R1 chip and that's doing all the machine learning processing from all the sensors
609. 1:31:33um so they now have their own silicon that's just doing machine learning processing so I think long term apple is going to be amazing but
610. 1:31:44if you want like a the thing George hotz is building like a six GPU monster machine you're going to want a PC
611. 1:31:53so it depends on what your goals are like I think I think just sort of out of the box Computing Apple's gonna be amazing at this [Applause]
612. 1:32:02um but PCS are obviously way more scalable and way more flexible um I gave it my menus it broke them down by
613. 1:32:15ingredients oh that's cool I had to make new recipes oh my God that's so cool were they good now to work it all out before someone
614. 1:32:23else does that's super cool wow huh it's just this that's the thing about code interpreter it's like like the fact
615. 1:32:36that I can upload an image and it can process data from the image I this is something I don't know if you know this
616. 1:32:43so so I went to go get data sources and and sometimes you'll get like like I downloaded all my LinkedIn data because
617. 1:32:51I thought it might be fun to like go like process my LinkedIn data and and see what I can do with it so your LinkedIn data it doesn't come
618. 1:32:59down as a single spreadsheet it comes down as a bunch of individual CSV files that are all individual and so I was trying to figure out well
619. 1:33:09how do I get them into a workbook and can I do that automatically and it was all just not fun and then it turns out
620. 1:33:17um you can upload a zip file you can upload a compressed file that has like a hundred different csvs in it and it will
621. 1:33:25write the code to unzip the file it'll read all of the csvs and then talk about them in context so you can upload
622. 1:33:33multiple files to code interpreter by putting it in a zip file and like and and then Ethan Malik did a thing where he uploaded
623. 1:33:44he uploaded a spreadsheet that had that had prompts in in one of the it had prompts in in the first row and then his command was
624. 1:33:58um his command was something like take the prompts in each of the cells in the first row and write the second row and
625. 1:34:07then write the so so he basically uploaded an empty Excel file and code interpreter filled in the Excel
626. 1:34:15file based on his prompt and then gave him a link to download it back so he's he's doing like I I gloo what to do with
627. 1:34:29it seems like it's it's like a Swiss army knife that you can do almost anything now it's only in Python so it
628. 1:34:39there's I'm sure there's libraries that it doesn't have I'm sure it's a little Walled Garden right now but just conceptually what it's doing
629. 1:34:46when they start expanding that out to all the programming languages into all the libraries and figure out where the
630. 1:34:53safety boundaries are within it it's just insane [Music]
631. 1:35:07don't even bother with stocks Aladdin has that coverage there you go somebody knows I don't know [ __ ] about investing
632. 1:35:13so there's I like like here's the other thing about AI you know we're talking about this AI stuff like it's only been
633. 1:35:20around for six months it's been around for like three decades and and like the stock trading um AI stuff's been around for at least a
634. 1:35:29decade probably more than that because those guys are on it you know stock Traders are the ones that like run their
635. 1:35:36own fiber because if they can get something three microseconds faster than the other guy they can make money
636. 1:35:41um so they're on top of this [ __ ] but if you're talking about sort of consumer grade level tools um if if Aladdin's a good one that's
637. 1:35:49great my mom asked me to call her home insurer chat GPT giving a list of questions for the insurance agent oh this is this is
638. 1:35:58one of the things that I'm most excited about is um the ability for all of us for the for the every man the every woman to um
639. 1:36:10[ __ ] with the machine right to be able to call United Healthcare and say well based on the policy here are your rules
640. 1:36:21and here are the things it says you should cover Well normally people don't read that part of the documentation well I did well chat GPT did
641. 1:36:31but you know here's my prepared statement would you like to talk to my lawyer no no well why don't we'll just settle you
642. 1:36:37know what that's covered now we're gonna get to do that soon we're gonna get to [ __ ] with people I like
643. 1:36:44that I got a mosquito bite right on my elbow look at that little [ __ ] what in Florida now they have malaria
644. 1:36:53did you hear this the [ __ ] is happening it's like God damn it how high the Moon
645. 1:37:07let's see foreign how high is the Moon
646. 1:37:23well I think it's obvious 384 400 kilometers or obviously 238 900 miles everybody knows that the measurement is taken from the center
647. 1:37:36of the earth to the moon I'm going to ask um does it change um depending on the orbit um
648. 1:37:51cycle or is it perfect perfectly circular um it's not perfectly circular ha
649. 1:38:08so at perigee which I assume is the narrowest yes 225 000 miles at the apogee 251. there you go see
650. 1:38:26there's chat CPT for you making us all a good scientist I can mansplain even better now [Laughter] it seems like he knows a lot
651. 1:38:39what's that earbud he's not listening to anything with that is he that's gonna be there's a good application just have a little earbud
652. 1:38:49that gives you dating advice you're on like your first date and it's telling you what to say so you don't look like a
653. 1:38:54schlub that would be pretty good what's a good eye to summarize audio files the the open AIS Whispers really good
654. 1:39:04what major challenge do you think AI can help solve in the next five years lots and lots of them um I think it's going to automate out thank
655. 1:39:15you very much for the hearty heart um I think it's going to automate out a lot of the low-level repetitive work that is
656. 1:39:23um kind of Soul crushing for people um I think the scientific advancements are going to be insane I think medical
657. 1:39:31advancements you know advancements in Cancer Treatments genetic diseases neurological diseases brain injuries spinal cord injuries um
658. 1:39:43I think I think the advancements are going to be insane um Sam Walton talks about um I should find this I should find this uh
659. 1:39:52this link and put it somewhere but it was an interview and he says he he thinks within the next five years or I
660. 1:39:59don't know if he said five years maybe it was ten years in the next short amount of time we're probably gonna have a single year where
661. 1:40:07we have as much scientific advancement as the previous 500. and if you want to get a little mind [ __ ] go to chat GPT right now
662. 1:40:17and search for what were the major scientific advancements over the past 500 years and so just look at all those look at the
663. 1:40:26list it gives you have it write you a table and then imagine a single year where the equivalent the modern equivalent of that
664. 1:40:34much advancement happens in a single year that's a very real possibility in not much time from now so probably
665. 1:40:42within five years it's a lot there's probably gonna be advancements in climate science in energy like like um like not needing oil energy
666. 1:40:58um yeah lots so I don't know I've created a community Kyle Amelia's wife is our mayor that's awesome [Laughter]
667. 1:41:09thanks for the answer you're welcome Aladdin is back it is like a little family in here it's cool I like it
668. 1:41:24what's what's nice is you know someone comes in and starts trolling me you guys you know like pounce on them
669. 1:41:29appropriately if they have if they have questions I already answered you let them know if I went to go deal with the
670. 1:41:34dog you're like he'll be back he's changing the shirt so we can read it's good I feel like you got my back it's nice
671. 1:41:41I was I was a bartender in New York um New York City for like a year was it yeah like a year and a half
672. 1:41:48and my first bartending gig was right near One Police Plaza um which is you know the the police station in in Manhattan that all the cop
673. 1:41:57shows are at right and so there were there were two bars like One Police Plaza was here and then there was this Irish Barco I think it
674. 1:42:06was called O'Neill's and then I was at this like low rent neighborhood bar called castillos and so all the police officers went to
675. 1:42:18O'Neill's and like all the lower level cops and administrators came to my bar and so one night I'm bartending and and they were they were
676. 1:42:29wild people they were wild like this one guy was this really short guy I forget his name but he was uh
677. 1:42:37he was like yeah yeah I became a cop so I could kick people's ass I just like kicking people's ass that's
678. 1:42:43why I became a cop so so anyway so one night I'm bartending in like these two like college punks come in and they start [ __ ] with me
679. 1:42:52and the the little short guy the one that said he liked to kick people's ass he was like he was like five four maybe and skinny
680. 1:43:01well a Puerto Rican guy and and he goes he goes hey um hey can I ask you something come you know let's can I ask you something
681. 1:43:11outside and so they they went outside and they were gone for like 15 minutes and he came back it was a little sweaty
682. 1:43:17he's like yeah they're not gonna [ __ ] with you again he drank for free for the for that week so it was pretty good it's good when you
683. 1:43:25got you got your uh you got your uh low-level cops to uh to have your back when some jerks come in um I gave it my menus oh oh it's already
684. 1:43:40here what's going on Tick Tock why do you move me around so there's the audio file who's that Aladdin
685. 1:43:54Aladdin is Blackrock yeah I know no no one's going to outperform Aladdin well I mean I know like like the early days of the World Wide Web
686. 1:44:07there were there was a lot of Shangri-La kind of possibility as well and all that sort of Consolidated like
687. 1:44:13all the open source stuff and all the really cool design stuff that all sort of went away and it all did consolidate
688. 1:44:19so I assume the same thing's gonna happen with this AI but I think if the big companies try to consolidate it too much the the the open
689. 1:44:29source Community I I feel like this is I feel like this is squeezing through the fingers of the control mechanisms but we'll see we'll see
690. 1:44:40uh uh now I'm gonna run her policy through chat EBT thanks for that oh cool Oh that's oh yeah oh that's great yeah
691. 1:44:49exactly yeah yeah take everyone's complicated Health Care policies and you know home and auto insurance policies
692. 1:44:56and yeah have code interpreter you know make you charts and graphs about you know how you could what are the like you could probably ask
693. 1:45:07it things like what are the sneaky ways that they make this process overly complicated that we could simplify and get [ __ ] paid
694. 1:45:19I bet that's in there that's actually something that would be fun to play with evil [ __ ] policy writing people and
695. 1:45:30lawyers that are like well if rather than having them fill out one form we had them fill out 10 forms at
696. 1:45:38six different times over the course of three months and then at each of those steps we try to deny the claim I would think that only Point
697. 1:45:480.625 of the people would actually pursue that right you know there's people thinking like that in those [ __ ] companies
698. 1:45:57so now we can think like that too we can actually read their [ __ ] and and decipher oh that's how the rule works okay
699. 1:46:06let's find another rule that contradicts that and then let's take them to court and watch them settle before we get there that would be good
700. 1:46:15thing they're genetically engineered mosquitoes that can kill off the spread of malaria oh you're being serious yeah
701. 1:46:19and zika yeah that's good they're doing they're doing all sorts of remediation right now down there about I
702. 1:46:27assume it's just lots of chemicals Monsanto is probably happy what if AI determined our government on alive JFK um
703. 1:46:37the conspiracy theorists would have more evidence to say that our government on the live JFK I don't know I feel like we
704. 1:46:45we're living in a in an era right now where um logic doesn't matter anymore truth doesn't matter anymore like we're in this post-truth
705. 1:46:54sort of world that's very very frustrating so I don't [ __ ] know I don't know what would be nice is if AI
706. 1:47:03could snap us the [ __ ] out of whatever pipe dream we're in the middle of right now we could just get back to being
707. 1:47:09human that would be good whatever happened to IBM's Watson it flamed out Watson Health was just a disaster apparently they just relaunched
708. 1:47:18because I so it's just remarkable that like the the amazingly quick adoption of chat GPT just forced everyone out of the woodwork so
709. 1:47:31IBM was really big they did their big blue thing they beat Gary Kasper off at chess that was sort of the first big
710. 1:47:38headline AI can beat humans because they said no one will ever be able to beat a grand master and they did it
711. 1:47:47and then they had this Watson brand was just [ __ ] brilliant and then they said we're going to do Watson health and
712. 1:47:52we're going to cure cancer and we're gonna do all this stuff and then it just kind of flamed out and you know it was
713. 1:47:57it was machine learning that was not this Transformer based stuff that we're playing with today so it was much more
714. 1:48:03specialized and so I don't know why it it didn't succeed but it didn't succeed um and then about four months ago IBM
715. 1:48:12announced Watson GPT I think they called it or or something like that they basically said we got AI too
716. 1:48:21we we did the chest thing and we Watson GPT and then I haven't heard [ __ ] from them since but I did hear from uh a buddy that that knows
717. 1:48:33um some folks at IBM they put together a what do they call them um a center of excellence around prompt
718. 1:48:42writing and their IBM they have 400 000 employees or 280 000 now it was 400 000 when I was working with them um and
719. 1:48:51um apparently they've put a thousand people on prompt writing because they can so I have a feeling we'll hear from IBM
720. 1:49:01but yeah it's it it sort of flamed out well well ain't you a cyber Sore Eyes thank you your show is something else I gotta tell you well
721. 1:49:12I would agree it's something else I hope it's good that was good I want you all to be happy you know what I mean
722. 1:49:23it would be nice if you was happy I don't want you to add a partner I'll tell you what you know uh all right
723. 1:49:37whatever happened to IBM SWAT so we just did well well same for Sore Eyes I am now I am now I am now you catch my drift sir
724. 1:49:47yeah that's your name and it's in your thing all right which bar did the defendants go to I think the defendants got the [ __ ] out
725. 1:50:00of lower Manhattan the defendants were like I'm out of here they got they got in there they got in there uh in the
726. 1:50:06subway and headed to the boroughs or to Jersey or to the Upper East Side [Laughter] um what's something you'd wished you had
727. 1:50:16chat GPT for 20 or 30 years ago ah [ __ ] writing screen place I I went through a two-year period where I wrote
728. 1:50:23seven feature-length screenplays with a writing partner and for me like the hardest part of writing is is the
729. 1:50:34low level [ __ ] like I love the concepting I love outlining while it's still creative and I like I like when the screenplay is either at
730. 1:50:51zero to fifteen percent done or 85 to 100 done because those are really creative things the part of the screenplay that is just
731. 1:51:02for me is just like grinding my face against a brick wall having have having some strong person grind my face against
732. 1:51:11the brick wall is that middle the grind the the you just got to keep writing you feel like it's a piece of [ __ ] you don't
733. 1:51:18know what's gonna have you you're not you don't follow the outline anymore so just what's happening with me with chat gbt right now is
734. 1:51:30the whole idea of creative blocks is a thing of the past like if I don't know where to start or I don't know what to do
735. 1:51:40um I can just with chat EBT and get to 80 like that so I can just stay at a much higher creative level so almost
736. 1:51:47all of my history of of all the writing I've done over the years um I just wish I had it for that I wish
737. 1:51:53it were just part of my life because it's it's it's it's staggeringly powerful to be able to have a concept
738. 1:52:02that historically would not have been brought into the world just because it was it wasn't a good enough idea or it wasn't
739. 1:52:12um it was just too much to deal with to actually pull it out of my head or I didn't have the energy or whatever now I
740. 1:52:17can just go I kind of have a brain fart of an idea and I can go huh let me see if I oh oh there's a oh that's not quite
741. 1:52:28quite the right tone but you know what would be the right tone boom there's the right time oh that's actually pretty good let me copy and
742. 1:52:35paste that and throw that over here so like I'm sort of getting ideas out of my head much more quickly and then there
743. 1:52:41I'm getting them to a higher level of completion much much more quickly what do you got to do you got to pee why are you whining huh
744. 1:52:49you gotta go pee pee no meow little fat dog you little fat dog no yes all right I'll be right back
745. 1:54:52foreign still here 30 people these are amazing all right the dog is happier
746. 1:55:10all right let's get some hearts going see if we can get some more people in here bang on the hearts share the live if you can
747. 1:55:18and I'll do another what time is it it's 10 o'clock I'll do another I don't know 15 minutes half hour or something like that
748. 1:55:24I love watching the art evolve in front of my eyes yeah it's amazing uh what's the big break on your show gonna happen
749. 1:55:33let's get this show on the road buddy Hey where's oh where's my big break on your show gonna happen I don't know man
750. 1:55:40let's get this show on the road I just started a new job and trying to compare the insurance choices is
751. 1:55:51are you so Emilio's wife if you started a new job and you're trying to compare Insurance choices um go play with code interpreter if you
752. 1:56:01haven't because I bet I bet you could grab you let's say let's say
753. 1:56:16you've got three three choices or maybe there's ten you could grab the PDFs of whatever the [ __ ] they're trying to sell you
754. 1:56:26throw them in a folder zip that folder and then upload that to code interpreter and have it compare them for you
755. 1:56:36I bet it does it I bet within an hour you could you could knock some stuff out the silver lawyers are going to be the
756. 1:56:43reasons why it's stalled more I agree with that with the big companies for sure I mean that's why that's why browse
757. 1:56:50with Bing is no longer in chat gbt they know what it means for them lawyer look the lawyers are they're pre-ordering their Yachts now
758. 1:56:59because so many lawsuits are going to be flying in so many directions when you have the level of disruption that's coming
759. 1:57:08everyone's going to be suing everyone you stole my business I mean elon's doing it right now with Twitter right with threads
760. 1:57:15um you know all the Publishers are now saying to to open AI you trained this on three billion documents so you owe us
761. 1:57:22all money okay this is just gonna the lawyers are gonna make a [ __ ] fortune on this stuff um yeah AI does not determine now ai is a
762. 1:57:36probability engine it will determine it doesn't right now
763. 1:58:26thank you what okay all right whatever and I'm back
764. 1:58:43uh-oh lots of loves loves going on a keyboard rephraser app I can talk like Joe Pesci sure Killarney Rose can I solve crimes like
765. 1:58:56who killed JonBenet there's enough well what what AI can do is it can take whatever evidence it could it can
766. 1:59:05analyze [ __ ] really well so if there's evidence to be analyzed sure if there's not it'll make it up I mean that's the
767. 1:59:15problem the hallucinations with this [ __ ] are are a lot although I what what it sounds like with code interpreter is
768. 1:59:23that a it can do math because it's writing python if you watch how it does what it does it's mind-blowing it's
769. 1:59:29writing python so it can actually do math and apparently it hallucinates yet less because it's analyzing a specific set of data
770. 1:59:39all right all right you bunch of pansies let's put an end to those damn creative roadblocks exactly Don Rickles dog ain't crying
771. 1:59:51champ is Hefty not fat what did the doctor call him she she goes the vet we took him to the vet and she goes well
772. 2:00:00he's doing okay she goes he's a chunky boy he's the best when your vet says your dog is chunky [Laughter]
773. 2:00:16he's cute though all right to show up uh felonious the the thing with Don Rickles and Frank Sinatra was hilarious that's a great story
774. 2:00:31um do you think teachers should prohibit the use of Chachi PT to write essays no I don't um go read Ethan Malik's piece called
775. 2:00:45I'm gonna find it uh the homework apocalypse m-o-l-l-i-c-k home work apocalypse yeah the homework
776. 2:00:59apocalypse go read that it basically says the internet made cheating really easy AI makes it it it's done
777. 2:01:13and so it's not that I think the intent is good with teachers saying if chat EPT writes your essay you're
778. 2:01:28cheating and you shouldn't do that so I'm gonna find things that determine that and I'm gonna block it and
779. 2:01:35the intent behind it is right they want the kids to learn but I think how kids learn has to change it has to fundamentally change
780. 2:01:45because education preach at GPT was about giving kids enough information so that they could memorize it
781. 2:01:54assimilate it learn it get the knowledge you now have a tool that can do the Tactical execution of almost anything at some level at some
782. 2:02:06level of competence um so if you're moving into a world of infinite um content creation and infinite sort of
783. 2:02:23tactical execution what's left critical thinking is left so the teachers that I see that are doing this right are bringing this [ __ ] into
784. 2:02:36the classroom they're saying let's have chat GPT write the paper and we're all going to look at the same paper and what you're going to do
785. 2:02:43tonight is you're going to go home and you're gonna you're gonna read how Chachi Pita did it and you're going to come up with ways
786. 2:02:51you would have done it differently and and you're going to find out things that did wrong and you're gonna find out if it used this logical
787. 2:03:00argument or that logical argument or some other logical argument um and we're gonna have discussions and and
788. 2:03:08and you know teach them to think critically so so I think that becomes more and more and more important have I
789. 2:03:15been to Casa Bonita since it's reopening I have not and I need to get there because apparently the food is better the food was so bad
790. 2:03:23at the old one that I was told by six people like it's not even like the kind of bad that's funny it's
791. 2:03:32just like disgusting don't go there so I've never been actually my uh my co-founder my business has been
792. 2:03:41and she said the food she said it was worse than dog food it was it was horrific but apparently it's good now so I want to get there
793. 2:03:48um I want to learn to code an app I looked up API pricing and it's all about tokens I don't understand so basically
794. 2:03:54so when you when you um make an API call so so let's say you're doing an app and you want your app you want someone to be
795. 2:04:03able to say oh what what was what what was it uh I think it was Silver Fox who's got the restaurant who said she took her menus
796. 2:04:12and popped them into chat GPT and kind of came up with new recipes right so let's say you want to do an app like that and you know you scan
797. 2:04:22you scan your current menu and it automatically creates a list of ingredients based on those things and it gives you new things
798. 2:04:32well you're gonna need to do an API call to translate the photo of the menu into something and that's going to cost you two cents and then
799. 2:04:45you're going to need to take the OCR text from the menu and you're going to send that to open AI so so let's say there's
800. 2:04:52a thousand let's let's say there's 750 words on that menu and so you you have a prompt that says take this menu
801. 2:05:01there's the menu and you send that to open Ai and then it sends you back a list of ingredients let's say between
802. 2:05:07sending the menu plus your prompt and getting the response that's 1500 tokens so your API will cost that the API call to do that will cost
803. 2:05:20whatever the tokens times a thousand is so it's like two cents or six cents or whichever model you're using
804. 2:05:26so if you did 1500 instead of two cents it would be three cents right um and so so it's basically just
805. 2:05:34quantity of [ __ ] sent and quantity of [ __ ] sent back and then they they add that all up and it's it's very very
806. 2:05:42reasonable if you're doing a small amount of thing like if you've got an app that only has a thousand users and
807. 2:05:49they're only doing you know on average 500 tokens a day it's not going to cost you very much if you have a hundred
808. 2:05:59million users and they're doing you know 10 000 tokens a day your bill is going to be big but it but it's it's quite
809. 2:06:06cheap if you're doing small prototyping stuff so like I like I've got I haven't been playing with it a ton lately but I I I
810. 2:06:16haven't had a bill more than like 10 or 10 or 12 dollars on any month that I've been playing with the API
811. 2:06:24stuff I don't but I don't call it a lot but it's it's pretty cheap I know you're the guy to ask sir can we collaborate where can I write
812. 2:06:35to you if you go to my um profile my there's a link Tree in my profile that's got my contact info it's also got a a survey for
813. 2:06:48this thing I'm putting together a workshop so some folks on here have have said hey why don't you put together a workshop
814. 2:06:58so I made a graphic for a workshop that counts right um but if you go to linktree slash real Kyle Shannon that's the link tree the
815. 2:07:10first link there is the um survey to just there's only three questions there I just want to find out what what do people want to uh
816. 2:07:19what do they want to learn if I did a workshop and then I'll I'll put something together so anyway my contact info is there I am now
817. 2:07:28[Laughter] what am I doing oh yeah um I am now you can DM Kyle it's true
818. 2:07:43I actually look at that [ __ ] because you know why not can AI build me a YouTube channel it can write you all the scripts
819. 2:07:54and then you can use other tools to automate all that [ __ ] you can use canvas canvas got their bulk create now
820. 2:07:59so you can bulk create and canvas have chat EPT write all the scripts yes oh here's the thing it's just we're we're moving into an era
821. 2:08:15where everything's going to be automated everything and then and then what gets really interesting I think is what becomes
822. 2:08:31valuable in in a in a world where you can automate everything from creative to music to writing to project management to just everything
823. 2:08:40what becomes valuable and my gut is telling me right now authenticity I think authenticity makes a comeback live music like in-person things
824. 2:08:54um real people telling real stories may rise above the noise I think that but you know it's going to take two to
825. 2:09:03three years for all that stuff to start to happen but that's where we're headed you have to know your thing otherwise
826. 2:09:12it's almost useless for for instance encoding yeah I agree that's what I say about the about this AI [ __ ] is
827. 2:09:20how to get involved if you're new with this stuff if you're trying to figure out how do I how do I get my head around this stuff what
828. 2:09:28there's kind of three phases there's the there's the dick around with it phase right the the parlor tricks phase write
829. 2:09:37me a song write me a poem write me a marketing strategy he's [ __ ] around with it kind of generally getting the capability then
830. 2:09:46they're sort of looking at well wait what do I do what's the thing you're good at and I think your point there is
831. 2:09:52really smart like if you're really good at engineering and coding you're going to be able to do things
832. 2:09:58with the engineering and coding this can do that are at a whole different level than what all the rest of us can do
833. 2:10:06um so start to understand how to use these tools to do what you do like like basically disrupt yourself figure out
834. 2:10:12how you can do what you do better and then I think the next phase after that is then start to look more broadly
835. 2:10:20and more horizontally and say well [ __ ] it took me 30 years to get to this level of being an engineer I don't have to
836. 2:10:29take another 30 years to get to the level of being a good illustrator because now I have mid-journey so if I
837. 2:10:34could do this level level of coding and this level of image Creation with mid-journing I can maybe put together a project I
838. 2:10:42never would have even considered before and then if you do that with five or six different disciplines or ten
839. 2:10:48all of a sudden you start to become a super being right so so these things become Humanity amplifiers right
840. 2:10:58and and some people are going to be really good at this the ones that are good at critical thinking and thinking
841. 2:11:02horizontally are going to be [ __ ] rock stars makes unbelievable errors it does you've got to be vigilant with
842. 2:11:14the errors although code interpreter I think is going to help thin that out if you're not aware you're lost agree
843. 2:11:26um yes I can build you a YouTube channel so how do I get an app created is there a prompt for that well so it depends
844. 2:11:39what you mean by an app if you want an app like that's going to show up on your iPhone then you need to do stuff with
845. 2:11:45xcode and Swift and things like that and and chat CPT can help you write all that but you still need to get it in
846. 2:11:52Frameworks it can teach you all that um if you don't feel like learning all that [ __ ] you know down at the at the Lego block level
847. 2:12:02um use a no code tool like zapier um which now has a a new feature called interfaces where you can create a
848. 2:12:09website and Bots and things like that or use there's another no code platform called Bubble which you can do you can
849. 2:12:17create mobile mobile apps with bubble and I think it handles all the all the back end [ __ ] and submitting it to app stores and Google Play
850. 2:12:27um or go find an engineer who knows how to do app development and ideally find one that's doing app development using these tools
851. 2:12:38where they can knock that [ __ ] out really really quick so I wouldn't put that all on yourself necessarily
852. 2:12:43you know find a partner or a you know a young person that's hungry like finding a young person that's hungry
853. 2:12:51right now is kind of hard because they're all a bit cynical and they're like I like working from home and I
854. 2:12:55don't really this whole thing where you entrepreneurs with your expectation of 40 full hours of work so if you can find not that
855. 2:13:07but someone who's actually excited about this stuff and thinking about an entrepreneur entrepreneurially
856. 2:13:13um you can probably get that app done very very quickly but go look at bubble I mean bubble and zapier
857. 2:13:28the no code tools are they are more simple than coding but they're still complicated so you you need to know your level of sort of
858. 2:13:39geeky geekiness tolerance right if you kind of like geeking out on stuff you'll probably enjoy them and you'll be
859. 2:13:47success like it's not hard to be successful with them you just need to be a good Problem Solver because you still
860. 2:13:53have to do a lot of problem solving even though they're no code yeah the problem with no code is there's still a learning
861. 2:13:59curve that's exactly right like zapier's really got all sorts of quirks and and weird things and yeah that the other
862. 2:14:06point is they're quite limited that's the the second problem with no code is you'll you'll you'll figure out the no
863. 2:14:14code stuff you'll get to 90 of what you want and you're like I just need this 10 that's the 10 that the no code tool
864. 2:14:21doesn't do it's like oh that yeah that you're gonna actually have to write some Code for ah [ __ ] really that's totally my experience with
865. 2:14:32totally my experience uh yeah zapier is more like a tool for connecting other tools yeah it is zappy so but I think bubble is is more of a
866. 2:14:42app development framework um but anyway um listen it is 10 18 here in Denver Colorado looks like I've gotten to the
867. 2:14:51the bottom of the list here got to the bottom of the barrel really nice hanging out with you all tonight I appreciate
868. 2:14:57your time as always have a fantastic evening peace out um I will oh Silver Fox there's also remote Engineers on Fiverr yeah that
869. 2:15:08could that could be interesting so peace out everybody thanks from Bucharest awesome bottom of the barrel I
870. 2:15:15meant bottom of the question Barrel not that you human beings are the bottom of the barrel how dare you Emilio's wife
871. 2:15:21you know better than that peace out everyone