aiLL 6/30/23 - Talking with ChatGPT About How it Works

Transcript

1. 0:18welcome welcome welcome welcome to the AI learning lab it's Friday night my name is Kyle Shannon thanks for
2. 0:26joining um we'll wait for some other folks to get in here while we're waiting for people to pop in just to give you a
3. 0:32little bit about my background I'm a a longtime entrepreneur I've started 12 or 15 companies in my day and uh in many many
4. 0:42years my current gig I'm the founder and CEO of a company called Story vine which is an automated video storytelling
5. 0:49platform um I'm relatively new to the AI space probably the last year or so I've been deep diving into it and have been
6. 0:57absolutely obsessed uh since chat GPT um had a rather large Epiphany that um paralleled The Epiphany I had uh when I
7. 1:08first discovered the worldwide web in 1994 and I started one of the first digital agencies back then and um ended
8. 1:16up riding that wave up so I think I think this generative AI stuff is is of a very similar uh it's there's a lot of
9. 1:24parallels to the early days of the web and the early days of this stuff although this stuff's more profound um
10. 1:29if you've been here before you know these URLs behind my head if you haven't been here before so here's what here's
11. 1:34what's going on here um I'm not affiliated with any of these companies but but it's it is my strong belief that
12. 1:41if you haven't played with chat GPT you owe it to yourself to go there and play until you have what I'm calling the
13. 1:49Kevin McAllister moment from home alone when you figure out what this thing can actually do um it's pretty amazing and then bing.com is just
14. 1:59Microsoft version of it so if you go to bing.com and then click on the chat button that's also chat GPT and po.com
15. 2:07has uh six different models you can play with and one of the models in particular at po.com is good enough that I think it's worth
16. 2:15paying for hang on I think my dog is whining he likes to hang out with me when I do lives hang on is that a
17. 2:24champy huh is that champy my door yeah he wants to see you champy where'd he go come
18. 2:38on lay down lay down we'll go live I'm already live buddy all right yes the dog wanted to hang out okay welcome oh
19. 2:47there's 161 people here welcome everybody uh my name is Kyle shanon this is the AI learning lab uh I was in the
20. 2:53middle of talking about what the hell this channel is about and the dog was like let me in dad so the dog's here now Champ's here
21. 3:00he's a fat tub of a dog he's a little Thug and um okay why did you use po to sum up the terms of service versus chat
22. 3:10gbt is the model more suited to it so I'll get back to I'll actually show you that I probably could have used chat GPT
23. 3:18with a with one of the link readers but I just it was just an experiment I did in post so let me actually go show you
24. 3:23what I did and then I'll talk to you about why I did that by the way my name is Kyle Shannon this is the AI learning
25. 3:29lab these URLs behind my head the top one is chat GPT if you haven't played with chat GPT go play with it um play
26. 3:36with it while I ramble on here bing.com is is another one worth playing for that's gp4 connected to the internet uh
27. 3:44not quite as good as the um official chat GPT but still it is gp4 and it is connected to the internet and it's free
28. 3:52and then po.com I'm going to show you right now if you're brand new to all this GPT nonsense this this document
29. 3:58here prompt. chat will teach you about prompting and how to use chat GPT and give you lots and lots of ideas of the
30. 4:05kinds of things it makes possible and future pedia.com
31. 4:20which I thought oh I didn't flip my camera doesn't seem like he knows how to use the Tik Tock Sally all right there we go okay so this
32. 4:31is chat GPT here if you've not seen it before this is what it looks like fancy huh it's actually quite simple um okay
33. 4:43so I'm going to go to po.com so po.com is similar to chat GPT it looks the same and if I if I flip the camera over here
34. 4:50to the left couple of things going on here at po so you've got explore and create bots so so one of the things you
35. 4:59can do do it pose you can make your own bot which is really fun and I can I can show you how to do that or you can go
36. 5:04explore Bots that other people have made there are a ton of them there in fact if I click explore we can just go here
37. 5:11there's games writing Bots programming Bots history Bots music Bots just tons and tons of them the the the top one
38. 5:17right now is a mid Journey bot where you give it a shitty mid Journey prompt and it'll write you a good one and there's
39. 5:23there's a bunch of them in there um if you go find one called I am Shakespeare that's one that I wrote that responds is
40. 5:29Shakespeare in I Amic pentameter so uh so anyway okay back to back to PO here so the first six options Sage GPT 4
41. 5:40Claude plus Claude instant 100K Claud instant and chat GPT those are all just uh basically like chat GPT variations
42. 5:48right two of them are chat GPT Claude is anthropics version and then Sage is POS version this one here Claude instant
43. 5:58100K what that 100K means is that means 100,000 tokens and that translates to about 75,000 words so you can copy and
44. 6:07paste essentially like a full novel into your prompt and and start interacting with it which is absolutely remarkable
45. 6:15so so I was I was uh messing around with Wonder Dynamics which is a super cool 3D um animation tool where you basically
46. 6:26upload video and it replaces your body with a robot it's it's actually quite amazing but they had this really long
47. 6:32terms of service and you know they're they're well formatted but like they just go on and on and on and on and on
48. 6:41and what hit me was I've never um ever read terms of service I've agreed to them all I've never ever read them
49. 6:51because I don't feel like parsing all this crap I have ad and I'm not going to deal with it so what hit me was I can
50. 6:57probably go over to PO and and go into Claud instant and just paste those terms of service in So based on these
51. 7:11terms and then I hit shift return and I'm going to just paste in all of those terms of service now if you pasted all
52. 7:18these into chat GPT it would break because it's way more than 3,000 words but I've got a 75,000 token window here
53. 7:25here context window so I can paste these in so I'm going to say um please tell me what I am a agreeing to in tell me
54. 7:42the most important things I am agreeing to in very short easy to understand bullet
55. 8:00points and I'm going to hit return and it's going to go read all of those terms of service which is if you've never
56. 8:07played with this crap it's just it still amazes me every time like it took all that in and I'm actually scrolling to
57. 8:13the bottom here are the most important things uh Wonder owns the services and all content on the services you only get
58. 8:21limited license to the services you grant Wonder a license to use your user content for the purpose of providing the
59. 8:28services to you so if you make something here they can use your content you represent that you own or have the
60. 8:34rights to your own content right so it's putting in very simple English what that big long thing was now I could also
61. 8:43probably do the same thing over at chat GPT so I'm going to copy so I'm back at the Wonder terms of service so I'm going
62. 8:52to copy the URL here and I'm going to go over to chat GPT and we're going to go to GPT T um plugins and I'm going to turn off I've
63. 9:03got some food ones turned on I I'll I can talk more about plugins in a bit I'm going to turn off the food
64. 9:09ones and I'm going to turn on a plugin called Web pilot and web pilot will go read a web page so I'm going to say based on these
65. 9:23terms and then I'm going to put in the URL for the terms of service I'm going to hit uh shift return to give me two line breaks
66. 9:36please give me a give me a um a a list of the most important things I am agreeing to
67. 9:57in um bullet points that are 6 to 10 words
68. 10:11long and now that should fire off that plugin right here yep there's web pilot so it's going off and looking at that
69. 10:23um at that thing so here are the the bullet points which are not in oh you know what I didn't put I didn't put easy to
70. 10:36understand acceptance of agreement policy in okay I'm going to say um only include the most
71. 10:50important ones and make them easier to understand and for a non-legal person and let's go see what it does you
72. 11:06accept their rules if they change the rules tough [ __ ] if there's a disagreement it's resolved through
73. 11:12arbitration not in court so this gave me a big long list so anyway um yeah so you can do it you can do it multiple places I just
74. 11:22chose po because the uh the ability to just copy and paste in a a giant document is just fun to do um but yeah but with
75. 11:31using web pilot or some plugin that can read a website or a PDF you can do the same thing there so good question
76. 11:38welcome everybody um my name is Kyle Shannon this is the AI learning lab it's Friday night we're hanging out if you've
77. 11:45got questions about AI pop them in the comments below and I will do my best to answer them and let me get my my groovy
78. 11:55background going there's my groovy background with my groovy URL so all [Music] right I love learning recipes on chat
79. 12:10gbt yeah it's good at that like it's read all of the recipes on all of the internet so it's probably pretty pretty
80. 12:16good at it and also if you if you have access to plugins um use the instacart plugin and um it will order your ingredients
81. 12:27for you so you can just come up with recipes and then have the ingredients show up on your doorstep I saw someone
82. 12:33else did something with recipes where they used the Woolf from alpha which is a mathematics thing but it'll determine
83. 12:40calories of your meals so you can you can actually have it generate calories as well the 2021 cutoff date is annoying
84. 12:47though yeah if you use if you use the internet connected version of gp4 it's less annoying but it that is annoying
85. 12:54what's the 3D robot program called Wonder Dynamics Wonder Dynamics here I'll I'll go show you what these things do it's pretty [Music]
86. 13:11amazing do I have it there it is so I think I have one here that I did but so so basically what happens is you
87. 13:21upload a video of like a person walking and then you choose one of these robots and it will replace the person in your
88. 13:30video with the robot so I think are these examples yeah so like you know see the robot walking so that that's just a
89. 13:39video of a person walking and this Wonder Dynamics swapped it out I did want it didn't come out super great and
90. 13:46it takes a while it takes like half an hour to process but if you know anything about Motion Graphics and animation and
91. 13:53things like that this is pretty amazing like that little pig robot like there it sort of flashed away but there's another
92. 14:04dude so so I only did a 10 10-second video but yeah pretty amazing Wonder Dynamics crazy it's crazy I tell you
93. 14:18crazy do me a favor if you would double tap the screen to to get some hearts go and see if we can get some other people
94. 14:24in here and if you wouldn't mind share share this live with your folks and say hey come hang [Music] out [Music] D here we come to save the
95. 14:41[Music] day spelling tell wrong yeah here's the deal about typos and chat GPT it doesn't give a [ __ ] i' I've Stu when I'm in chat
96. 14:53GPT I just type right through typos it doesn't care it understands everything you could I I am convinced if you got
97. 15:02each word 60% right it would understand what the hell you meant it's amazing so typos I I no longer care it fixes them
98. 15:12it just fixes them it fixes capitalization are are plugins free add-ons no so so for chat GPT to get GPT 4 you need you need to
99. 15:25pay the 20 bucks a month for GPT plus that gives you access to gp4 gp4 connected to the internet and plugins and when they released this new
100. 15:35thing called code interpreter you'll also get access to that and when they upgrade gp4 to be multimodal and include
101. 15:42images you'll get access to that so the reason to pay it and and I'm not affiliated with it in any way like I
102. 15:49talk about it so much they should [ __ ] pay me but I'm not so I don't care if you do it or not um I think it's
103. 15:55worth it just for Early Access to their Tech and I think the GP pt4 implementation at open AI is better than
104. 16:01anything else Bing if you click on the chat button if you go to bing.com and click on the chat button that's gp4 for
105. 16:10free and it's connected to the internet and it will support plugins in the future it just doesn't right now but you
106. 16:18can generate images at bing.com you can just say make me an image of a rooster on a I don't know roof or wherever in a
107. 16:26field and it will do that um and it's connected to the internet and it's free and then po gives you access to all
108. 16:32those different models that I talked about before so uh so it's not free but it will be uh the plugins will be free
109. 16:40at bing.com they're just not there yet I'm in it and I'm not retiring for another 10 to 15 years so I'm soaking up
110. 16:47as much of this as I can yeah I I you know I think I think in it right now um AI Ops is going to be a huge thing like
111. 16:57which services do you use which ones are secure you know are you doing Standalone versions of chat GPT that you're
112. 17:04spinning up on Azure or whatever it might be um you know increasingly companies are going to want to GPT all their data right so being in
113. 17:16it right now and getting your head around this [ __ ] I think it's a brilliant move really really very smart because at some point
114. 17:25someone's going to go who knows about this [ __ ] and if you can raise your hand even a little uh it's going to serve you
115. 17:32well but I think AI operations is going to be a it it there's going to be a there's going to be a lot a lot lot lot Mr fribble is also in
116. 17:41it I want to get into Tech now because of this um and so so what's what's fun about where we are right now with with
117. 17:50AI is depending on like you know how geeky you are right so if you've got your little geek meter if you're pinning
118. 17:56all the way over on you know can't talk to people at parties and I'm I'm an Uber geek you can get into the
119. 18:03machine learning and the model training side like there's all sorts of projects you can play with that are that are more
120. 18:09technical things like Lang chain and and autog GPT the autonomous agents they all require a fair amount of technical chops
121. 18:18to play with right now um and and you know companies like open Ai and anthropic and um one of the founders of
122. 18:27Deep Mind he just started a new company that got funded for $1.3 billion today or yesterday so so AI companies are hot
123. 18:40right now um so if you're on the geeky side you can do that but what's also cool is because of chat GPT and because
124. 18:47of the fact that 100 million people started using chat GPT in this in the first six weeks it had hit 100 million
125. 18:54people in six weeks because of that there's this inflection point where you don't even have to be a geek if you're
126. 19:00just AI literate in how to use chat GPT for business that's going to serve you well as well so um so yeah so you you
127. 19:09can be anywhere from an Uber geek all the way to someone who you know is trying to figure out how do I do
128. 19:14marketing plans with this thing or how do I do project planning or whatever it might be so yeah it's really interesting
129. 19:21times it's free to share on Twitter I'm not sure what that means oh cool I need plus then definitely yep some sites are
130. 19:29Banning copy and paste because of AI plagiarism any way around it some sites are Banning copying and P oh copy and paste because of AI
131. 19:44plagiarism I don't even know what that means I guess be I I guess I guess if if if I've written an article they don't
132. 19:54want someone coming in and copying and pasting that article into chat GPT rewriting it and getting traffic for it I get
133. 20:03it I listen I don't know I think it's incredibly shortsighted I don't think that'll last long are they also blocking
134. 20:11any plugins from accessing the site and can they they probably can but I don't know I feel like that shortsighted all
135. 20:20this stuff I all of the efforts right now of like AI detectors and you know AI is plagiarism and all that sort of stuff
136. 20:30are all very shortsighted for one thing it doesn't acknowledge how these tools actually work um and and and also these tools are
137. 20:40going to be built into everything we use Microsoft is is already rolling them into all of Office 365 Dynamics Azure um
138. 20:49everything Windows um Apple will likely have them everywhere Google's going to have them everywhere like these tools
139. 20:55are going to be ubiquitous right so it's not going to be something you can stop so I I think a lot of this stuff is very
140. 21:02shortsighted um have I played with auto GPT or baby GPT I've played with them a little bit all of the autonomous agents
141. 21:09right now kind of suck ass because they're just it's it's it's new and the the underlying tools that they're kind
142. 21:17of gluing together to make them are a little janky uh for those of that you don't you that don't know um Auto GPT
143. 21:24and baby uh what is it baby GPT I think it's baby AGI um are are what's called autonomous agents so instead of like you
144. 21:33going to chat GPT and typing in a prompt you go to one of these autonomous agents and you kind of give it a goal so the
145. 21:40goal like if my my company's name is storyvine so I might say go find me the marketing taglines of all my
146. 21:47competitors um and their annual revenue for the past three years and it will start prompting itself it will give
147. 21:55itself a set of three tasks and then go start executing those like it'll go fire off a Google Search and say who are the
148. 22:02competitors to storyvine and then it'll go find a site that's got revenue of you know companies like that and then it
149. 22:09will pull that in and they'll give it s a new set of tasks um they're conceptually insanely cool um I think
150. 22:17autonomous agents are the future I think we're absolutely going there it's just that the current ones kind of suck
151. 22:24there's a new one that came out last week called um GPT engineer which is super looks promising they it's basically like Auto
152. 22:37GPT but they've kind of limited it to programming and I think because it's kind of a smaller goal set it's probably
153. 22:44easier to kind of contain it and get it to to be a bit more predictable and then I played with one 3 days ago called GPT
154. 22:52author which writes um a a complete fiction novel it's [ __ ] insane like you just give it kind of a topic and a uh it's
155. 23:06just crazy you give it a topic in a writing style and the number of chapters that's all you give it it comes up with
156. 23:12a title it writes the outline writes each chapter and then uses stable diffusion to create a um a cover image for the
157. 23:22book and the books weren't bad I mean you know not something I would buy necess necessarily but uh but it writes
158. 23:30a book um so are you basically saying Bing is using a premium version of chat GPT yeah Bing is using GPT 4 as is if
159. 23:40you go to KH Academy there's this GPT teaching assistant called conmigo that's also gp4 that's been very highly tuned
160. 23:48in that case for Education uh but bing.com is gp4 so so it's free gp4 now if you you use gp4 here and you compare
161. 24:01it to gp4 there they're not the same why cuz they they have different sort of uh safety guard rails up basically um
162. 24:11Microsoft got their ass handed to them when when they first launched Bing a reporter from The New York Times
163. 24:18interacted with it and and it didn't have as stringent guard rails and it kind of lost its mind with this New York
164. 24:24Times Reporter and told him it it was in love with it told him its name was Sydney and it was in love with him and
165. 24:31he should leave his wife for Sydney that made the cover of the New York Times so they kind of lobotomized
166. 24:39uh gp4 at Bing for a while it's getting better but it's still not as good as the one up there but if you don't want to
167. 24:46pay a monthly fee it's free and it's it's really good and it's connected to the internet so yeah that's exactly what
168. 24:52it is um what do I think about buying a AI domain names right now I bought um I bought storyvine doai which is my company's name um as an
169. 25:07investment I don't [Applause] know I have no idea about the domain Market um anymore because there's so there's so many of these top level
170. 25:23domains now Biz and this and do that and and a and nft and do you know eth I bought I bought a couple of eths I bought my name eth
171. 25:40um I the the problem is right now is a lot of these AI companies are are just using so many of these different domains io.
172. 25:52axyz do this. that that it's really hard to find sites right now and it's it's really easy to end up at a at a site
173. 26:00that's not the official site which is terrifying um I don't know if there'll be a good a good investment I think that
174. 26:08all this AI stuff is going to be ubiquitous it's just going to be everywhere um so so I think I think every site's going to have ai rolled
175. 26:16into it in some way so I don't know that that saying this is going to be the AI version of you know xerox.com
176. 26:25I don't know I don't know I I I just I don't know enough about Don the the the domain name buying and selling economy
177. 26:35to know if it's worth it anymore um I wish I'd bought a bunch of docs back when like you know when I started agency.com agency.com was
178. 26:46available like everything was available it was it was Insanity um so um I there there was a buddy not a buddy there was
179. 26:56a colleague that that worked for time ink we worked in the time ink building the early days at agency.com and one of
180. 27:03the guys who worked for for this thing called Pathfinder which was the time ink um like service for for internet stuff
181. 27:12um he bought all the domains that were typos of the most famous domains that that were driving traffic and then he
182. 27:21just redirected from the typo to the real site and it paid him like whatever a nickel every time someone got
183. 27:28redirected to the real site he was he was bringing in like 50 Grand a week cuz he had like hundreds of them it was insane it was
184. 27:38amazing there you know there there was there were some things I should have done a lot more of back then and didn't
185. 27:45I was too busy building a company God damn it all right AI related crypto blockchain technology is a is huge
186. 27:53upside I think that AI is going to drive adoption of blockchain and nft technology I I agree with that um but
187. 28:01you oh you're talking about AI related crypto and blockchain Tech um yeah there's going to be that as well like
188. 28:09like Sam alman's World coin is is an interesting one to look at um but but I also think that we're we're going to
189. 28:16need um some sort of tokenized content verification and human verification uh infrastructure after the
190. 28:242024 election I think it's going to be quite clear that um having infinitely powerful um writing and image and video generation tools
191. 28:37um are causing some issues so I I think the ability to say you know this piece of video was created by this verified
192. 28:46human being and it was verified that it was created on you know that site and it's verified that it hasn't been
193. 28:53modified that kind of [ __ ] we're going to need soon so I think I think blockchain could be a big part of that
194. 29:00can my folks use open AI for Apple TV yeah if they've got a if they've got a browser if if Apple TV you can do a
195. 29:08browser right so if they've got a browser in Apple TV they just point it there and like on my phone I don't even
196. 29:14use the the chat GPT app there's an official app but there's a bunch of scam apps I just point chat my my phone
197. 29:22browser to that and then make it a um a shortcut on my desktop um so yeah they probably can assuming
198. 29:30that I I I use Apple TV but I've never I I I don't remember if you can do a browser there but I assume you can it's
199. 29:36just basically a website it's it's like a big web engine um has Auto GPT actually worked for anyone I got Auto GPT when I first tried
200. 29:47it it within like 30 seconds I typed in that thing go find the taglines of all all the competitors of story vine.com it did
201. 30:01like six different operations and returned me a list of competitors taglines in like 30 seconds and then I
202. 30:08couldn't get it to work again for like 3 weeks and I said [ __ ] it so um so a I think it got overloaded but but B it's also just it's
203. 30:19just it's the components aren't sophisticated enough right now like the the quality of the prompting that that
204. 30:27does what it does it's just not there if you go look at um GPT author and and it sounds like GPT engineer both of those
205. 30:37projects GPT author and GPT engineer I know I know GPT author because I've I've I've written two books with it
206. 30:46um it's really easy to understand and it's a very clear concise set of operations that it does it's not it's
207. 30:53not sort of free wielding the problem with auto GPT is you can tell to do anything and then it kind of goes off
208. 30:59and tries to do [ __ ] and it runs into a pay wall here and a login there and and you know it just gets wrong information
209. 31:08there and it just it gets caught in these loops and it just you know it's I don't know it's just a mess right now I
210. 31:14don't think it has enough framework around it right now to make it successful but you know what I would say
211. 31:20is if you're geeky get involved in projects like that get involved with Lang chain projects get involved with
212. 31:27auto GPT and baby AGI and GPT for all there's a bunch of projects out there that what they're up to is really really
213. 31:35interesting where they are in the development is like zero you know. version 01 you know they're they're very
214. 31:44early um so you could sell it and make money I assume that's about the the a domains you can potentially sell them
215. 31:51but I don't know not a fan of being I I I like it concep but every time I go to use it I'm like this is shitty and I go back to
216. 32:02open.com decentralized AI far preferable to corporate authoritarianism yeah and I think there'll be that there I mean the the
217. 32:11good news about chat GPT getting so popular so fast is that meta released their large language model the the including the weights the for
218. 32:24the core model so the open- source community with large language models right now is on fire right so there's
219. 32:31going to be all sorts of projects decentralized projects and installable projects probably within a year or two
220. 32:38you'll be able to install gp4 or the equivalent on like your phone without internet connection like you'll have
221. 32:46full gp4 level capabilities on any device you have um including probably raspberry pies and [ __ ] like that like
222. 32:55the the the work that the open source communities doing in making these things more efficient and smaller and there's a
223. 33:01talk with the CEO of uh stability AI yesterday that I watched and he he was saying the same thing he's like these
224. 33:10things are going to be incredibly powerful incredibly fast and Incredibly small and just they'll be everywhere so
225. 33:17yeah so I don't disagree with you but at the same time it's going to be interesting to see what Google does and
226. 33:24what open AI does and what this new company Pi does I think it's called inflection or something like that this
227. 33:30was the the exf founder of uh of Deep Mind his startup just got funded $1.3 billion it's going to be interesting to
228. 33:39see what he does with $1.3 billion he's creating agents right personal agents Pi um Pi you're your little Pi agent um so
229. 33:49it's going to be interesting to see what they do that answers my AI domain question I was on the fence thanks great yeah I
230. 33:57like I wish I had a more definitive answer it's good or not I mean if if you've here's the good news about AI
231. 34:04domains well okay bad news they're expensive they're like 100 or 200 bucks they're not cheap at all but if you've
232. 34:11got an idea for a business there's actually a because they're expensive there's a decent chance you can get your
233. 34:17business name a probably not a bad call um I I wouldn't buy tons of them though unless if you got a big trust
234. 34:25fund hanging out you want to go spend some money on AI domains go for [Laughter] it all right I actually support my
235. 34:34company's tag management system and web analytics tools oh that's interesting yeah and like all that stuff being able
236. 34:41to autot tag things and and take unstructured data and get it into structured data if if you haven't been
237. 34:49playing with the function calling with the with the new GPT apis with the function calling you can do things like
238. 34:58you know grab a pile of unstructured data and have it output Json that you can pull right into your tag Management
239. 35:05systems that could be some interesting stuff to do since I'm guessing you know what those are what do I think AI is
240. 35:11going to affect that space well I just talked a little bit about it um yeah so so go learn about code interpreter go learn about code
241. 35:21interpreter man it's like it's like I want to get my hands on that thing so I can see if it's really as as it looks
242. 35:27but every video that I see on YouTube of code interpreter where you've got some data analyst looking at what that thing
243. 35:34does is just their mind goes it's like and this stuff's crazy enough as it is but I think that's going to be crazy crazier
244. 35:44um yeah I mean these GPT tools are are really fantastic at um at analyzing you know um things where you would normally have to write
245. 36:00special you know SQL calls and you know sequel searches and and specific code to be able to pull things out and python
246. 36:09code to do this and do that um I think the place that I would probably explore if I were you is using large language
247. 36:19models as a kind of a custom report generating tool like like like an like an easy to understand front end for all
248. 36:27the data that you're managing right so if you've got if if the SE suite at your agency is asking you you know for
249. 36:35interesting questions or Trends or things like that really start experimenting with gping all of your data so that you can very quickly um
250. 36:46pull up custom reports and and things like that I I think that's going to be there's going to be superpowers there
251. 36:53the accessibility will it clear out any small fries the accessibility will it clear out any small fries I don't quite know what you
252. 37:02mean the the accessibility of the large language models like the open source large language models I think that will
253. 37:07Empower small fries I don't quite know what your question is if you can if you can re rewrite that Dash around the
254. 37:14world good name I'm a fan of the big good names um will it clear out any small fries I don't know what that means um I can get
255. 37:24funding for that what can you get funding for tell me what can get funding for I've got I've got a bunch of ideas
256. 37:30right now that I'm like Jesus this is fundable the fact that the the I mean listen he was the founder of Deep Mind
257. 37:37which you know deep mind was acquired by Google and they they they cre created Alpha fold which is the big protein you
258. 37:45know the protein predicting thing so so he's got some pedigree but 1.3 billion dollars for a [ __ ] startup like
259. 37:53that's the seed round it's like the [ __ ] I mean it's just insane ah the internet gets disrupted and telecommun Telecommunications has
260. 38:07crashed can we use modified shortwave well I I mean I mean here's the
261. 38:21thing if if there's um let let's say that the the CEO of of stability AI is right and and and what it looks like is happening with these op
262. 38:34Source large language models let's say within a year to 18 months um they've gotten these large language
263. 38:43models down to like a a small install like a 100 megabyte install and it's the quality of gp4 it's got the knowledge of
264. 38:52G gp4 and the quality of gp4 and you can put it on a thumb drive or on on any phone not connected to the internet if
265. 39:02all of that [ __ ] gets you know gets interrupted and you've got essentially you know all of the internet queriable
266. 39:12on a chip on your device on your Raspberry Pi on your laptop then yeah I think you could process stuff locally shortwave whatever you
267. 39:24know short shortwave the output so yeah I think there's a very real possibility that yeah you could you could do that
268. 39:35um oh you could sell the domains later no well you can sell them later is there a market for them I don't know so are
269. 39:42you saying that when GPT author helps write the books you can sell them well GPT author probably so because it's just
270. 39:49an open- Source it's just an open- Source set of scripts basically um yeah yeah probably I don't know I mean who's going to sue you not
271. 40:01the open source project um I I mean the the thing about so so the thing about all the copyright issues going on right now just let's assume
272. 40:13that the next 20 years are going to be nothing but lawsuits both directions right I should be able to copyright this
273. 40:20you stole that like that's going to happen for the next 20 years the the IP attorneys are going to get loaded more loaded
274. 40:30um there are lawsuits like open AI was just served a lawsuit two days ago that they trained their model on published
275. 40:40books and so either they're going to settle or it's going to go to court and if it goes to court at some point one of
276. 40:47these companies you know might get some sort of ruling that says hey if you wrote a book with this particular model
277. 40:54from this particular company then you're going to have to pay royalties into this pot I can't imagine that happening at at
278. 41:03any kind of scale that would affect an individual anytime soon so yeah go write a pile of books and stick them up on Amazon eBooks why
279. 41:15not I I mean listen if nothing else do it to learn it like like if nothing else just do it to to to see what's possible
280. 41:24like that's when when I saw the GPT author thing I'm like okay let me go try this it's it's a a Google collab
281. 41:31notebook so you got to know a little bit about Google collab and how to run a collab notebook but that's not
282. 41:37particularly you know you don't have to be super technical you have to be a little geeky and you have to be a little
283. 41:42obsessed but chat GPT can tell you how to [ __ ] do it so go you know run your Google collab and and watch it write your book and then
284. 41:53read what it gave you and then look at the prompts what's cool about it it's open source the notebook's open you can
285. 41:58see all of the prompts that they use to write the entire thing so you could go modify that and say rather than it
286. 42:06writing um you know fiction books I want it to write um selfhelp books and you could just go in and modify all the
287. 42:14prompts and create a version of GPT author that's GPT self-help author um that's the cool thing about
288. 42:23where we are right now all of this [ __ ] is just like out there you can play with it no one's figured out the [ __ ] rules yet so go
289. 42:34play don't come to me though if like you get a letter that says you got to take your book off Amazon because it's you can't copy I don't
290. 42:42know um because is that going to happen yes is it GNA happen to everyone no it's going to happen to the big boy they're
291. 42:48they're going to go after the you know companies the company that just raised 10 billion from Microsoft there's going
292. 42:54to be some lawyers knocking on there their Door Tech startups established companies wanting to invest in AI I'm currently
293. 43:03working on a half billion doll request for a company are you an investor Becky that's cool I've got some I've got
294. 43:11some I've got a so I've got an 11-year-old um video technology company that I've got an AI version of it an
295. 43:18omni Channel version of it that I'm trying to get funded through clients right now but I think I could probably
296. 43:24spin it up as a separate company so if you're in that world I'd be interested to talk to you to learn more
297. 43:30um what I was asking was would the accountability cancel the small mom and pop shops in the AI World well so this is so what what you're
298. 43:44asking Dash around the world is a really big question and it and and I don't think it's just relegated to Mom and Pop
299. 43:52Shop so there's a you would probably find there's an article out there from a Google engineer um that that was
300. 44:00apparently a leak I don't know if it was a leak or if it was something you know Google was just putting out into the
301. 44:05world but but go to Google and search for um the phrase we have no moat and it's an article from a Google
302. 44:13engineer that basically says both Google and open AI don't have any moat they don't have any defenses against these open- Source
303. 44:22projects that are getting for $100 in training the open source projects are getting these cheap little models as
304. 44:31good as the model that took Google six months and $10 million to to train and they're doing it for like a hundred
305. 44:38bucks and they're doing it in a week it's a really interesting article um whether or not it actually plays out that way I don't know but
306. 44:47but the challenge with the generative AI stuff right now is it's so easy to replicate anything right if someone
307. 44:56comes up with a really interesting tool that automatically generates sales proposals in in a new and unique way
308. 45:06where you answer three questions and it generates a sales proposal for you that's the best sales proposal you've
309. 45:11ever seen the ability for someone to be able to reverse engineer that and knock it off is like that and so I think every
310. 45:19single company right now is going to be struggling with two things one is where's the competition coming from for
311. 45:26our current business right like what do we need to do to our current business to survive that's that's the first thing
312. 45:34and then assuming that you figure out what to do with your current business to survive how do you how do you maintain a
313. 45:41Competitive Edge in a world where everyone can just look at what you're doing and instantly go knock it off not
314. 45:48like in the next three years if I invested enough money I could copy that company I can copy that company probably
315. 45:56over the weekend even if I don't know what I'm doing because of these tools so that's the environment we're
316. 46:02going in I think I think investors are going to be trying to figure this out like what do you invest in do you invest in a
317. 46:10company that's got some really cool proprietary thing that someone can just knock off tomorrow easily like that's
318. 46:17that's the world we're entering um you know are we there right now no but like you know I'm looking at my current
319. 46:25company I'm looking at the current value propositions of my 11-year-old you know video technology company and I'm I'm
320. 46:33looking a year out and I'm saying at least one if not two of the three major value propositions diminish in value
321. 46:40within a year and and one of them dramatically um so yeah I think it I think it'll hurt them now Sam Alman in talking a lot of
322. 46:55the a lot of the conversations Sam alman's been having recently with the government about you should regulate
323. 47:00open Ai and you should regulate Google and you should if if anyone gets any large language models that are stronger than
324. 47:08gp4 like if you get to GPT 5 or actual AGI those should be licensed like you have to license nuclear power plants so
325. 47:16he's talking about that but what he's also saying in those conversations is you shouldn't regulate the small AI
326. 47:22companies because they need to be free to innovate um so CU basically all the small AI companies are going to are
327. 47:29going to take the models that these guys Google and and thropic and and this new company that just got funded and open AI
328. 47:38they're going to take those models and they're going to build a whole economy on top of these models that these guys
329. 47:43build and that he's saying those guys shouldn't be regulated so I think that I think the competition's going to be
330. 47:49insane and and wild and creative and Innovative but but which of those companies can actually maintain that
331. 47:58competitive Advantage I don't have a [ __ ] clue so it seems what was that all the corporate Empires hold the
332. 48:08strings not with the open source thing anymore go go look at the open source large language model projects go read that go read that
333. 48:17article we have no moat it's it's or that it's not really an article it's like a leaked internal post from from
334. 48:28Google it's it's pretty interesting I want to learn as much as I can more than anything that that's I mean listen
335. 48:36that's I I've been like just like a like a crackhead with with this stuff for the past six months and I I don't see it
336. 48:44stopping I like like I just I feel like it's actually really important um at least for me personally but I think it's
337. 48:52important for a lot of people get yourself as educated as you can on what these things make possible and and what
338. 49:02how they align with your skill sets and like what you can make them do because if you can take what you're good at and
339. 49:08figure out how to use these tools whether it's mid Journey or chat GPT or one of these video tools or one of the
340. 49:14music writing tools if you're good at that [ __ ] um there's going to be a place for you in the workplace and probably
341. 49:21you know an elevated place for you in the workplace so like I feel like the investment the investment's relatively
342. 49:28low because these tools are simple right now um like learn as much as you can and I I think it's going to pay dividends
343. 49:36for anyone that does it that's that's why I'm here that's why this channel exists is that option in the chat. openen AI is what option is
344. 49:50the um code interpreter it might be I I mean code interpreter is going to be in chat. open.com so so what you get in
345. 49:58chat. a.com that's chat gpt's official website if you pay for GPT plus you get gp4 gp4 connected to the internet and
346. 50:09gp4 with plugins and then plugins are basically like there's a plug-in store there's almost 600 of them now they're
347. 50:16just little extensions to chat GPT that sort of connect it to the internet in very specific ways that's essentially
348. 50:22what it is um but you also get Early Access to whatever other Technologies are coming so we know this one's coming called code interpreter
349. 50:31that's this really powerful tool so those are the things that are available there but I don't quite know what you're
350. 50:35asking so if you clarify that I'm happy to do it I know investors we should talk all right um connect with me Becky on
351. 50:43like go to my profile my contact info is in there or you can just you know DM me on on Tik Tok here um we'll set up a
352. 50:50time to talk be be interested to to speak with you sounds to me like I could can try to make comics by AI writing
353. 50:57plus images yeah go look at there's a the the copyright office there's a there's a woman that did a graphic novel
354. 51:09um in the in the last year that the copyright office said the book is copyrightable but the images aren't or
355. 51:16something like that um it's worth looking into whether or not you can copyright it um but if you just want to
356. 51:23create it and put it out there and you're not at worried about all the copyright stuff I would just go ahead
357. 51:27and experiment with it if you really want to lock that [ __ ] down um go read up on what's happening with the
358. 51:33copyright office right now right now they're basically saying if the machine created it it's not copyrightable my
359. 51:41question to the copyright office is where's the boundary between the machine created it and I created it so if I
360. 51:48create if I if I take a a photo with my digital camera and I open that in Photoshop and it's this big and I use
361. 51:57generative fill to fill out something that's this big but it's my image in the middle at what point is that image not
362. 52:05my image is it 10% bigger 50% bigger 80% bigger that that's AI generated I don't know like like a lot of the work that
363. 52:17I'm seeing that that people are putting out in the world is kind of hybrid work between their work product and AI work
364. 52:23product and when you have hybrid stuff like that I think what's going to end up happening is the copyright office is
365. 52:30going to start to set standards of here's what it looks like to be human created and and here are where those
366. 52:36boundaries are those boundaries don't exist right now they're like if it's AI it ain't copyrightable those boundaries aren't
367. 52:44clear to me and I think they're going to get less and less and less clear over time so but yes you can absolutely go
368. 52:51create uh a graphic novel or a comic right now the IP lawsuit will never succeed the argument is college degrees
369. 52:59are earned via published books I I totally agree listen I'm not saying that any of this these copyright and
370. 53:06trademark lawsuits are valid like I totally agree like if I if I go to art school and I study um you know pop art
371. 53:14and I love Andy Warhol and I graduate college and I go to a bunch of art that looks like Andy Warhols like that's just [ __ ] I learned
372. 53:24and that's my style and and you know some of the foundations for these artists like I think Keith herrings is
373. 53:31one he's got a really specific look they're suing the [ __ ] out of people so anyone can sue anyone in America um so I
374. 53:38think there's going to be a lot of lawsuits but I don't I don't I think ultimately this will kind of normalize
375. 53:43kind of like music sampling did it'll take some amount of years and people will be pissed off but it it'll normalize at some
376. 53:51point um actually they're not investors they're funders and they don't want to own a percentage of your company oh
377. 53:58that's interesting all right yeah happy to talk about that you can own 50% copyright prompts aren't copyrightable I don't know why
378. 54:07you'd want to copyright prompts I guess if you wanted to sell them there's a there's a website right now prompt something
379. 54:13prompt prompt base is that it that that they're selling prompts I think that's kind of ridiculous cuz I I think prompts
380. 54:21are going to drift into the background ultimately um it's good learn them right now it's good to understand how these
381. 54:27tools work what they're good at what they're not good at um but I think ultimately prompting drifts into the [Music]
382. 54:38background collage is not copyrightable in most cases interesting thank you for the Roses Olu 2533 cheers some some fresh
383. 54:52AA hey you want to hear something crazy so I live in Denver and the first 6 months of 2023 Denver has gotten more rain than
384. 55:04Seattle that's never happened before in history it's a it's a an El Nino we just exited Alania and we've
385. 55:13entered in El Nino and I think there's just a giant hose from the Pacific Ocean coming coming over the Rocky Mountains
386. 55:20apparently uh but yeah Denver's gotten more rain in the first 6 months of 2020 three than Seattle [ __ ]
387. 55:28insane prompt AI I think oh interesting okay the princess oh okay you're talking to each other cool can we see the back yes sorry about
388. 55:38that okay so if you're new here oh we got it so some new people here uh my name is Kyle Shannon this is the AI
389. 55:43learning lab these URLs behind my head what are they why do I have these here I'm not related to these in any way the
390. 55:50top three the top three are just basically variations on chat GPT the top one is chat GPT chat. open.com that's
391. 55:59the official chat GPT website if you've not played with it go there now just go and start playing with it um prompt.
392. 56:06chat is a little tutorial document that'll teach you how to do prompting and give you a whole bunch of ideas to
393. 56:13just go play there so just go play and do par parlor trick what I call parlor tricks have it right you poems and songs
394. 56:20and and tell it to create a table that compares two different topics you you know about um have it write you code
395. 56:27just start to see what this thing is and how it's different than Google cuz it kind of looks like a Google search box
396. 56:33and you're like oh it's just like fancy Google it ain't fancy Google it's it's profound right but you got to dig deep
397. 56:41enough what's interesting about chat GPT is if you give it a generic concise prompt it'll give you back shitty answers if you give it
398. 56:52really specific High context I want you to act like a you know special mathematician in this brand branch of mathematics and I want you to
399. 57:02help me learn about this branch and B if you give it a lot of context it gets really good it's like the harder you
400. 57:08push it the better it gets so that's that bing.com is also chat GPT but it's it's on Microsoft's search engine Bing so if
401. 57:19you click on the chat button here that's actually gp4 and it's connected to the internet so you got to pay for gp4 with
402. 57:26these guys you get it for free with Microsoft and then po.com gives you access to six different um chat GPT
403. 57:34models from different companies and there's some cool stuff there and there's also a bunch of bots there that
404. 57:40you can make your own bot like you could make I've made a Shakespeare bot I made an Emily Dickinson bot I've made a bunch
405. 57:46of bots you can make Bots it's fun prompt. chat will teach you about prompting and then futur pedia.com AI tools
406. 57:57so all right thank you for the very valuable info you're welcome saves me a lot of time stumbling around there is so
407. 58:04much [ __ ] right now that's it's why I'm doing this channel I will change out these URLs by the way I'm not affiliated
408. 58:11with these in any way I'm not getting any money from them I'll change these out when I think it makes sense like
409. 58:16when there's some other tool that you should be playing with but I think right now if you're new to AI just chat GPT
410. 58:24just go there start playing with it start learning it um start using it for other people start using it um to solve problems for
411. 58:35yourself and you'll see there there's there's there's some there's some gold in them their heels some gold in them
412. 58:44near heels you know what I mean Buck you know what I mean I don't know about this AI stuff ain't the robots going to kill
413. 58:51us ain't the robots going to I'm pretty sure I I saw a movie once where the robots I think they called it an AI I
414. 58:57think they called it an AI and AI killed all the people killed all the people I think that's what these are so I
415. 59:03wouldn't I wouldn't touch him I wouldn't go there don't be that guy cuz he talks weird for one thing he's got a friend named
416. 59:13Buck all right now I'm going to get hate mail from people who have friends named Buck um I meant that author tool you
417. 59:21mentioned the author tool does not do it doesn't do Comics it does but you could you could modify the author tool to do
418. 59:31Comics um but you you'd have to geek out a lot but what it writes right now is a fiction novel and then it does the cover with
419. 59:39stable diffusion so it does it does the cover graphic and the their their prompt for their cover graphic kind of sucks
420. 59:47cuz the two books that I did the covers that It produced were like you could sort of see where it was going but it
421. 59:54had nothing to do with the title and like it just it just wasn't good um so I I would you know I think the it doing a
422. 1:00:01cover right now is a gimmick um if they got their prompting better it would be interesting and you could probably
423. 1:00:07pretty easily modify that to make a graphic novel to to at least write the text of a graphic novel and describe what the images
424. 1:00:16are um but I'd go play with it it's called GPT author and and uh that's one of those things where keep going back
425. 1:00:25and you know checking with that project cuz that's one that they'll probably update that thing a lot and it's going
426. 1:00:31to get better and better and better I'm one of your students I started yesterday very cool students in what in
427. 1:00:42this this is welcome welcome to to to old guy University where are you learning I don't know some old guy on t Tick
428. 1:00:55Tock what's he charging you seems to be free but anyway cheers I actually did a class today um with a with a with a bunch of guys that
429. 1:01:10I work with doing um automations on zapier it's and it was really fun really really fun can you tell chat GPT there's
430. 1:01:21Amelia's wife hello there um can you tell chat GPT to only use certain sources like math props not the students so Dr J what you're talking
431. 1:01:34about there's actually um there's a dude named Andre karpathy who's one of the founders of of chat GPT of open Ai and he was speaking
432. 1:01:46at the Microsoft developer conference and he used that exact example to explain why having chat GP act like a
433. 1:01:55certain Persona works and so I'll I'll explain it to you the best I understood it this is I'm dramatically paraphrasing
434. 1:02:01what Andre carpy was talking about so so I might might not have some of the some of the pieces right but this is the basic
435. 1:02:08concept so the way large language models work and and the the metaphor that I've come up with is imagine the Grand Canyon
436. 1:02:16is full of words okay so so they they took these large language models they trained it on all of the internet right
437. 1:02:24and so it looked at all of the words on all of the internet and it put it into this basically mathematical formula um
438. 1:02:30called a vector space which you know you know like an XY coordinate thing where you plot numbers on a on an XY
439. 1:02:36coordinate it's like that only instead of two Dimensions it's thousands of Dimensions so so just imagine this
440. 1:02:42massively complicated mathematical space that's all these semantic clusters of words so you might have the word dog in
441. 1:02:50this cluster it means it's a fuzzy animal you keep in your house and in this cluster words it means it's a guy
442. 1:02:56that did his girlfriend wrong right he's a dog right so so it sort of knows all the words and it's got all these sort of semantic
443. 1:03:03clusters So when you say here's a math problem I want you to solve it it's kind of like you're in a helicopter above the
444. 1:03:11Grand Canyon and you're really high up and it's looking at all of the Grand Canyon so you're seeing all of the words
445. 1:03:18and you're seeing all of the possible semantic areas and and exactly what you said there what what what karpathy said
446. 1:03:25in this talk to to Microsoft was um in all of those clusters the professor that solved that problem correctly that's going to be in there
447. 1:03:37and then all the students who didn't solve that problem correctly are also going to be in there so if your
448. 1:03:42helicopter is too high it's G basically going to average in all of those math problems and give you some sort of
449. 1:03:49answer there's a good chance it's going to be wrong cuz it pulled in the student stuff as well but if you said say I want
450. 1:03:54you to act like a Stamford mathematics professor that specializes in this branch of mathematics and I want you to
451. 1:04:00help solve this thing what you're essentially doing is lowering lowering your helicopter into one of the side
452. 1:04:05Canyons that's just you know about Stanford Professor mathematics and your chances of getting that thing go up
453. 1:04:12dramatically so great question and that's exactly how it works and that's why so this prompt. chat thing is really
454. 1:04:20just a long list of act like a certain kind of act like a novelist act like a mathematician act like a somaler act
455. 1:04:29like a gardener and that's exactly what it's doing is it's narrowing that semantic region so that your answers are more accurate and also
456. 1:04:40gp4 is much better at not hallucinating telling lies than GPT 3.5 I was asking you to lead me to the author tool you mentioned oh um it's
457. 1:04:56called GPT author let me go see it should be on GitHub GPT
458. 1:05:13author with no [Music] space hold please let me go find
459. 1:05:30it let me go to Google
460. 1:05:48collab it's GPT Dash author hold please GPT Das author
461. 1:06:07AI okay there's a medium article here hang on I'm going to flip my camera around so you can see this because it
462. 1:06:14it's a fun tool to it's a fun tool to play with you you have to dig through getting it set up on Google collab but
463. 1:06:21I'll show you how that works um so this it's a medium article GPT author a constrained AI tool for generating
464. 1:06:32instant fantasy novels so go search for go search for GPT D author and Ai and this this will come up this medium
465. 1:06:42article and then it will link you over to the GitHub link so the GitHub link is GPT Das author mumer github.com
466. 1:06:59mumer dgpt [Music] athor now is this the
467. 1:07:16one yeah this is the one so there's so I've got one that's a copy of GPT author version 1.1 I would assume that that's probably higher than
468. 1:07:26that right now but but the way um Google collab works is you have these little cells that have little play buttons and
469. 1:07:34the play buttons are actually running code and like loading libraries and stuff like that so like see how it's
470. 1:07:42twirling there and then it'll get a little green check mark when it's done see it's it's like it's like doing some [ __ ] it's
471. 1:07:50installing some things this so so this is running on my Google Drive it's Google collab so I ran that cell it's
472. 1:07:58got the code in it and then the next cell I run it and it's you know it'll do its thing building a wheel for Google trans
473. 1:08:09whatever that is that gave it a check mark now I'm choosing my primary language and then here this version that
474. 1:08:18I have you type in your author The Prompt itself like what do you want the story to be about what the writing style
475. 1:08:25so those are your inputs and that's all you have to input like you know what's your story about what's the writing style and
476. 1:08:34then you keep running these cells and is this the last one no it's not the this is sort of loading the prompts and I think is this the one
477. 1:08:51no so this section with all the these codes in it these are all the prompts that write all the chapters so so this is where you could
478. 1:09:00modify like all the things in red here are all the prompts that write the the book and then the final thing book
479. 1:09:08writing is hit play and now it's going to start writing my book so that's kind of how collab works is you've got these cells with code in
480. 1:09:20them that people that are geekier than me wrote and then you hit the play buttons on the cells and occasionally it
481. 1:09:27will screw up and then you can go to chat GPT and have it help you debug while it's screwed up um and what the other cool thing this
482. 1:09:35does is as it's doing each of the steps it's telling you how much they cost so it's calculating the cost from open
483. 1:09:42AI so this is each step here is about a third of a cent you know a fourth of a cent a third of a cent a tenth of a
484. 1:09:52cent generating storyline with chapters and high level details so so basically now it's going through and writing a
485. 1:09:59book so it's just just absolutely insane and this is like that thing of of you know it having all these steps in it and
486. 1:10:09and basically I just say I want you to write a book kind of like this uh it'll just go do that there are all sorts of
487. 1:10:15crazy fireworks going on right now or we're being attacked in Denver I didn't I don't know what's going on Friday night oh maybe it's a
488. 1:10:25ball game it's a Fourth of July weekend for a ball game or something maybe the Rockies won all right so hopefully that helps let me
489. 1:10:36turn on my My Little Secret background [Music] here is someone there no what is the most fun thing you can
490. 1:10:52think of with with using AI right now for me it's being with people when they have this
491. 1:11:07moment that's why I do this it for me it's all about the possibility
492. 1:11:22of I think it's really easy to fall into the Trap of this is evil it's not perfect it's going to [ __ ] everyone's
493. 1:11:32job up like it's really easy to go into a negative place with it and what I'm finding is the more I dig into it the less it
494. 1:11:40feels like that the more hopeful I feel like the more I learn the more hopeful I feel because I see what it makes
495. 1:11:50possible and so for me it's not any specific thing it's like it's like it's that Journey it's the Curiosity and the exploration
496. 1:12:04and imagining what the future could be like if this bends in a positive direction and I think that every person
497. 1:12:18every individual right now that gets curious about this and bends it even slightly in a POS positive
498. 1:12:25direction will have a disproportionately positive impact on the trajectory of AI so I think that's it like I get
499. 1:12:37really philosophical about this stuff because I there's a couple of things I think we are incredibly lucky to live at
500. 1:12:46a time when we get to play with tools this powerful we there there has never been anything in in the history of human
501. 1:12:53it that's been remotely close to as powerful as what that is it's it's like a Humanity amplifier it's like a
502. 1:13:00Humanity reflector if you think about what that is it was trained up on all of the internet so think about all of the
503. 1:13:06data and content that's been jammed into the internet for the past 30 40 50 years it all got sucked up it all got vacuumed
504. 1:13:14into that tool and now we can play with it oh there is someone over there hang on I'll be what you
505. 1:13:30do I'm not listening to you okay that was crazy those fireworks were nuts weren't they I was like is there someone over there there
506. 1:13:39was um so yeah so so for me that's it um someone asked about future pedia.com get out it every day um most
507. 1:13:56of the AI tools right now are just horrific pieces of [ __ ] so just know that out of the 3700 tools there there's
508. 1:14:05probably 200 that are good but the 200 that are good are going to be profound right so so you're going to
509. 1:14:14have to sift through some crap it's like finding that an app in the Android store like you know it's you know 98% [ __ ]
510. 1:14:21it's it's going to there's going to be a lot of that for a while with this stuff cuz everyone's so excited about it I got
511. 1:14:27an idea for an app oh chat GPT wrote it for me let me put that out in the world oh it's a horrific piece of [ __ ] yeah me a lot of that
512. 1:14:38um yeah I I you know that's fun for me right now is uh like future gazing with AI does 100 pounds of flour make a Big
513. 1:14:51Biscuit that's you s Ross per does 100 pound of flour make a biscuit that's a Gen X joke nobody knows who Ross per is anymore where is the
514. 1:15:00fiction tool I just showed it I'm not doing Comics LOL um I thought you said Comics but maybe not maybe I misread he answers questions I
515. 1:15:17do can someone catch me up just got here hey an how's it going Kyle can you host workshops I can I've thought about it
516. 1:15:28um I keep thinking about it it's I I like I don't want to host host one if it's shitty so um amazing how incredible new
517. 1:15:40discoveries we can learn can open opportunities okay cool I'll be checking that out how does AI categorize the data
518. 1:15:49correctly I.E it really comes from a processor of mathematics yeah I can if do you want me to if you guys want me to show you I can actually
519. 1:15:58show you how it works I've got a slide that I did for a presentation on how chat jpt works when you understand how
520. 1:16:04it works it's even more insane that it does what it does thank you for sharing your brain with us um if you want me to
521. 1:16:11do that I'm happy to share it um to to show you how it works um he's talking AI he answers questions I know but which
522. 1:16:23subject I'm just talking about I'm just answering stuff okay uh hang on I'll be back like does he make me for a long
523. 1:16:35time yeah he wants you to take off I'll do that later let me finish okay I'll all right I'm back I got to go soon
524. 1:16:48got family duties all right um this is great will you make a this video available so we can rewatch it and learn
525. 1:16:55more so here's the good news I found out three days ago that all of these lives are actually recorded and it makes
526. 1:17:03little Snippets of them and I've been posting them to my channel and I'll I'll be able to download all these videos so
527. 1:17:09I will probably start a YouTube channel or a playlist and I'll put all these up there so at some point I will they're a
528. 1:17:17bit rambly right now so I don't know the value of that but but we'll see so I don't know that I'd want to watch
529. 1:17:23whatever the [ __ ] I'm talking about for two hours cuz CU I tend to stay on here a lot and I don't really know what I say
530. 1:17:29but I'll go I'll go take a look at them so at some point I'll do that yeah and I'll think about doing a workshop um I
531. 1:17:35just need to figure out like what's the what's the model where do I do it what's this what's the topic things like that
532. 1:17:42um looking at it glass half full I agree I've started using it a lot more over the week since great oh since watching
533. 1:17:48your stream before this about six months ago I thought it was just Google it's it it's so [ __ ] deceptive it looks
534. 1:17:56like it's just Google and then you and then you see someone it writes a poem and you're like oh it's fancy Google and
535. 1:18:02it's not until you go well wait a minute what what's what's something I do I'm a project manager and then you type in you
536. 1:18:10know give me a project plan for blah blah blah out it comes and you [Laughter] go that's the best [ __ ] moment that's
537. 1:18:20the thing that makes me happiest is watching that I had I had to have with a 10-year-old kid who literally went like
538. 1:18:28this when he looked at me and goes this changes everything it's crazy it really is crazy who created this I don't know what
539. 1:18:39created what is there an AI platform that can create characters for a book visual characters um Mid journey and stable diffusion can um
540. 1:18:54um it's a little problematic to get that character to repeat consistently in different locations it there there's
541. 1:19:01ways to do it but it's it's really complicated so but yeah if you want to do initial character development for a book you can absolutely do
542. 1:19:09that um Mid journey is the one that I would recommend you have to generate your images on Discord so to use midj
543. 1:19:17journey you have to learn Discord and then you have to learn how to use mid journey within Discord and if you've
544. 1:19:23ever seen Discord which is just like geekier version of slack um you're like why are they doing that there and that is the right
545. 1:19:33question but it's the best and then if you want to use stable diffusion you need a computer science degree not
546. 1:19:40really but you got to be able to run notebooks and do some geeky stuff to install it all right is there an AI
547. 1:19:47platform okay that that one I just answered AI to build Advanced what was that advanced technology sounds great yeah it
548. 1:19:57yeah I mean so here's here's part of the challenge of where the technology is right now especially with this
549. 1:20:05generative AI stuff the the the new stuff like in the last 6 months is it's it's so powerful that you can see
550. 1:20:15glimpses of what it's going to be and you're like oh this is going to totally transform whatever project management or
551. 1:20:22programming or whatever and then when you actually go to implement it it's like 90% brilliant and 10% like it's not
552. 1:20:30quite right that last 10% is like if you've ever remodeled a house it's like getting to 90% we're pretty much on
553. 1:20:38track that last 10% takes you like four years I think I think that's what's going on with AI right now that it's
554. 1:20:46really easy to get conceptual stuff close but then to get it to actually be good software that's that's going to take a
555. 1:20:53while I like your personality please host a workshop oh thank you I appreciate that I will okay all right
556. 1:20:59I'll put one together I got to figure out what it what it is how long it is is it like multi- weeks how deep do we go
557. 1:21:05is it just you know basic [ __ ] I I don't know you know here here's here's my request to anyone that's here that that
558. 1:21:12wants a workshop is is connect with me on Tik Tok here and and DM me and just say you know if you're if if I were to
559. 1:21:21do a workshop what would you want it to focus on what would you want it to be maybe I can get some input because I
560. 1:21:26don't want to just put something out there and have it be a waste of anyone's time um he has to he has to do a workshop
561. 1:21:34fine I'll do a workshop that'll be helpful please don't discount the value this information is awesome great well I
562. 1:21:42always discount everything because shame is my love language just it's the way it is don't overthink immediate practical Zoom yeah
563. 1:21:54I know I yeah you're right I'll do that I'll put something together um I don't think you're rambling this is hard to
564. 1:22:01understand because it's at its infancy yeah it really is I'm all about a workshop okay great um Workshop Workshop
565. 1:22:08it seems to be that seems to be a a theme emerging in in the comments can I make a professional logo for my company with AI you can [Music]
566. 1:22:19um there's there's some issues with logo stuff so so doing things in mid journey and stable diffusion it's just
567. 1:22:25generating bit map images it's not generating Vector images right and if you just like what I would say is this
568. 1:22:34what what you can what you can do in mid Journey with logo development is you can concept brilliantly um but if you really want to
569. 1:22:45like tie in a logo and get the typography right like it won't do typography at all like it just these
570. 1:22:51things don't do typog photography at all the image itself you could probably do quite well um but in the end you're
571. 1:22:58going to have to hand that to a designer to really turn that into vector and fine-tune it and things like that
572. 1:23:03depends how anal you want to get with The Branding um but yeah yeah you can what I would what I would recommend with
573. 1:23:10that is go to um if you're on Facebook look up the mid Journey official group on Facebook and then within that group
574. 1:23:19look for logo development and you'll find find a whole bunch of people doing it and probably within there you'll find a whole bunch of
575. 1:23:27prompts um the other thing to do with mid journey is because it's on Discord there's a whole Community there
576. 1:23:35and there's all these different channels so there's probably a channel within mid Journey actually let me go look there's
577. 1:23:41probably a channel within mid Journey on um like logo development or you know let's see abstract characters image
578. 1:23:56prompts group jams yeah no I don't see anything specific to logo development but there's a show and tell there's paintovers
579. 1:24:07there's image jams there's prompt FAQs ideas and features so that's so so within the mid Journey Discord there's
580. 1:24:15there's lots and lots of things you can also join the AI Salon that that I created if you go to the salon. I in
581. 1:24:23fact let me put put this up behind me [Music] here this if you go to that URL that's a link tree to um a community that I've
582. 1:24:40created of AI enthusiasts and things like that we meet every other week There's a Discord server where you can
583. 1:24:46geek out and ask questions but I would do that so you can you can definitely do that yeah I made logo but it doesn't
584. 1:24:53like letters Shameless Works um are there any certifications you know about that generally that you about that
585. 1:25:01generally proves you know a lot about AI not right now well okay two different flavors of AI there's there's pre-chat
586. 1:25:08GP Ai and there's post that's kind of how I'm I'm I'm I'm looking at it right now pre-chat GPT you're talking about
587. 1:25:17machine learning and Ai and things like that if you're an engineer and you're and you want to go more on the tech
588. 1:25:22technal side that's a whole different conversation and you know you should have engineering background and
589. 1:25:28understand models and neural networks and [ __ ] like that if you're talking post chat GPT um if if if there are uh
590. 1:25:39certificates they're not worth anything because anything you learn right now is going to be you know worthless 6 months
591. 1:25:45from now because everything's changing so fast what I think is a really good best practice right now is get really curious
592. 1:25:53about this stuff get as good as you can about it and start posting what you're learning start sharing what you're
593. 1:25:59learning um if people at work are curious about what you're doing say hey do you want to learn something sit down
594. 1:26:05I'll I'll sit down with you for an hour I'll teach you how to do it start putting yourself out there that you're
595. 1:26:10learning about this AI stuff that you're doing it demonstrating that you've got some Acuity with this stuff is the
596. 1:26:18certification right now that's how early we are a year or two from now there's going to be you know decent courses and
597. 1:26:25things like that but right now it just everything's in flux everyone's trying to figure out what the [ __ ] this thing
598. 1:26:30is how it works how you do it how you do it well how you get it not to hallucinate what the hell hallucination
599. 1:26:37even means all that sort of stuff like um so so so I think it's on individuals right now to get curious about it and
600. 1:26:47start teaching one of the one of the people that was in here um in in the lives has she she set a goal for herself
601. 1:26:53to teach a thousand people about Ai and in doing that she's now doing um workshops and things for corporations at
602. 1:27:02like 10 grand a clip so like just go just go am I an educator or teacher I am an ex actor I'm actually a founder and CEO of
603. 1:27:17a video technology company called storyvine and I I haven't done any formal teaching but I'm I'm really
604. 1:27:25passionate about this stuff and I think it's I'm passionate about demystifying it because it drives me [ __ ] crazy
605. 1:27:34when I hear people demonize it or diminishing it without having tried it it makes me insane because this shit's
606. 1:27:41profound it's going to change everything it's going to change everything it's going to change the definition of work
607. 1:27:46it's going to change the value of what we do and how we contribute it's going to change the economy it's going to
608. 1:27:52change businesses and it's coming a shitload faster than I think anyone's prepared for and so I'm just super passionate
609. 1:28:02about you know I someone was on the live last night he like is all you talk about chat GPT for now because the thing that sits at that URL
610. 1:28:16is so profoundly powerful I could talk about it for the next decade and not wear it out I don't understand what it is or or how
611. 1:28:25deep it goes at all like at all at all like I barely scratched the surface are there 8,000 other things to
612. 1:28:36learn yeah that's why I put this URL here if you want to geek out on all the other [ __ ] but but understanding just
613. 1:28:42the basics of what large language models are and what they do and what they make possible and how they're going to change
614. 1:28:49your job and your job and your job that's what I think we all anyone who doesn't do that is going to get
615. 1:28:58blindsided by this thing and it's going to be really [ __ ] painful and that makes me very sad I don't like I don't
616. 1:29:04want it but it's coming and so I want as few people as possible to be hit by that Title Wave uh let me do a couple more questions
617. 1:29:18here I'll do another 10 10 minutes or so um oh that was you that was you and that's cool yeah the the the the goal of
618. 1:29:27teaching a thousand people like far out yes that [ __ ] awesome um what's the elevator pitch for AI that's interesting well so I wouldn't
619. 1:29:40call it AI like so there's so chat GPT H that's really interesting it's the all this stuff is more machine learning than AI
620. 1:29:53um chat GPT is is the word version but like the it's a great question the Transformer model hang on
621. 1:30:07dog's being loud um the Transformer model that Google invented in 2017 is kind of the foundation for all of the
622. 1:30:15tools that we're playing with right now so chat GPT um Mid Journey stable diffusion the video things the medical
623. 1:30:23things all the [ __ ] that's happening right now is is because of this thing called the Transformer so they're taking
624. 1:30:28these machine learning models they're they're they're basically pointing them at massive data sets and then when you
625. 1:30:35point it at like Radiology images it gets really good at reading X-rays and when you point it at brain scans it
626. 1:30:42understands what the brain is seeing and when you point it at all of the words on the internet it [ __ ] understands all
627. 1:30:47of the words and when you do it with pictures it understands all of that or it doesn't doesn't understand it it it
628. 1:30:53it lets you create it with that okay so the elevator pitch the thing I've been saying recently is it's it's a Humanity
629. 1:31:06amplifier or it's like a talent accelerator or it's a
630. 1:31:23like like like I keep going back there's a quote from Arthur C Clark that any sufficiently advanced technology is
631. 1:31:28indistinguishable from Magic it's that it's that when you like here watch watch
632. 1:31:45this like I've done this a lot and every time I do it it still blows my [ __ ] mind all right here's chat GPT we're just using GPT 3.5
633. 1:31:59um write a python application that
634. 1:32:14um calculates the um um navigation for a
635. 1:32:29rocket to land on the moon I don't even know if it's possible that it can do that
636. 1:32:45so calculating the navigation for a rocket to land on the moon involves complex calculations and consider ation
637. 1:32:52well I can provide you with a simple example of a python application please note that the actual moonlanding
638. 1:32:57navigation requires sophisticated algorithms and data right that go beyond import Mass so it Imports a library then
639. 1:33:04it's like it's got the moon's radius it got it's got the moon's gravity it's got total L latitude and longitude if you're
640. 1:33:10an actual astrophysicist this is probably comically simple but if you're someone like me that has no [ __ ] clue
641. 1:33:18how I would even begin to approach this I at least can look at this and go okay I understand at least what they're trying to do there
642. 1:33:27um you know let's say um write uh let's see write um a sub routine um that will orbit the Moon
643. 1:33:44three times before landing can it do that yeah apparently I mean and I mean the thing about
644. 1:33:59code is you can run it right so is that code correct I don't know it's probably got mistakes in it there's probably
645. 1:34:07libraries missing it's probably oversimplified it's probably not great I know that if you do code three times in
646. 1:34:14a row it'll take three different approaches but probably within an hour I could get some sort of simulation working with
647. 1:34:23some sort of rocketry simulator which it will tell me exists and put that code in it and it will do something that's
648. 1:34:31[ __ ] miraculous and it and like that's that's a teeny sliver of something it does the same [ __ ] with
649. 1:34:38language H have it write you Shakespeare like stuff it can do that have it write you in the style of Mary Shelly have
650. 1:34:46it write you a marketing plan have it write you tweets and social media a copy it's like it's it's just this Limitless
651. 1:34:58hole of possibility and and and it's just like it's just one tool that in the next three to five years is going to get increasingly
652. 1:35:09sophisticated it's going to be multimodal like right now it just does text it's going to be able to spit out
653. 1:35:15images and then animations and then films and then music and then it's going to connect to all the [ __ ] that you're connected to
654. 1:35:24and it's you're going to give it permission to access your bank and your travel accounts and your grocery [ __ ]
655. 1:35:29and you're this and you're that and you're just going to like this is like Jetson [ __ ] you just talk it and it will go do things for you
656. 1:35:40like that's where we're going um yeah so I so back to what's the elevator pitch for AI I have no answer for you
657. 1:35:53that was not an elevator [ __ ] that was a [ __ ] that was a [ __ ] di tribe I think it's a human it's a it's a
658. 1:36:03it's a Humanity amplifier I think it takes who we are and what we've done and repackages it so that every single
659. 1:36:12person can tap into that genius maybe that's it I don't know it's a great question with a shitty answer let's what
660. 1:36:21application allows to go beyond the text restrictions of GPT well so with plugins you can go beyond the the text restrictions here
661. 1:36:31let me let me show you another thing I I'll this is this is actually quite simple what it's doing but this this is going Beyond
662. 1:36:40GPT the the text restrictions of there's a couple of them okay so we're going to we're going to play with some plugins
663. 1:36:47now so if you haven't played with plugins on chat GPT the the way it works is you go to gp4 if if you're if you've
664. 1:36:55paid for the plus thing and you've enabled your beta uh Services you can you can play with plugins so I'm going
665. 1:37:02to go to plugins and when you play with plugins you have this little drop- down menu and here's the ones I have
666. 1:37:07installed and then in any s in any session I can turn on up to three so I'm going to turn on instacart I'm going to say make me a meal
667. 1:37:20plan that looks like dog food but is healthy I don't know you should also be having fun as you do this it sounds like you're
668. 1:37:35looking for a meal plan that might resemble the appearance of dog food but is actually healthy all right scrambled
669. 1:37:41tofu and vegetables lentil stew this is good ground turkey vegetable Skillet um and then you'll notice what
670. 1:37:50it did down here would you like me to create a shopping list for these meals on instacart and I'm going to say
671. 1:37:57sure and now what it's going to do is it's going to fire off the instacart plugin and it's going to take my little meal plan and turn that
672. 1:38:10into a shopping list and it's going to give me the the hyperlink to the shopping list so this is all text but now it's going to pop me
673. 1:38:20over to in cart and there's all of my ingredients for my dog food meal plan right and those could be on my um porch within the hour so so
674. 1:38:38there's one um there's also a plugin I haven't played with this one very much so I don't know it but let me
675. 1:38:45turn off this one there's a plugin here called show me diagram I'm going to say um create
676. 1:39:00um a flow chart um of the House of Representatives or no let's say of the US Senate because it's
677. 1:39:18smaller and that should find off that um show me diagram thing and it will go I assume what it's doing in the background
678. 1:39:28is doing some sort of search of something mermaid graph so so it pulled a graph in oh there you go so there's voters State selectors
679. 1:39:44bills you can view this F diagram in a new tab so there's the diagram in a new tab so so it's remedial right now but
680. 1:39:53and then and then and then there's other things in there there's there's hooks into stable diffusion things like that
681. 1:39:59so um so right now all of these tools are very very remedial but but there's there's ways that you can kind of you
682. 1:40:07know fake what they're they're where they're headed where they're going to be so I don't know if that actually did
683. 1:40:15what you wanted it to do but you know but but it's free this is a free class uh What applic uh okay we just did
684. 1:40:25that one Microsoft has a certification for teachers I can send you the link Mark cool um 500 nonprofit fundraisers
685. 1:40:34trained on chat GPT so far this summer oh that's so cool that's so cool this I mean this is going to change
686. 1:40:41philanthropy it's going to change [ __ ] everything it's going to change everything it really is it's going to
687. 1:40:48change everything all right I'm going to do two more questions and then I got to go I got to go deal with the kids the
688. 1:40:54dog the wife all of it got to go deal got life life's not just Tick Tock you can't just ticktock your life away what' you
689. 1:41:04do Friday night I don't know some old guy on Tick Tock weren't you listening to him last night shut up can you show the slid you mentioned
690. 1:41:17about how chat GPT is built yeah okay so all right this is I'll try to do this quick um this is how chat GPT
691. 1:41:27works it's staggering it's staggering how it works it's it's like the more you understand about it the more wild it is
692. 1:41:38when you see what it actually does the great
693. 1:41:53Renaissance which is that date this was sort of the money slide from this presentation I gave so I gave this talk
694. 1:42:00at at this thing called creative mornings and um compared the early days of the web to the early days of
695. 1:42:08generative Ai and this is the date that changes everything in my opinion and we are entering a new age um of like
696. 1:42:18dramatic human productivity and creativity and this was the slide I did on how chat GPT works so let me make this a little smaller so
697. 1:42:30we can see it all right so how this stuff works so so this gets a little geeky but but I I think it's
698. 1:42:43worth it okay so first is what does GPT stand for GPT stands for gener generative pre-trained transfer forer so
699. 1:42:51generative meaning meaning it's generating something it's generating original content I'll show you how that's what this is all about
700. 1:42:59pre-trained means it's been pre-trained you don't you don't need to provide it with data it's a large language model
701. 1:43:05that's been pre-trained on all of the internet and then Transformer is the technology so Transformer is the
702. 1:43:11technology that Google invented that allows these models to be trained on massive amounts of data so so this is
703. 1:43:18the thing that Sam Altman discovered the more data they threw at it and the more gpus they threw at it the better it got
704. 1:43:24so gpt3 and four are as good as they are because of that recognition that you can basically overload these systems with
705. 1:43:33massive amounts of data and then it compresses them into these very small usable things that we get to play with
706. 1:43:40large language model that's what these things are called in general the large language model is is these big models
707. 1:43:45trained on this that we can interact with then there's this thing called the latent space in the latent space this is
708. 1:43:51a dramatically oversimplified representation of it but think of this as an XY coordinate like a a two-dimensional grid XY coordinate each
709. 1:44:00of these little clusters of dots are mathematical um uh vectors of you know semantically related um words now it's
710. 1:44:11not actually the words in the database it's just the numbers so it's the numbers in in this in this case it's a
711. 1:44:17two-dimensional space the real space is thousand of Dimensions it's incredibly complicated and Incredibly profound but
712. 1:44:25but for this example it's it's that simple okay and then how this stuff works is that when you put in a prompt
713. 1:44:31it turns it into a mathematical weight or mathematical probability that then goes and looks at all of the words
714. 1:44:41in What's called the latent space and says based on that prompt which is the next most probable word to fit in that
715. 1:44:50sense sentence and then it does that over and over and over again so it's it's almost identical to when you're
716. 1:44:56looking at um you're typing in Google and and they have that auto complete at the end of a sentence that it sort of
717. 1:45:02finishes your sentences for you um it's like that only way more intense so if I type in the prompt the quick brown fox
718. 1:45:10jumps over the lazy blank you know we all know that's dog but how how this system does that is it says it takes
719. 1:45:17that and then turns that into a mathematical formula that then generates a probability and basically says okay in
720. 1:45:25all this latent space you know all these other clusters are not relevant but these light blue ones here you can't
721. 1:45:30really see them on there but there's light blue ones there um are are most probable and one of those that says 96%
722. 1:45:37that little dot there one of those is the most likely and if you look at what that actually is it's a big long string
723. 1:45:43of numbers it's a vector and then that long string of numbers is tied in a database to a word and that word is dog
724. 1:45:50and in this case and then the thing puts dog into that sentence and it does that over and over and over so when you see
725. 1:45:58it right like like a three-page story in 30 second or in 3 seconds it's doing what I just showed you word after word after word so it's it's
726. 1:46:09truly generating writing original content word by word just based on probability this is why it hallucinates
727. 1:46:18because what we experience is it writes something that looks coherent right we we look at it and we like oh my God it's
728. 1:46:25like how does it know that right all it's doing it's just a calculator that's the next word that's the next word
729. 1:46:31that's the it's got no awareness of what it actually wrote so when it writes something wrong if it's got too much
730. 1:46:37crap in its Laten space it'll just write in [ __ ] confidently it's like you asked me for this this is the next most probable
731. 1:46:47[ __ ] you're like that's not even close that's why it hallucinates so anyway hopefully that explains it and uh that's
732. 1:46:56a good that's a good thing to end on let's see here I'll pop up these URLs here at the end so you can see those I'll just
733. 1:47:06scooch out of the way let me see if there's any comments here thanks I needed to needed a pitch to reach
734. 1:47:18people as chat GPD to write a sherian on it for me nice elevator pitch for a very tall building yeah that was that was not
735. 1:47:26an elevator pitch I failed on that one but whatever sounds like most English food according to my non an English partner that's
736. 1:47:36hilarious man the first time I went to London I'm like what the [ __ ] is with the food cuz I was living in New York at the time I was like
737. 1:47:45really like people pay for this that's that's exactly my diet all right you're a Fun Bunch all right listen I
738. 1:47:55got to run um I hope this was helpful it was a pleasure have a great Friday night peace out everybody I'll think about a
739. 1:48:02workshop if you would you know just pop in what what you'd like me to do it on I'll think about that I'll figure out a
740. 1:48:08way to do it so I don't know maybe I'll do like a four- week thing or I I don't know I will think about it it's a good
741. 1:48:15idea so appreciate it have a good one everybody