aiLL May 1, 2023 - AI is Accelerating Faster Than We Can Keep Up

Transcript

1. 0:00foreign with the AI learning lab let's see notifying viewers that I am live
2. 0:15let's see there's a person hello Kyle Shannon with the AI learning lab logging in late at night well not that
3. 0:24late it's nine o'clock here in Denver hey everybody Welcome in if you have any questions about AI jump in ask me I'm
4. 0:31happy to answer anything I can uh if I can't answer it I'll tell you I don't I try not to make crap up and I don't know
5. 0:39what's going on with my hair it's a little puffy that's a bit of a nightmare so I apologize for that but what are you gonna do
6. 0:47what are you gonna hey at least I got hair right you know at this age I mean my forehead is it looks like I'm getting
7. 0:53smarter because my head's getting bigger but uh anyway yeah so uh let's see what's today today's Monday
8. 1:01um there was some big news item what was it let me go see if I can find there was an interesting news item
9. 1:07that I just saw where did I put that ask any questions get any questions about AI feel free to dump them in the
10. 1:16chat and I will answer what I can let's see what did I oh I know they did lamini a-i-l-a-m-i-n-i dot a i
11. 1:33this thing here and apparently it's really good at training your own data which I think I think that's going
12. 1:42to be I think that's going to be the real big the thing that's gonna really bring these large language models into
13. 1:51their power is when we can personally train them on our own data and then as businesses we can train them on our own data data
14. 2:00um have I have I done anything interesting with agent GPT I have not um I've played around with it a little bit uh you know kind of
15. 2:11hinting it figuring out how to get it up and running but I just I just haven't things have been busy at work because I
16. 2:18have Incorporated AI into my company vision and we've been presenting that and that's been going really really well
17. 2:26where clients are like we want to talk about this I've been really busy doing that um but I think I think all of the
18. 2:32autonomous agent stuff is going to be huge um and uh generative agents are really really interesting and then things like
19. 2:41Auto GPT agent gbt baby AGI all of those things well I mean what's amazing is those tools are um about a month old I mean you know in
20. 2:55their in their entirety um and and they're they're already having a dramatic impact um so you know I can only imagine six
21. 3:03months or a year from now where they've actually got those things dialed in and and you know they don't they don't stall
22. 3:08out and they don't go into recursive Loops that just end up running forever um if you have any questions about AI
23. 3:14feel free to join hey everybody it's uh Kyle Shannon with the AI learning lab uh I like talking about AI stuff because
24. 3:21it's all I can think about anymore and I am insufferable people ask me about the weather and I'm like well you know what chat gbt would
25. 3:30say about that uh but but it's you know the reason I'm so passionate about this stuff is I I think this is um
26. 3:40I think it's one of the more profound technical advances in humanity um I was just too early to be part of the
27. 3:48PC Revolution I was right in the middle of the web Revolution and did a lot of uh innovating during the mid 90s and early 2000s and
28. 3:58this feels a lot like those days um but it is happening significantly faster and way more profound in fact one
29. 4:08of the stats I've said this a couple of times on the channel and even on these lives but but I think it's it's quite
30. 4:15important the the World Wide Web to get to 100 million users it took eight years so I started one of the first digital
31. 4:23agencies in 1995 I was in it for um hi larisha larisha that I work with I guess is in my life she's sending me gifts
32. 4:34hi Larry [ __ ] that's really funny um I started one of the first digital agencies we grew really really big
33. 4:42really really fast I was only in uh it was a company called agency.com I was only in a part of agency.com for seven
34. 4:48years it took eight years for the World Wide Web to get to 100 million users it took chat GPT um a month and a half six weeks to get
35. 4:58to 100 million users so the the speed of adoption is uh absolutely insane and oh I got a Rose thank you larisha for my rose
36. 5:09I got one rose I don't I don't know I've this little my third lives I don't know how any of this crap works but if you
37. 5:15want to send me things I guess you can send me things hi Tim if you're watching uh let's see which AI tools am I excited about at the
38. 5:25moment I learned love to learn more about Auto GPT yeah if you go to well there's there's lots of videos um
39. 5:35there's lots of videos on YouTube about Auto GPT there's one guy in particular I can't remember his name hang on auto
40. 5:45GPT typing in public is always stressful Larissa and I talk about this all the time because we're always both typing
41. 5:53live with people watching us make typos um let's see which dude is the dude that I like see if I can find him e let's see auto GPT
42. 6:06tutorial or maybe it was a lang chain tutorial because uh Lang chain is within Auto gpt2 tour y'all uh is this the guy that's appy Dave he's
43. 6:20pretty good that's not who I'm thinking of is this the guy um rabbit metrics no
44. 6:36well this is boring watching me trying to find a video but uh anyway um the tools I'm excited about um I just saw a thing about mid-journey
45. 6:48more of a mid Journey guy love me some art I um I yeah mid Journey for me is the is the the top of the the top of the Heap when it comes to
46. 6:58image generation I did um stable diffusion for a long time with my my art project and I like I like stable diffusion but mid journey is just
47. 7:06consistently really good art um it's a bit of a pain in the ass because you have to do it through Discord but I'm I'm sure within the next
48. 7:13year they're going to have a web interface or an app or something like that so mid journey is where it's at for me there um
49. 7:21what am I excited about I mean I mean you know quite frankly just the more I use um chat GPT gpt4 um just any of the large language models
50. 7:32but gpt4 in particular is you know absolutely mind-blowing it continues to be mind-blowing for me um so so like that's the thing I'm most
51. 7:42excited about so if you you don't need to get more fancy than that I think there's also um I had a a question on the on the channel today
52. 7:52um that I made a video about about you know someone's like where do you find chat EPT when I search for it like
53. 7:57thousands of apps come up there's all sorts of crappy apps that um are using gpt3 under the hood um that are just you know trying to jump
54. 8:07on the Gold Rush of people talking about chat GPT so they're trying to get you to you know Fork over 10 bucks for the
55. 8:12shitty piece of software so just go to chat.openai.com um or go to bing.com the chat in bing.com is actually egpt4 it's based on
56. 8:23gpt4 and it's connected to the internet so both of those are free and you can pay for um chatgpts plus that gives you access to
57. 8:32to gpt4 which I do but um you get access uh did I get access to the plugins for for chat EBT yeah I did um I haven't played with them a lot
58. 8:43because I just haven't had time to go build one um I have started to have chat cheapyt um create a tutorial for me to teach me
59. 8:53how to create a chat GPT plugin I just pasted the documentation into chatgpt and told it to create a step-by-step
60. 9:01plan for me so I'm about halfway through that um they're very very limited right now they're an alpha um I've heard that they're they're
61. 9:09rolling out access a bit more um but um the the one that I'm excited about I don't have access to and that's
62. 9:15the the one that's connected to the web um where you can go off and search the web so I think that that that that I'm excited about
63. 9:24um I think the plug-in architecture is going to be huge like I think this is totally a game changer it's just really
64. 9:29really early right now so they're very raw and and not a lot of interesting stuff that I'm actually interested in
65. 9:35for work uh more of a mid Journey guy I love me some art encourage your viewers to share your live and invite more
66. 9:43friends okay hey if you're viewing um share this live and invite your friends all right thank you thank you
67. 9:50for oh wait I think that was from I think that's just a I think that's the robot telling me how to how to do this whatever
68. 9:57um will I will AI takeover software engineering jobs yes um absolutely if you you are not using copilot on GitHub you should but but
69. 10:09um it's absolutely transforming uh it's transforming programming um in in lots of lots of ways um and and you know I think that
70. 10:24I think the people that have deep engineering experience are are absolutely you know always going to be
71. 10:30needed but but I almost feel like it's like the lower level tactical stuff is just going to get automated out so I
72. 10:36would say if you're an engineer get good at product get good at sort of the big picture and strategy um I I think that's that's where
73. 10:44ultimately we're gonna we're gonna live I think a lot of the the Tactical stuff's gonna get automated out um which
74. 10:50for for me as a sort of hacky non-technical guy that pretends I know what I'm doing with software I'm really
75. 10:57excited about it real Engineers there's there's a lot of change coming um let's see I can pin a comment
76. 11:06what's that say I can pin a comment oh let's just uh Tick Tock telling me what to do will AI take over I already read that one um myself
77. 11:19have I seen interesting use cases for generative AI working with robotic process automation no but I haven't been
78. 11:27looking for it it's not a world that I know I know creative and I know marketing um I I don't know the the automation
79. 11:34world and Manufacturing and things like that um but but you know Tesla's got their humanoid robot Boston Dynamics been
80. 11:43doing it for a decade um I saw an interview I think it was Lex Friedman interviewed the CEO of Boston Dynamics and
81. 11:51um he said that there are a lot of robotic startups right now so I think I think large language models and robots are gonna
82. 11:59do this very very quickly and I think I think we're going to see advances um I think we're going to see pretty dramatic advances in uh
83. 12:08in robotics so yeah uh with the strike coming any thoughts of the future of screenwriting yeah I I've written seven screenplays I know I
84. 12:18know this world pretty well um I think that the um the uh the writer's Guild just did something about trying to block
85. 12:31um any GPT generated stuff I think that's very very short-sighted um GPT is going to be built into every
86. 12:41single writing tool out there if it's not already it's in notion it's going to be an office 365. it's it's likely going to be in the
87. 12:49in the in the um in screenwriting software so I think I think just even being able to figure out where the
88. 12:56boundary is between you know what lines were created with GPT I just think it's I think it's a pointless exercise right now um you know
89. 13:09we we here's the here's the problem with with any of the creative work creative arts and any of any of knowledge work
90. 13:18is that we are moving from a relative world of scarcity when it comes to creative output and knowledge output
91. 13:28um to a world of abundance that that there's just going to be tremendous amounts of content and whether it's completely
92. 13:37you know autonomously generated or completely human or some hybrid in between we're we're going to be able to
93. 13:46tell less and less that there's that boundary there right now because chat GPT is the separate tool it's you know I I feel like we're in a
94. 13:55very very short-term you know block the block the robots the robots are coming for our jobs I I don't feel that at all I feel that
95. 14:05these tools are um Talent amplifiers right so if you're really good at screenwriting if you're really good at storytelling
96. 14:14and you learn to use AI tools you're going to be really good um you're going to have your talent Amplified using that tool so I think I
97. 14:21think the the quality um and um and quantity of content is going to go up it does make it a lot easier for you
98. 14:32know completely non-talented people to get garbage into the marketplace so I think it's going to get noisier but you
99. 14:40see this with with art you see this with mid-journey and with image generation tools that um really talented people that learn to
100. 14:48use these tools tend to make way more interesting work than less talented people right um so uh you know and I actually think
101. 14:58writers you know anyone who's got the ability to synthesize ideas generate ideas are going to be really good at telling these tools
102. 15:06um how to give them what they want um ask any questions you want what are my macros I don't have any macros I
103. 15:14don't really use any macros are you a software engineer no I'm a CEO of a software company I would call myself
104. 15:19more of a product guy than well I'm definitely not a software engineer um I've got an acting degree and a writing
105. 15:26background I started one of the early digital agencies so I did a lot of early HTML work but that's not really
106. 15:32engineering the the we had Engineers that did the actual engineering and I've got a tech team that that develops my
107. 15:38current software though I am the one that um is the product guy right so I have the idea I know what I want it to
108. 15:45do and I know when it sucks and I I think I'm a pain in the ass product guy to my Dev team because you know my
109. 15:53definition of it should work is it should work like it's you know it's that simple if it doesn't work then it sucks
110. 16:00um I talk about that a lot um and ask any questions you've got about AI I'm happy to talk about them and what else I'm trying to think what
111. 16:11else is going on I haven't seen a ton of news over the weekend because I was busy kind of recuperating from a a last
112. 16:19minute trip to New York to go out to see a client who got very excited about the new AI stuff we rolled into my company
113. 16:26software my company is called storyvine and it's got um what do we have we have an automated uh video storytelling platform so people
114. 16:34answer questions in an app it goes up to the cloud a few minutes later you've got a fully edited video and so we are
115. 16:41rolling a lot of AI functionality into that and and actually I'm completely rethinking the platform and what it is
116. 16:49and what it can be and what what it can do and it seems to be resonating people seem to be getting very excited about it so that's cool
117. 16:57um let's see any opportunities to govern AI I think I think that governing AI is is I I think there's going to be massive
118. 17:09Industries built around it I think it's absolutely required I think we're entering a massive [ __ ] show where where
119. 17:17governance uh you know open AI is sort of trying to govern it and and sort of slow roll their releases but you
120. 17:27know last week or the week before four different open source large language models launched and metas alpaca or metas llama
121. 17:37um models all got leaked so so those are out in the wild and alpacas based on llama and so there's a bunch of it's
122. 17:45it's the it's absolutely a wild west right now um the the one company in this space that I I think is really interesting is
123. 17:53a bunch of X open AI Engineers that made um GPT three um started a company called anthropic an anthropic has what's called a
124. 18:05constitutional learning model which so so the way open AI works is they've got this large language model and then they
125. 18:11use human reinforcement learning so humans look at an answer and they say thumbs up or thumbs down and then the
126. 18:19machine learns which answer was better right and if you do that over tens of thousands of people it gets really smart
127. 18:26really quick it's one of the things they found is you don't need to train these large language models a lot before they
128. 18:31get dramatically better um so that's opening eyes thing um the the challenge with human reinforcement learning is that it's very subjective
129. 18:38and all the biases human biases um get built built into the into the reinforcement so so you can reinforce bad behavior anthropic
130. 18:48their model is called the Constitutional learning model and and the way their system works is they write a
131. 18:54constitution so they write a set of rules so first do no evil and you know do good for Humanity and whatever
132. 19:00whatever the rules are and if you go to the anthropic site I think they've got six or seven of them up there
133. 19:06um and then and then the way the system works is you ask it something it generates an answer and then it Compares
134. 19:12that answer against the Constitution and if it if it ranks poorly it'll send it back um and say no that violates something in
135. 19:20the Constitution and it'll it'll keep iterating until it passes the test and then it'll give you the answer
136. 19:25um I think I think that approach is is really really interesting um you know it's it's probably not perfect right now I think what we'll see
137. 19:33in the future is we'll see um there's not going to be a single governance model that works for everyone so I think what you'll see is you'll
138. 19:43have lots of different models and you'll be able to choose from uh here's a model that aligns with my you know values and
139. 19:49I'm going to use that one and here's one that doesn't align with my values and I'm not going to use that one I think I
140. 19:54think that's how we'll start to see these things the corporate ones are going to be very boring and then there's going to be wild
141. 19:59and zany ones and then there's going to be ones that are actually funny and ones that are actually smart and interesting
142. 20:04and ones that are actually dangerous and you know things like that all right let me see what other questions are here
143. 20:10um what are the best wait sorry lost it what are the best AI software for images that progress into other images like if you can mix images
144. 20:23um mid-journey right now does really good um mixing of images and stable diffusion with dream Booth you
145. 20:33can train it with your models there's also a piece of software called leonardo.ai um that I really like a lot I think it's
146. 20:41still in beta testing I have access to the beta but let me see here the oh Nardo um this one's really good login
147. 21:00watch as I log in to a piece of software don't you hate it when people try to show you something in Tech and they're
148. 21:07like oh it's going to be really good and then it takes them forever to log in like it is for me right now let's see
149. 21:15AI image generation so this is one that um oh yeah that's some disturbing stuff I did let's see um 70s muscle here I'll flip just like this
150. 21:3170s muscle car in abandoned Factory let's see how this does
151. 21:51[Music] yeah that's pretty good yeah so so Leonardo is pretty good and you can uh oh the reason the person asked was um how do I go back
152. 22:06was uh you can upload an image so right here there's image to image so you can upload an image and then base something that
153. 22:17you do on that image so um so most of the most of the image generation things are going to have um are going to have the ability to mix
154. 22:26images c x a i is what I have no idea if anyone knows let me know thoughts on auto GPT yeah Auto GPT agent GPT baby AGI all of these
155. 22:41autonomous agents are going to be huge they're also all um incredibly new and Incredibly raw and um there's kind of good news and bad news
156. 22:56right now the good news is because of the popularity of chat GPT the development Community is bouncing on anything AI so they're developing
157. 23:07all sorts of new Innovative things these autonomous agents are one of them they're essentially a month old so even
158. 23:16though they can do really remarkable things so so what they are if you don't know what an autonomous agent is is
159. 23:22rather than you having to prompt a large language model and say give me this specific answer and then it'll give you that
160. 23:28you basically just give it a goal you say I want you to go find the taglines of of my 10 biggest competitors and it
161. 23:36will create a list of tasks and prompt itself it will send its own prompts to either a large language model like like
162. 23:44chat gbt or it will go to a Google search or whatever it needs to do and so it will say okay you know who are the
163. 23:51top 10 competitors for storyvine and it'll go out and find them and then it'll go find their tag lines and then
164. 23:58it'll you know build that into a table using a large language model um they can obviously do stuff that's way more sophisticated than that
165. 24:07um so they're quite quite powerful and so people are using them to do things like create website Builders where you
166. 24:13literally just say make me a website for this kind of business and it will go and it will write the copy it'll generate
167. 24:19the code it'll go generate images for the design and package it all up and launch it and you know potentially even
168. 24:27go grab you a domain I you know it'll do all sorts they do all sorts of crazy things because they're so raw and so new
169. 24:36they're really janky they're really just bad software because they're brand new um but but what what it's doing is it's
170. 24:45stitching together a series of tools in a way that you don't have to prompt it you just give it a goal and turn it
171. 24:51loose right so that's terrifying also because you know imagine um if if you wanted to say you know go find me a bunch of servers that have a
172. 25:02particular um you know hack exploit and go exploit them and then send me all the URLs where I have root access
173. 25:12you know they'll just go off and do that so um but they're going to be amazing they're going to be amazing there's
174. 25:18another thing that came out last week called generative agents or the week before and generative agents the the paper
175. 25:25talks about like the little NPC players the non-player characters and games and these NPC characters use large language
176. 25:32models to actually talk to one another some sort of you know autonomous agent where you'll will each have our own
177. 25:38agents and we'll just tell the agent go do some stuff for me and it will go do it and I could see a world where you
178. 25:45know I'll have my agent talk to your agent and our agents will go off and you know coordinate things and figure out if
179. 25:51we've got interest or availability or whatever it might be and just do stuff for us um and they'll get to know our
180. 25:57preferences and it's it's going to get very very interesting thoughts about copyright and AI every image is unique and and
181. 26:04irreproducible if you don't know the seed yeah so um yeah the copyright and trademark issue the the the only thing that I can predict
182. 26:16right now is that the lawyers are going to make a [ __ ] fortune in the next 20 years that's just gonna make a fortune
183. 26:24um the lawyers are really excited about copyright and trademark right now because it's such a mess um it it's I it's gonna get better you've
184. 26:34got companies like like Adobe jumped in into the image generation thing recently with firefly and and you know they
185. 26:43trained their model on open source public donate a public domain images um and so the idea with with Adobe is
186. 26:52that they're going to make it friendly for artists where artists can opt out of the training set they can opt into the
187. 26:57training set they're probably going to have compensation models I think that we're going to see compensation models
188. 27:03um where if like I'm a photographer so I could potentially generate um uh you know training sets for my
189. 27:10photographs and then officially license them so if someone liked my photographs and they want to generate images that
190. 27:16look like my sky photos they could do that and I would get paid on the generation of the image rather than the
191. 27:21usage right now all the copyright stuff lives on the usage side and I think that's going to be a train wreck because
192. 27:26of like all the synthesized music we see coming and all the synthesized images we see coming I just think it's going to be
193. 27:33a morass for a lot of years but but I think I think probably within the next two to three years um there'll be systems out there where
194. 27:42where attribution and compensation are going to be handled um I also you know copyright I mean this is all it's already not what
195. 27:57it was intended to be I mean copyright was intended to protect people when producing work was really expensive and really slow and
196. 28:04required an incredible amount of technical expertise to actually make it happen um that got very much watered down when
197. 28:14we moved into a digital world when we went from analog to digital and then it got more watered down when we when we
198. 28:19moved to the internet where you could just immediately distribute things and now I think I think kind of the the the
199. 28:26nail in the coffin to to copyright being really relevant um is is um this AI stuff I think it's just gonna
200. 28:35it's gonna blow it out of the water like I don't know what it looks like I understand that people need to get paid
201. 28:40I understand that you you might want to protect something really amazing that you put together um I just don't know how you protect it
202. 28:48I don't know how you're protected in a world where where um you can generate anything instantly if you look at some of the songs that
203. 28:54have been put out like the the one that was put out that synthesized Drake and the weekend their voice and
204. 29:02um it was an AI synthesized song and it got like you know 600 000 or a million or however many views on Spotify a lot before they took
205. 29:10it down I mean you know we're gonna start seeing fan generated art that Rivals the original artist um and they don't care if they get paid
206. 29:19for that they're just doing they're just putting stuff out there so I don't know how to compete in that world so um
207. 29:25so yeah my thoughts are the lawyers are gonna get rich and it will get figured out but it's going to be an absolute mess for 20 years um
208. 29:33what are some of the predictable industries that may come as a result of AI technology um I think we're gonna see a lot of
209. 29:48incredibly powerful companies that historically would have taken a hundred million or 200 million or 300 million dollars to Launch
210. 30:00will get launched by two or three people you know over a couple of weeks I think incredibly sophisticated disruptive
211. 30:10companies are are just gonna like spring up out of nowhere I I also think there's there's a lot of talk right now about AI
212. 30:17is going to take jobs and there's going to be all this disruption in the in the workforce I absolutely believe that that
213. 30:22to be the case I think I think it's going to be a bloodbath but a lot of those displaced people also have access to the tools that displace
214. 30:32them right so what you're going to see is there's going to be a lot of companies that Embrace Ai and do it right there's going
215. 30:39to be a lot of companies that resist Ai and they're going to get hit like a tsunami with companies that just come in and and
216. 30:46absolutely obliterate them I think a lot of those companies that come in and obliterate the incumbents are going to
217. 30:52be started by people that got displaced out of those those bigger companies I think I think the agency holding companies are in real trouble
218. 31:00um not not because you know well they're in trouble for a lot of reasons but but they they can do good work
219. 31:08um but I think how they charge and the amount of time they take um I don't think I don't think they're gonna last that long I think they're
220. 31:16gonna get innovated out really quick if they don't if they don't disrupt themselves um what are some of the predictable
221. 31:24Industries um I think there's going to be a lot of Industries around AI safety I think there's going to be a lot of
222. 31:30Industries around identity and authenticity um so for example my company storyvine we we have a technology that allows
223. 31:39people to share their stories right so most of the content produced on our platform is authentic human beings sharing their stories
224. 31:47um the the synthesized human thing right now you can still tell they're synthesized but I would say give it a
225. 31:53year and it's going to be tough to tell if someone is real or not um and so I could very easily see us
226. 32:01doing you know sort of an authenticated maybe it's blockchain verified that this is a real human and I think authenticity
227. 32:08is going to go up so I think identity and authenticity there's going to be lots of stuff there I think security is
228. 32:15going to be a massive industry that comes out of this because the hackers have way more powerful tools as do the
229. 32:22the white hat hackers right the the ones battling them so I think those battles are going to get really interesting um boy I I think
230. 32:34any knowledge work I think consultancies are going to face real struggle and real change any knowledge work I think any
231. 32:41knowledge work I think any of the creative Industries I think Hollywood's gonna get upended and and they're
232. 32:48fighting it like that I was talking before someone asked about the screenwriters Guild and the writer's
233. 32:51thing and they're trying to block the use of GPT which I think is really short-sighted but um I I think I think the creative arts
234. 32:59and the knowledge you know any businesses that rely on on knowledge workers are going to be seriously upended
235. 33:08um Blue Willow yeah someone asked about uh image Technologies blue Willows one that's it's a Discord based thing like
236. 33:15mid Journey but it's free I think it's still free um yeah mine has been getting stuck in a do do nothing Loop
237. 33:23um not too consistent that's I assume that's in reference to Auto GPT like I said it's very very early there's not a
238. 33:29lot of backstops in there if it gets into a bad Loop uh it just loops and they they give you very bad answers I've had
239. 33:37I've tried to do I don't know I've maybe messed around with it 20 times and I think I've had two successful
240. 33:44completions so two out of 20 10 percent it's probably lower than that um how do I feel wait sorry things are scrolling too fast told
241. 33:55how do I feel about data security in an Enterprise environment well right now well so uh two days ago
242. 34:02um chat GPT announced that you can turn off um sharing of your data and your history for their training purposes so that's
243. 34:11good you will lose your history if you do that though so make sure you export your data so you have your history if
244. 34:16you've got any good good prompts in there um openai has also announced I assume sitting on Microsoft Azure that that you
245. 34:27can do Standalone servers um so I think I think Enterprise Securities major we've got um Pharma clients are our primary
246. 34:35clients at storyvine so this is a really big concern especially as we're talking about adding generative AI into our platform
247. 34:43um so so we will likely do Standalone servers that are that are effectively hermetically sealed probably for each
248. 34:49client but we might just end up doing a standalone uh thing for our company and then just uh you know assure security um but but I think I
249. 34:59think data security in an Enterprise environment's huge huge Samsung just got a bunch of stuff leaked because
250. 35:08um people put meeting notes and some proprietary code in a chat GPT chat someone hacked their history and got a
251. 35:15bunch of that stuff so someone got fired there so I think it's huge um what about AI content creator accounts we probably won't need real
252. 35:24people to create content anymore well to create the content no but to instantiate the content to come up with
253. 35:33the idea to come up with a strategy to for every image that you see that someone shared on Twitter there's
254. 35:40probably 50 images they created that weren't good enough to make the cut so curating content refining the content
255. 35:46you know you know combining different images together either in Photoshop or illustrator or whatever it might be adding type to it um
256. 35:55my experience with these tools you know on the surface you say oh these tools these tools can do creative work really well
257. 36:04they're going to replace all the creative people my experience has been I'm finding myself more creatively alive than um
258. 36:14then I've been in a long long time because what the tools do is they do the low level creative work that always got in
259. 36:22my way before so I would have an idea and I'd say oh I want to go do this creative idea and then I'd say but I
260. 36:27need a piece of software that does that and then I'd Google that and then I'd find you know two pieces of shitty software
261. 36:34that were just ads and then I'd finally find a piece of software and I'd go to install it and it wouldn't install right
262. 36:39then it would be buggy and then I'd go to YouTube try to figure out how to use it and 45 minutes later I don't even
263. 36:44know what the idea was anymore what's happening with AI is I have an idea and then I can articulate it with
264. 36:50chat GPT and I can visualize it with mid-journey or whatever tool I'm using like like that and and that gives me new
265. 36:57ideas and so I I'm finding myself being way more creatively alive than I was uh before it so so I I think I think we're
266. 37:06actually entering a great Renaissance I think we're we're entering an age of incredible creative expression but it
267. 37:15is going to be done through these tools it's going to be creative people using these super powers to create truly
268. 37:21remarkable stuff so that's that's my my two cents on the creative thing um building fintech and want to use Ai
269. 37:29and ml to prevent fraud and money laundering any advice uh yes go do that start testing it I mean the thing about um
270. 37:38large language models that the the the the biggest shift that my sorry inarticulate um the big difference is historically if
271. 37:53you wanted to do something with you know fraud and money laundering prevention you would have to explicitly program the
272. 38:02logic to say you know here are all the kinds of money laundering schemes we might see and hear all the you know the
273. 38:10fraud things we might see and let's write all these exceptions and you have to explicitly write the logic and what
274. 38:17happens with these large language models is you can say act like a you know money laundering you know fed
275. 38:26um who's expert at identifying money laundering schemes and go find the schemes and now give me 20 ideas for how
276. 38:33to prevent those schemes now write the code to find the schemes prevent them and send me a report daily and by the
277. 38:41way I'm you know I'm gonna run a Cron job that we're gonna we're gonna we're gonna update that thing every 24 hours
278. 38:50right so I think your ability to iterate um features within the company you're trying to create accelerate dramatically
279. 39:03your fraudsters and your money launderers also have access to these tools so they're going to come up with
280. 39:09increasingly complicated schemes so so you know you'll have your fraud prevention agent talking to their fraud
281. 39:16agent and and uh you know they'll be a constant War but I think you can iterate very very quickly so my advice would be
282. 39:24just start playing start playing with the large language models um really explore navigating between conceptual things
283. 39:33like what are the kinds of schemes that I might find out there and then how do I code against those and let the tools
284. 39:40help you come up with those and give those to your engineers sorry late to the party can I tell us can you tell us
285. 39:47what your background is how you got to where you are today yes I have a degree in acting I moved to New York New York City out of
286. 39:54school started and ran a theater company for four and a half years uh wrote seven screenplays in a one-man show after that
287. 40:02and then in the mid 90s um I was doing desktop publishing as my day job and um I stumbled on this thing called the
288. 40:10world wide web and I thought Hmm this thing's interesting and I had an idea for an online magazine and started one
289. 40:17of the first digital magazines a Zine called Urban desires and um and then from that met a guy that I ended up co-founding a company called
290. 40:28agency.com with which was one of the first digital agencies and so built a lot of the first websites for the Fortune 500 so one of the
291. 40:35reasons I started my Tick Tock Channel um at the ripe old age of 57 that I am uh last thing on the planet I thought I'd
292. 40:45do start a tick tock Channel um but I think that what's happening with AI is akin to to [Music] um what I discovered with the World Wide
293. 40:56Web that that it's going to change everything and I felt that in 94 when I discovered the web and I feel that way more profoundly now
294. 41:04I think that I think the changes that are coming with AI are um significantly um more profound and they're coming way way way way way faster
295. 41:15um than any of the any of the internet stuff that that happened over the past 30 years um the past decade I've been founder and
296. 41:24CEO of a company called storyvine which is an automated video storytelling platform and then I would say the last
297. 41:29year I've been pretty active in AI stuff and in the past six months I've just been absolutely obsessed with it I cannot
298. 41:39um I cannot uh I can't get my head around how fast it's moving and how profound this stuff is which is why I'm
299. 41:46doing this account and even when I'm doing these lives I don't know why I'm doing a live um let's see AI to grammarly
300. 41:55um to remove AI detection yeah again I think all this AI detection [ __ ] is going to be [ __ ] these large
301. 42:01language models are gonna literally be everywhere I think this is a very short window where everyone's panicking and
302. 42:07they have their panties and a Twist about yeah it was written by a robot it's not real writing and there's some
303. 42:13truth to that but these tools are just going to be like having a spell check or having grammarly you know plug-in in
304. 42:20your writing where it checks your you know these things that check your spelling and check your um your grammar
305. 42:26Rai right so the generative AI stuff is just a slight variation on the theme um these tools are going to be be everywhere
306. 42:34um so you will likely always be able to find a tool that will be able to skirt around AI detection but I don't think I
307. 42:42don't think AI detection you know whether or not something was generated with AI is going to be all that relevant
308. 42:48and issue a year from now I just don't I you know I think this is going to be ubiquitous this is it's going to be just
309. 42:54like having a word processor or a calculator right back when calculators were invented everyone freaked out about
310. 43:00no one's going to need to know math anymore because I got these fancy machines what are we gonna do so all right uh image generation and AI
311. 43:10takes a lot more skill if you're trying to create something specific exactly so so the idea that it's one of the one of
312. 43:17the my pet peeves is that people talk about these machines like they're just behaving on their own Auto GPT will just
313. 43:24behave on its own so I guess I should be a little careful where I tread here but but um a creative act is an intentional act
314. 43:34right you intend to create something specific you know for a project or or for a goal or I mean you might just be creating stuff
315. 43:42for for fun and that's fine um but yeah those people that can say hey I want an mg speeding down a road you know in the in the highlands of
316. 43:53Scotland um and they can turn that around quickly those people are going to be more valuable so absolutely I think there's absolutely skill
317. 44:01um you see it all the time you see people put stuff on Twitter that's just all the same and then you see someone
318. 44:06else that does something you're like oh my god well that probably took them you know a month of sort of you know
319. 44:14thousands of images to kind of get to a look and feel and figure out how to do that I know I know someone that I that I
320. 44:20work with that is brilliant at this stuff it took him a month or two to figure it out um what are the top three threats to AI
321. 44:30um an EMF would be a big one if all of our computers stopped working it would shut it down pretty quick um I would have said
322. 44:40you know prior to um prior to chat GPT I would have said something about government intervention or or monopolization
323. 44:50but I think I think the fact that chat CPT went to 100 million users in a month in in a month and a half
324. 44:58um has forced a lot of companies out of the woodwork and it's forced a lot of Technology into open source and even you
325. 45:05know stability just released their their large language model open source and their image model is open source I have
326. 45:12a feeling stability is doing that because they probably have an inkling they're getting the pants suit off them by Getty
327. 45:19um for their for their training models so if they don't settle that that could put them out of business so I have a
328. 45:24feeling companies like stability understand that that they might not survive litigation and so I think they're putting things
329. 45:35out in the open source to get them out there and you know the fact that meta's large language model leaked and is out
330. 45:42in out in the open now it's in the development Community um the fact that there were four large language models
331. 45:48um released last week the fact that some of these image models are open source um I don't think that governments are
332. 45:56going to be able to get their hand around it their hands really around this stuff um it's possible but I don't see that as
333. 46:04a threat I see a bigger threat being government incompetence um kind of regulating the wrong things um and and you know just doing stupid
334. 46:14[ __ ] like you know you can't use chat tpd to write um okay what are some of the other threats I mean if
335. 46:29there's enough of a negative kind of panic out there because of Hollywood that you know AI is gonna destroy Humanity
336. 46:39that you know if someone takes one of these Auto GPT things and kind of turns it loose in the right way and someone
337. 46:47ends up you know dying or you know people end up dying because of it I could see there being some kind of panic
338. 46:52where where a lot of this stuff gets locked down temporarily but I just I don't think it's gonna stick so um
339. 47:07I don't know I don't know I where where I'm trying to live is in the place that any individual who gets curious and
340. 47:18explores AI right now is going to have a disproportionately positive impact on AI moving forward and so I'm trying to get
341. 47:27as many people up to speed and enthusiastic about trying it and at least understanding it right now is
342. 47:33possible because I think it's going to precipitate lots and lots of change and if you're on the inside of it using it um
343. 47:40you can help precipitate that change in a positive way rather than just a destructive way right if we just leave
344. 47:45it up to the capitalists they're just going to cut staff there's another model that says hey with the same staff we can
345. 47:52do way more work or way better work with the same amount of staff so there's there's lots of different ways to
346. 47:58approach these tools so I don't know if that answered your question um you articulated Tech frustration for
347. 48:06creatives perfectly by the time I figured out the ideas gone yeah exactly exactly that thing where you're like
348. 48:13you've got an idea and then you're just trying to like just jump the hurdles to be able to even get to a like The
349. 48:18Whisper of an articulation of that idea and then it's gone so yeah absolutely that's that's where I mean these these
350. 48:26generative AI tools are just a [ __ ] superpower do I know Howard cash I do not um there's uh it happens don't take it okay hi cheers hello
351. 48:40um age is just a number thank you mine's 57. it's a big one [Music] um thanks for your content interesting background thank you
352. 48:53thanks to Mom and Dad oh I guess that's your your name all right okay it just I'm gonna love it let's see what else
353. 49:01good point about stupid government Regulators yeah I mean like if you watch those Tick Tock things that like the the
354. 49:06the Congress people didn't even know how Wi-Fi worked much less how Tick Tock worked much less how incredibly
355. 49:14transformative powerful AI tools work like it's [ __ ] terrifying we did um the everyday AI newsletter that I do every
356. 49:24week we did it one of the um one of the topics this week was um this military tool where they're using large language models to do
357. 49:34military logistics and it was just absolutely terrifying but I'm sure the military's got you know AI tools that
358. 49:42make what we have look comically um simple let's see they're a very knowledgeable AI developers all of a sudden
359. 49:50um where have they been all this time well I mean so so so listen when the oops when the uh when the World Wide Web
360. 49:58first came onto the scene the internet had been around for a number of decades same with AI ai's been around for a bunch of decades um
361. 50:07a couple of things have happened so so when gpt3 came out um in the documentation of gpt3 were the instructions for how to build chat GPT
362. 50:17and Sam Walton assumed that someone in the development company would build it and they didn't and then
363. 50:23um Microsoft was about to announce that they were going to make this billion dollar investment into open Ai and Sam
364. 50:29Altman said to his Dev team you've got two weeks put together chat GPT so they built that in two weeks so if you feel
365. 50:36like chat GPT is kind of like uh like a thin piece of software it's because it is and you know Sam Walton has said it's it's a bad piece of
366. 50:44software um but but it got to a hundred million users in a month and a half and so what that's meant is that the
367. 50:56entire development Community has just flocked to um this generative AI stuff right so so the tech had been there for a long time
368. 51:05and a lot of developers have been developing things with gpt3 and and you know using a lot of the tools that are
369. 51:11sitting underneath it for for like a year and a half before before chat GPT came out um but but because it got so popular
370. 51:20that's the thing that's generated all this Buzz that's the thing that's forced Google out of the shadows and and sort
371. 51:26of force them to put I mean they're the ones that invented the Transformer that all of these generative Technologies are based on
372. 51:33um and they were sort of keeping it behind the firewall and they were sort of forced out in the open um AWS has now
373. 51:39been forced into the game Apple's got something uh some announcement coming in June that I assume is going to have some AI stuff with it
374. 51:46um you know meta has been forced out of the out of the Shadows with this stuff um so so I think that that the developer Community um is is
375. 51:56um is just pouncing all over it the other thing about these large language models is with a hundred lines of code you can
376. 52:04generate applications that are incredibly sophisticated right so so I think you've got even Junior developers
377. 52:11um are able to whip whip up some pretty remarkable stuff very very quickly so that's where I think they're I mean you
378. 52:17know part of it is just there's there's all this buzz and fervor but but it's because the tools are that good and and
379. 52:24uh there was all this you know visibility because of chat GPT um how is it possible to technically lock it to lock what to to lock
380. 52:34um large language models well you could I I you can lock it if you run a large language model locally on a machine
381. 52:41that's not connected to the internet you can now do that with like with alpaca and and some of the things that are
382. 52:46built on oops my machine went to sleep uh alpaca and some of the some of the things that are based on um llama and and some of these other
383. 52:54open large language models they're they're getting them I forget what they call the process what's it called um
384. 53:02I don't know where they shrink it down not miniaturization but quantization quantumization I don't know um
385. 53:11I heard of one large language model that um they've got down to a five gigabyte install so a five gigabyte install you
386. 53:18can just put that on a laptop and run it so if you're just running a large language model locally there's no
387. 53:23there's no real risk you're not sending your data out to anyone so if that's what you mean lock it down you know you
388. 53:29run it on a virtual machine or you run it locally um let's see please go live again soon okay I will all right
389. 53:39um I've been at this for about an hour now if anyone has any other questions jump in there pop them in
390. 53:45otherwise isn't it kind of frightening that even expert AI researchers don't really understand how AI works yeah it's frightening um
391. 53:55I mean they they understand technically how they work they don't quite understand how these emergent behaviors are happening
392. 54:05so they don't really understand you know they shouldn't be as good as they are right so so that's a little disconcerting but
393. 54:17the thing that the Transformer did the Transformer technology was allowed machine learning neural networks to be
394. 54:25trained on massive data sets and so what you've got with GPT three and four is you've got a tool that is essentially learned
395. 54:35all of the internet like like think about from like the mid 90s until September 2021 think about all of the
396. 54:45content that has been published in Wikipedia and every blog and every non-paywald news site and and every
397. 54:54website anything that can be crawled um has essentially been crawled and it knows all of that right so so when you have that amount of stuff
398. 55:05um even though it's doing something that is you know um probability based um it's it's a massive massive uh data
399. 55:17set that it's learning from and then the other thing that happened which I find this fascinating so they had gpt3 for a while and it was
400. 55:25okay and then they put together this code model I forget what it's called codex I think or something like that what what
401. 55:32what GitHub co-pilot is based on they basically trained up you know these large language models on all of the computer code right
402. 55:39so over here it learned all of the words and over here it learned all of the code and a funny thing happened was once it
403. 55:46learned all of the code it got a lot better at the words now my thought there is that it learned logic with the
404. 55:53programming and it could now take those patterns of logic and apply those to language right because language in a lot
405. 55:59of ways is human programming language right it's you know when we talk to each other it's it's in code there's there's
406. 56:06lots of different things but language is really powerful for us as humans so um so they saw these emergent behaviors
407. 56:13where the language model got better when it got better at programming so um I find it a little scary that I don't
408. 56:21understand or that they don't understand how that actually works um um but but I don't know these tools are
409. 56:29growing so fast they're they're going to very quickly dramatically outstrip our capability as humans to understand them so I think you know
410. 56:42we're lucky to be able to play with them I feel incredibly lucky right now um my comments about the government could
411. 56:53clock the AI Trend I don't think they can I don't think they can either um no I think because everything's open
412. 57:00source now I don't think the government could get their hands around this anymore wait till it's used to hacks on nuclear
413. 57:05code and then see how fat well yeah but so so that's the thing is is like so so even if someone hacks the nuclear code and something truly
414. 57:14horrific happens and the government says we gotta lock this [ __ ] down um the fact that so many of these things
415. 57:20have already been open sourced means that they're open source they're they're out there um they can make them illegal and and
416. 57:27things like that but it's not gonna it's not gonna stop people from using them so um I I think the big commercial
417. 57:34companies are probably going to get sacked with regulations right I think I think open Ai and Microsoft and
418. 57:41whoever's using these things you know Google might get them locked down but the open source stuff they won't be able to lock it down
419. 57:49um is this that Brokeback Mountain [ __ ] I don't know what you're talking about [Music] um [Music] yeah but uh thanks for your hard work
420. 58:03you're welcome um how fast do I think the progress is to AGI I don't know I mean I think that you know there's there's a
421. 58:17Microsoft paper that was released a month or so ago that that says gpt4 shows is showing signs of AGI um signs of it uh it's it's a really
422. 58:28long paper um it's it's worth exploring if you don't have ADD like I do I'll just rely on summaries of others um
423. 58:39the the problem with with the term AGI is that there's not an agreed to definition of what it actually means
424. 58:48um I I think that we're past the Turing test at this point I I some of my interactions with chat GPT um
425. 58:56it's actually the thing that makes it makes it clear that it's not a human is how how fast and how good it is like no
426. 59:03human could be that fast um but but AGI you know artificial general intelligence is is very undefined right now I think this combination of
427. 59:17autonomous agents and memory the ability for these agents to act on their own behalf remember what they did learn from what they did
428. 59:28um I I think it's going to accelerate quite quickly I mean I mean also the fact that so many developers are doing
429. 59:34so much work on this technology it's it's going to come um quite quickly I think that Ray Kurzweil predicted
430. 59:44um 2029 and I think a lot of the AI scientists have now um shortened their prediction to 2030 so Ray Kurtz file has been at 2029 for a
431. 59:56long time the AI scientists are now down to 2030. um I'm guessing it could come faster than that I I think I think we're going
432. 1:00:03to start to see exponential Tech development um what's my next move for a product or service to monetize it given this uh AI
433. 1:00:13technology for me personally it's rolling AI um into my product but doing it in a way that it it dramatically expands what
434. 1:00:24what my product is so I've got a video technology company that that makes it automates video production and allows
435. 1:00:32our clients to to have people tell their stories really powerfully we have an app that guides an end user through telling
436. 1:00:37their story goes up to the cloud a few minutes later you got a fully edited video and with generative AI dramatically
437. 1:00:44expanding the output side of what our company does but also dramatically expanding the input side of it so for me
438. 1:00:50personally it's about rolling this stuff both technically and conceptually into what my company's all about because if I
439. 1:01:00don't the value of what we do now is going to drop to zero very very quickly and it's terrifying um oh this is great so I found myself
440. 1:01:11thanking chat GPT so I find myself thanking it I find myself getting a little guilty if I ask it to do too much
441. 1:01:19like if I ask it to rewrite something for like the sixth time if if it were a human I would just not I wouldn't have
442. 1:01:26asked them the third time to rewrite it you know so um there there's something about um talking to chat GPT as if it's like a
443. 1:01:37creative collaborator um where you where you talk to it like a human and you give it more context um it I think it generates better work
444. 1:01:46um whether or not you actually need to thank it I don't think it means anything to it um but I think if if there's something
445. 1:01:51about the tone of how you write um it's doing predictive you know answering on the other side of it so the more context that you give it
446. 1:02:01um the more it colors what you get back so maybe it gets a little more pleasant in its responses to you as well because
447. 1:02:08it because you're doing that so I do too like I you know I think we anthropomorphize a lot of things anyway
448. 1:02:15we certainly do it with pets right um so I think it's going to be really easy to do it with these tools
449. 1:02:20especially if you see some of the some of the tools that are you know synthesizing you know people's faces and
450. 1:02:26voices and putting a chat GPT behind it and if it does it quickly where you're just talking to someone and it's talking back to you
451. 1:02:33um yeah we're gonna pretty quickly get to the someone asked that before about how close to her do I think we are I think we're probably
452. 1:02:42hell of a lot closer than we were than we were a year ago I mean I would say probably within within three years we'll have robots
453. 1:02:54that are not as weird as they are now and within five we'll probably have ones that are that are pretty remarkable and
454. 1:03:01I think probably within two years we'll have virtual you know on-screen you know like you're chatting with
455. 1:03:08someone on FaceTime but it'll be a bot and it'll just be like talking to someone I think that's probably within
456. 1:03:13two years have I used the Chrome extension web chat GPT there's a bunch of extensions out there um I've used
457. 1:03:20some of them my problem is things are moving so fast I like I installed like every time I install something it shows
458. 1:03:27up as a feature in in some tools so I'm just I'm just trying to slow roll what I install I'm not located in New York I'm an ex
459. 1:03:36New Yorker I miss New York um I will definitely go back to New York but I'm not there right now I'm in
460. 1:03:42Denver it will collapse the stock market I think it's going to have dramatic impact on everything I don't know if
461. 1:03:49it's going to collapse the stock market it could it could um expand it but I don't know AI will be doing the trading I agree with that it
462. 1:03:57to a lot of degree a lot of those speed trading systems are all AI anyway like that's the thing none of this AI stuff is new like ai's been
463. 1:04:06around for decades right what's new is us as real people now have access to these tools that are you know otherworldly powerful um
464. 1:04:20AI to analyze technical analysis for stock trading yeah if you want to if you want to play with that
465. 1:04:25um go to gpt4 3.5 or 4 whatever chat GPT three or four um tell it to recommend a stock trading site that does paper trading right where
466. 1:04:35you can do it without real money I mean if listen if you got lots of money and you want to go blow it do that
467. 1:04:41um but but go to a site that's got paper trading that has Bots and then either go find a bot or two that you like copy and paste the code
468. 1:04:49out of it um put it into chat GPT and have it you know you can have it Mash two Bots together or you can just say hey cheche
469. 1:04:57EBT give me five different strategies for trading and then and then say write me a bot for this particular site you
470. 1:05:03know in the following code and it'll do that and then you can go play with it and try it so yeah I think you can
471. 1:05:09um you can absolutely use AI to do technical analysis um and and you know and you know buy and sell triggers and stop losses and all
472. 1:05:19that sort of happy horse crap I'm horrible at that stuff so I'm not going to touch that oh I would do it with um
473. 1:05:26with paper trading right uh but but with real money it's just it's ridiculous for me to do something like that advice for
474. 1:05:33someone experienced in server development for 20 years next career move I think right now uh you know the
475. 1:05:38if if you're in if you're in server development I mean there's like the need to have service they need to
476. 1:05:48have computers running you know isn't going to go down it's going to go up um just get um get as knowledgeable as you can about
477. 1:05:59this AI stuff just learn as much as you can in your field or in something you're interested in even if it's not in your
478. 1:06:05field just start getting um as educated as you can with it um with things like security and coding I I would I would start digging deep into
479. 1:06:15autonomous agents I think things like server maintenance and and um keeping like devops kind of stuff keeping
480. 1:06:24um libraries up to date all of that stuff is going to go from really complicated hard-coded systems that you
481. 1:06:30either have to pay for or build yourself to just have some version of Auto GPT just go do it for you right or an
482. 1:06:37autonomous agent could do it for you so if you understand those tools and can go build those tools um uh you're gonna you're gonna be in
483. 1:06:45high demand I think um so um just get get going hang on um yeah we'll do the trading I'll be doing the learning it will collapse the
484. 1:06:57stock market let's see I got chat 250 to write a piece with extreme emotion and emojis yeah it'll write with Emojis it's pretty
485. 1:07:05good it'll write tables too if you haven't gotten chat GPD to write a table for you it's pretty pretty amazing
486. 1:07:11um you do realize that the more people that hand over their thinking process to AI the quicker we go no I don't agree
487. 1:07:16with that and I don't I'm not handing over my thinking process to AI I'm handing over the crappy low-level
488. 1:07:23tactical execution [ __ ] that I hate that gets in the way of my creative thinking I'm giving that to AI I'm
489. 1:07:30taking the high-end strategic stuff so I you know any any anyone who it's so easy right now to to give into the robots are going to kill us
490. 1:07:44um I think it's actually a tougher place to go to go to the to the Light Side come to the light come to the light it's going to be okay
491. 1:07:54um I like my new t-shirt thank you oh it's got crap all over it what's on there dinner [Laughter] um thank you for your advice how about
492. 1:08:04digging into the development of chat gbt apis and other type of apis um yeah I should do that I should do
493. 1:08:11some stuff on apis I do I do a lot of work well not a lot of work I've built a number of tools using zapier um doing
494. 1:08:18automations no code tools where I'll take inputs from a form and then um take those inputs roll them into a
495. 1:08:27prompt for chat GPT send it off get the result back and then send out an email so I've built I built a number of
496. 1:08:34content generation tools with that um but yeah I mean listen the apis for these things have been out for years so
497. 1:08:41so a lot of developers have been using them so this isn't as new as it seems um just what's new is our access to it um
498. 1:08:47like you're thinking very professional and pleasant to listen to thank you thanks for your time I've been using a
499. 1:08:53chat gbt to use the API and it works well yeah the apis are clean the apis um for um DaVinci 003 are quicker than chat GPT
500. 1:09:03because jet gbt is overloaded I think that's going to balance out pretty soon I still don't have access to the 4.0 the
501. 1:09:10gpt4 API which is bumming me out but whatever uh um I'd make a great host for an AR AI maker type show thank you very much I
502. 1:09:20appreciate that I'm gonna have to do something with the hair [Laughter] by virtue of asking your questions to AI
503. 1:09:27it begins to analyze human thought process yeah I think I think rather than it's it's not really analyzing anything um it I think
504. 1:09:35I think what these GPT tools are actually doing is reflecting Humanity right it's been trained on most of the
505. 1:09:43crap we've written for the past three or four decades so and and you know way before that right obviously all the
506. 1:09:49historical stuff that's there um the hair is perfect yeah the hair is special the hair looks like I could run a jet ski off it
507. 1:09:58um all right I gotta run thank you all for your time and hanging out and uh yeah we'll talk soon I'll make another video or two but thanks
508. 1:10:07everybody take care