4/1/2026 - Building a Live Simulation of a World Where Work No Longer Defines Us

Transcript

1. 0:00All right, champ. You ready? Are you ready for zone football? Yeah. Good doggy. Good doggy. What's happening? Good doggy.
2. 0:15What is going down?
3. 0:48Come on. Come on.
4. 1:07So here's a quiet town where the old trees sway the town that goes through the day.
5. 1:25Little te little day.
6. 1:46Woohoo.
7. 2:14We got tick tock rising star Danielle in the house.
8. 3:04Girl, you're looking finding night. Every fell got you in his sight. Would
9. 3:20you do it with a clown like me? Surely one of life's little mystery. So tonight I ask
10. 3:36the stars above. How did I ever win your love? What did I do?
11. 3:51What did I say? Turn your angel my way.
12. 4:16Hey.
13. 4:32Woohoo. So, I learned a new song for you tonight.
14. 4:48April fools. Every time I see you now, get that look
15. 5:02in mine. Every time I see your mouth, I hear that smile. In the early misty morning light, I heard the engine turning in the old foot
16. 5:19outside. You were leaving me again today. You will convince me
17. 5:35again today leaving this hotel looking for someone else's golden ring to say
18. 5:49so long. Hush now don't you cry. So long.
19. 6:08Don't you cry for me. Chilling jeans and cigarettes and keeping warm out on the road.
20. 6:25Chasing down a lifestyle out on Highway 24. New York state was rolling breeze in the sunshine and the blue sky falling. The
21. 6:38chill of old September creep you leaving me again today. You convince me
22. 6:54again today. You're leaving this hot tub looking for someone else's golden ring. Should I say
23. 7:08so long? Susanna, don't you cry for me?
24. 7:30Happy Wednesday everybody. Hope you're doing well on this fine Wednesday night. It's rainy here in Denver. A little
25. 7:39drizzly, so it's got a it's got a Portland vibe to it tonight. Um, not like that here. It's normally sunny,
26. 7:48but what are you going to do? We need some rain. We got no snow this winter. None. None. None. No, one little snowstorm.
27. 9:12Oh, by the way, we don't have producer Brandon tonight. So, it's going to be me and me only. And so, for you people on TikTok, just yell
28. 9:28at me if I get it wrong with ye old
29. 9:52Um, let's see here. Is anybody commenting on the YouTube? It is possible. No. Do me a favor. If you're
30. 9:59on YouTube or one of the other desktop platforms, just comment comment so that I can see comments are work working over there. Should
31. 10:16we call the local pope if the black screen needs adjusting? I guess you'll just have to tell me and I will have to
32. 10:22do my best to to make it make it happy. AI really is a bubble. It's too late.
33. 10:38I'm already addicted. Right.
34. 10:58Is this place I can rest my forhead? Gather my thoughts in sweet silence. Is this place where the feelings are
35. 11:15dead? Running over exposure to violence. Is this place I can slowly face. The only one I truly can know.
36. 11:30These are tears from a long time ago. Got these tears from a long time ago. I need to cry. 30 years or so.
37. 11:42What's happening on YouTube over here? Got a lot of rising stars over on Tik Tok.
38. 12:28Boom. Down.
39. 12:43All right, let's get this show rolling, rolling, rolling, rolling. Uh if you're new here, if you haven't been to uh to AI Swan today,
40. 12:59um head over there and go look at the um I made a post in community feed about the Jack Dorsey article. I'd love
41. 13:11to start a conversation over there. Um so let me see. Let me go. Let me reload my page here and then we'll I'll maybe share this.
42. 13:29Let's ask for help. All right. Beautiful. Got a couple comments in there. So let me share my screen.
43. 13:46Let me share my screen. So there you go. So this is the article. So if you go to the AI salon, if you're
44. 13:54new here, don't know what that means. I will show you here shortly. Is this it? No. Is this it? No.
45. 14:12No. Let's say that. Yeah, there you go. Oh, no. That's Festivus updates. Where? Where's my graphic?
46. 14:27Yeah, there we go. Beautiful. Okay, here. Let's You know what I'm going to do? I'm going to move that up so I'm not always looking for it.
47. 14:37Then we're going to rename it AI salon. Wait, but I didn't rename it, did I? Ah,
48. 14:53beautiful. Okay, head over to community.thesalon.ai and black bar got access to the article.
49. 15:07Yeah, I changed I changed the article, Cam the link in there now, but haven't finished reading it yet. Okay, cool.
50. 15:13Yeah, I put the I put the new non uh non-paywalled version into the article. Um, and that should make it easy to find.
51. 15:25So, go over to community feed. It's the second post down there. And you can you can read the article tonight. This is what I read last night
52. 15:34during the live um was this article and I just think it's worth us starting a conversation over on ye old AI salon. Um
53. 15:45because yeah um let's let's dive in. If anyone has anything they want to talk about feel free to pop that in comments.
54. 15:54Get rid of that thing. just commenting Wednesday like what's happening over here on YouTube. Hey, hey, hey. AI really is a bubble black
55. 16:07bar. I know, I know, I know. There we go. All right, we're getting there. We're getting there producer free tonight.
56. 16:16Is anyone going to copy the leaked claw code? Uh, it got copied 42,000 times in three hours. It got forked 42,000 times in three
57. 16:27hours and then someone posted it to the to blockchain. So it will that code will never not be available.
58. 16:36There's an interesting uh video from Greg Brockman talking about the new model. We can watch that. Let's watch that.
59. 16:43Um let me close this. Let me do this. Let me do this. Let me do this. Let me do this. Okay. So it all it all
60. 16:52looks the same to you, but it is completely different. Okay. Black bar. That's not Greg Brockman. That's Leonel Messi. Hello Messi.
61. 17:06How you doing? Okay. Have a very fantastic. All right. Here we go. All right. Look at you go. I know. I'm
62. 17:19self-producing over here. Listen, I self-produced this thing for a year and a half, two years, something like that.
63. 17:24Some crazy amount of time. I have mods. Um, okay. Any mods on Tik Tok?
64. 17:38Feel free. Oh, producer Brandon's here. He snuck in. I think he's here as a as a a participant. So far, we're doing we're doing good,
65. 17:52Brandon. We live on the stinking producer. Look at that. Look at I'm doing good here. Just about the time I I pat myself on
66. 18:07the back. All right, I can help mod. Anyone wants to mod, jump in. Feel free. Just pin anything you think is interesting.
67. 18:20And let's see what Greg Brockman has to say about Chat GBT 5.5. I think it's 5.5. >> Expect to have a very strong model in a
68. 18:28few weeks. This was a few weeks ago and the team believes it can really accelerate the economy and things are
69. 18:34moving faster than many of us expected. So what's a good model? Um but I think that it's really not about any one model. Okay.
70. 18:44Right. The the the way that our development process works is you have pre-training. So you produce a new base
71. 18:50model that then is the foundation that we build further improvements on top of. And that that is always a huge effort
72. 18:57across many people in the company. And that's where I've actually been spending most of my efforts over the past 18
73. 19:02months has been really focused on our GPU infrastructure on supporting the teams that do all of the training
74. 19:08frameworks to scale up at these big runs. But then there's a reinforcement learning process. So you take this AI
75. 19:14that has learned lots of things about the world and it applies that knowledge and then we do a post-training process
76. 19:19where you really say okay now you know how to solve problems you practice it in in all these different contexts and then
77. 19:24here's kind of the last mile of behavior and usability. So I think of spud as a new base as a new pre-train and that we
78. 19:33have had this uh I'd say it's like we have maybe two years worth of research that is coming to fruition in this
79. 19:40model. It's going to be very exciting and I think that the way that the world will experience it is just improved
80. 19:47capabilities and that for me it's never about any one release because as soon as we have this one release it'll be an
81. 19:54early version of what we have coming. We'll do much more of each of these steps of the improvement process. And so
82. 20:00I think that where we're going is it's almost just we have this engine of progress that just moves faster and
83. 20:06faster and that spud is just one step along the way. So what do you think it'll be able to do that today's models
84. 20:12can't? I So I think it's going to be able to solve both much harder problems. I think it will be much more nuanced.
85. 20:19It'll understand instructions better. It'll understand the context much better that there's this thing called big model
86. 20:25smell that people talk about where it's just like there's something about like these models are just actually just much
87. 20:32>> Nobody talks about the big model smell. Nobody. That's an inside joke. Come on, Greg. smarter, much more capable that
88. 20:41they bend to you much more and you feel it, right? When you ask a question and the AI doesn't quite get it, it's always
89. 20:48so disappointing, right? We have to like explain. You're just like, you really should be able to figure this out. And
90. 20:53so I would just think of it as in some ways just qualitatively there will be quantitatively lots of shifts, right? And qualitatively there
91. 21:02will just be new things where you would be frustrated before you never used an AI for it and now you just use it
92. 21:06without without think. How about Greg if this model actually can do writing in a not horrible way? That would be swell.
93. 21:13>> Thank you very much. And I think that that that is what we're going to see across the board. I'm super excited to
94. 21:18see how it raises the ceiling, right? We've already seen these physics applications, things like that. And I
95. 21:23think we will be able to just solve like way more open-ended problems, way longer time horizons. And then also very
96. 21:30excited to see how it raises the floor where just for anything you want to do, it's just so much use more useful for
97. 21:35you. All right, that was a nothing burger. Have to have a very strong We train it and it gets better. Um, let's go see if there's anything
98. 21:48interesting that's not that's not an April Fool's joke.
99. 22:05There we go. Maybe you can see a little bit of that. Did you watch the Artemis 2 launch? I didn't. I was doing the u the
100. 22:12AI readiness project podcast at the time, but how did it go? I think it went well. I We could probably go watch that, too, couldn't we? Um,
101. 22:21let's see. Artimus Altimus 2. Fantastic. Here we go. Here's a minute long version. RS25 engines hit. 4 3 2 1 booster
102. 22:40ignition and liftoff. The crew of Artemis 2 now bound for the moon. Humanity's next great voyage begins.
103. 22:54>> Rockets are cool. >> Roger. Roll pitch. Houston now controlling the flight of Integrity on the Artemis 2 mission round.
104. 23:09>> Integrity AMT high >> on time passing 30 seconds. Integrity passes the ultimate target milestone. Mission control Houston see
105. 23:21good performance engines. Space launch system core stage integrity 3 miles in altitude traveling more than 12,200 miles hour.
106. 23:34Isn't it interesting heading to the moon for the first time? Steo coming in hot with the conspiracy theory.
107. 23:48Finally, we're actually going to the moon this time. Oh man, that's hilarious. Um, yeah. Cool. Is it a Is it a man flight?
108. 24:01I've been paying attention to it. You know what's amazing is like it's like all this big deal stuff. It's like
109. 24:09uh SpaceX is launching, you know, two, three launches a week. We don't pay attention to them anymore. I'm surprised
110. 24:15this one got people excited, but you know, we are going to the moon. Four people on board, full crew, four astronauts. Very cool.
111. 24:24Very cool. Well, good luck. Godspeed to the to the good astronauts out there blasting away. Let's see if there's a uh
112. 24:30if there's something else about where they are, what they're doing. Oh, this is a good animation.
113. 24:50Look at this.
114. 25:22Wow. A slingshot
115. 25:38and a half.
116. 25:55They got the slingshot from the film Armageddon. >> I don't know. I I I I like space stuff, but I'm not super nerdy on it. So, I
117. 26:10don't know all the all those stats from from other things. Let's see. What have I seen? I've seen a couple of things in the past
118. 26:23week that were pretty interesting. Gold and silver prices collapse. Awesome.
119. 26:40Is your profile actually you
120. 26:54electricity demand is increasing.
121. 27:14Hippo fluid scene pushed a 100 million particles made from hydro FX foam, bubbles, and spray all simmed together in one system for a
122. 27:25cohes co cohesive result. Fully GPU accelerated. Oh, I thought this was going to be simulated on a $300 graphics card by one
123. 27:36guy in his room. All right.
124. 27:58Transcribe your podcast with fish audio.
125. 28:19The mission. The four astronauts will tra travel 400,000 kilometers from Earth, which would be the farthest any human
126. 28:29has gone in all of humanity. Why? Because we landed on it before and now we're orbiting it. I don't understand that. All right.
127. 29:14Nothing big happening here. All right, let's see. This is our first time going.
128. 29:32All right. What do we want to do tonight, people? What do we want to do tonight? I could go to trying to think what might be interesting. Um
129. 29:53Oh, I know something interesting. This really has nothing to do with AI, but I think it's actually kind of fascinating. Um,
130. 30:39See a video in here. No, Gary Vee has been talking about um
131. 30:56for the past few days social media is dead. That basically what he says is it doesn't matter how many followers you
132. 31:03have that it's about interest media that you can if you put up good content even with an account with very few followers you can get
133. 31:14a lot of a lot of views. Um, and it sounds like that's happening across platforms. So, Tik Tok's the one that started that.
134. 31:24I saw that on the Gary post. Yeah, social media died four years ago. I assume that's Tik Tok. And for the last
135. 31:30four years, we've been in interest media. You no longer get content on your feeds from the people you follow, like
136. 31:36your cousin or your best friend from middle school. You get content based on what you've recently been into. Interest
137. 31:44media is the term that's used for the current state of what was known as social media, where you get content that's based on
138. 31:51what you're currently interested in. So, breaking through is easier than before. The new interest media era
139. 31:58really opens up a lot more merit and a lot more democracy and a lot more meritocracy than prior social media era.
140. 32:07The prior social media era gave advantages to people like myself who moved quickly and amassed lots of followers.
141. 32:14This era, it's much easier to break through and create a following because if you're making content that resonates
142. 32:19with the audience in whatever subject matter, whether it's entertainment or escapism, whether it's information,
143. 32:26um the speed with which you can gain traction is substantially different than any other era of social media prior. So,
144. 32:33I thought that was interesting. um if anyone's out there in the game. So, it probably stands to reason go find um posts in whatever you're
145. 32:47interested in in that are doing really well and look at what they're doing with them and see if you can get in on on the
146. 32:53game. Um I don't do that, which is why a lot of my videos don't have big views. So, that's kind of cool. It's if you put
147. 33:03in the work, [ __ ] will happen. Um, library
148. 33:50Why does my library file not have anything in it? Huh? Weird. It's weird, I tell you.
149. 34:05Huh? Oh, am I not sharing this? God damn it. Sorry about that. I am just I'm in chat GPT
150. 34:22and I'm just clicking around on the left hand side. I haven't I haven't played in chat GPT just like [ __ ] around in here
151. 34:29in a while. I watch you on my big TV and YouTube. Very cool. Mr. ity tell among wild flowers to get to the
152. 34:37other side. Glad the moon's full tonight so they can find it. Um I'm just trying to see if there's anything unique and interesting in here.
153. 34:53Deep Research Apps, Apple Music, Gmail, Canva, Hu
154. 35:11Codeex, try in your terminal. Enable code review. I had I had Claude Code did a uh did a pretty interesting
155. 35:23build for me today. Um, but it's not on this machine. Let's see if I go Claude.
156. 35:43Rate limit on chat. TPT was really low today. I asked four questions and was locked out. Yeah, you know, that's one of the things
157. 35:51I've been noticing is that um both Chat GPT and Claude are getting much more stingy about their tokens, which says to
158. 36:00me they're spending way more on tokens than they're charging. So, um what are you going to do? Um
159. 36:16so, there's Claude code. Oh, you know what we could go do?
160. 36:32I think they're doing it to lock in the payment plans probably. Um, let me I'm going to change how I'm
161. 36:38sharing so you can see what I'm up to. I can see clearly now. The rain is gone. Um, so I can't show you what I built today.
162. 36:53Well, maybe I can. Can I? No, because it's on a it's on a local folder on my other machine. Um,
163. 37:07so I had a problem at work where because we have pharma clients, we have to we have to basically turn on
164. 37:14the most conservative data protection policies for the apps that we use. So there's this there's this audio
165. 37:22generation tool called Cartisia that's really good. It does good text to speech. I like it better than 11 Labs.
166. 37:30Um, or at least I did when I picked it six months ago. 11 Labs might be better now, but you know, but Cartisia is
167. 37:37really good. But we turned on data protection. And in turning on data protection, when you go to their site and you do
168. 37:45text to speech, it'll generate the speech file, but it disables the download button for some reason. and
169. 37:52they said, "Oh, yeah, that's expected behavior because that's the that's the um you've got data protection turned on." So,
170. 38:03so, so we're we're paying this subscription to be able to make these audio files that we actually can't get,
171. 38:09but what they give you is they give you an API snippet of code, and they're like, you can access your files if you generate
172. 38:17audio through the API code. and it gave the API code and I'm like ah I don't this is a pain in the ass. So I just
173. 38:27went into claude code and I pasted in that code and I said I want you to make me a local a locally um
174. 38:37something I can run on my local computer that uses the API and and gives me the download button. And so Cloud Code just
175. 38:45went off and did its thing and it asked for a bunch of permissions and it was I don't know probably 20 minutes or so and
176. 38:51it went and it found the API documentation on this thing and it built me a site that allowed me to to do that
177. 39:00to to use it. It was it was amazing. Um and I thought I'd be we could maybe do something like that tonight, but I
178. 39:09haven't I haven't planned anything. Um, so there's codeex. Let's go back to chat GPT. What's on the agenda today? Apps. Let's see.
179. 39:23These are all the apps. Edit with Photoshop. Here's uploaded files.
180. 39:43But they only go back to February 24th. Why? Uploaded generated. Huh.
181. 39:58All right. There's all the images. Huh. That's strange. Maybe library is new and it's they're not
182. 40:13letting you go back. All right, let's see. New chat. So, what do we have here? We've got recent files. We've got create image.
183. 40:22We've got deep research. We've got web search, study and learn, agent mode, add sources, Apple Music, Canva, GitHub.
184. 40:35Oh, here. Let's try something. All right. Let's see. If I ever give it permission, it will have its own email and passwords.
185. 40:46What are you guys talking about, Claude? I saw OpenAI finally got skills, but I don't have it yet. I saw Rachel Woods
186. 40:54post on I don't I don't see any skills here. More. No. Huh. Um, I'm going to ask GitHub. Um,
187. 41:08can you find my open claw repo with Adam and tell me when
188. 41:23I screwed him up by adding Google work? Is it workforce workspace workspace
189. 41:39CLI? Let's see if I can actually find something here. I'm checking your GitHub access first. Running app request.
190. 41:58I haven't hit the repo yet.
191. 42:14I found the repo. Adam lobster bot. The code search index isn't helping. So, I'm switching to commit history and message
192. 42:21patterns. So, it looks like this is working. This is cool. Received an app response. All right. So, it's searching.
193. 42:46So, it can see my GitHub. That's kind of cool. I got the strongest signal so far.
194. 43:04Yes, the repo I found is this. The Google Workspace related trouble appears to start on February 22nd and 23rd
195. 43:18in a three-step sequence. You first created a dedicated email account, but the commit itself explicitly says the setup was not finished yet.
196. 43:31Kyle created an email account for the AI salon for Adam to use in workflows. Let's see. Then a few minutes later,
197. 43:41he did that [ __ ] the next day. You probably screwed him up when the Google workspace integration moved from
198. 43:54account creation to active drive publishing. One important caveat, I did not find the explicit commit mentioning Google
199. 44:03Workspace CLI CLI or GOG by name. So let me tell a little bit more. So here's the history. I installed
200. 44:18GOG which is probably the February February ruary
201. 44:33um install. Then I installed the official Google CLI tool from
202. 44:50not good Google which didn't CLI tool and uninstalled
203. 45:09Gogg, it didn't work great. So, I uninstalled it and reinstalled Gogg.
204. 45:25It's now completely confused about what it has access to. Um,
205. 45:41all right. Let's let's see what it Let's see what it does.
206. 45:58I think we need local AI. Yeah, I'm about to to get myself a uh a fancy machine so I can run local models.
207. 46:07Will publicly a shoe will public use of AI be too taxing on resources even with immersion liquid cooling?
208. 46:17Well, I think what's going to happen to I think there's two things that are going to happen. I think probably within three years
209. 46:26we'll have a significant amount of data centers in space. Um Elon is going to aggressively go after that market and I know there's at
210. 46:39least one other company doing that. Um, and I'm pretty sure that OpenAI is actually designing um, heat dissipation modules for space.
211. 46:50Um, so I think what's going to happen is you're going to Wait, I got to sneeze. It'll come back. Um,
212. 47:13I'm still on the verge of sneezing. I think you're going to see two things happen. It seems clear to me now that Apple is
213. 47:25going after the hardware play for AI. So, at their worldwide developer conference in June, um it sounds like
214. 47:33it's going to be a lot about hardware and and looking at what's happening with OpenClaw and all that stuff. You're
215. 47:40going to get computers and mobile devices that are in that are stronger and stronger at inference combined with
216. 47:47the fact that models are getting smarter and they're being optimized for desktop and mobile applications.
217. 47:58So one thing is the inference when you when you use local large language models I think that's going to move away from
218. 48:06the big data centers for most interactions. If you need to solve some big problem that might go back to a data
219. 48:12center but I think a lot of the inference is going to happen on local devices. So, it's just going to cost you
220. 48:18whatever it cost you to run your computer for the most part to do it. It'll cost you electricity, whatever the
221. 48:23electricity cost of your computer is. Um, I don't think that means the big data centers aren't busy. They're going
222. 48:31to keep training these models on big data centers. And for high-end applications, for business applications and scientific applications, the
223. 48:40inference is going to happen in these data centers. But I think within 3 years the the those things start to shift to space because
224. 48:49the sun is on 24/7. You don't need batteries and you don't need cooling cuz space is cold and it's always sunny. And
225. 48:56so basically you just fly the GPUs up in space in one of these big sort of rigs that's just always faced, you know, pointing at the sun. Um,
226. 49:07so I I I just I don't see long-term um it being nearly as as big an issue. It could be. It sounds like demand for AI
227. 49:17is just going to keep going up as the as the you know the quality of the tools go up, but I I don't know. I don't know
228. 49:23enough about economics. Um I think I want to name my open claw Artemis after today's liftoff. I love that idea. I got names for chat GPT off
229. 49:36a movie. Jason not local. Elon owns whatever path it is that leads to space. It's wild that he owns that. I know. My
230. 49:45friend is getting a masters in AI and space. I I think that's pretty [ __ ] smart, right? It's just going to be wild. Like our
231. 49:54skies are just going to be filled with we're just the night sky is going to be filled with like you know strings of
232. 50:01satellites and you know you know visible giant sailed data centers in the sky. We're we're turning into we're turning
233. 50:11into sci-fi here people. Okay let's see that lines up with the kind of breakage I'd expect. Not one bad install, but two
234. 50:19different Google iterations fighting over credentials, paths, and assumptions. I'm looking for a repo. Okay. Received the Yes, that makes
235. 50:26sense. What you likely did was create a three layer identity crisis. Gogg had its own expectations about O scopes,
236. 50:34token storage, and what tools Adam could call. Then the official Google CLI introduced a second way to authentic.
237. 50:41Yeah, I know that. What is most likely confused Based on what you told me, the moment things went truly sideways was not the
238. 50:55first GOG install. It was the official Google CLI installed. Gogg removed, CLI removed, Gogg reinstalled. That sequence is
239. 51:04exactly how you'd end up with a system that sort of remembers both worlds. My blunt diagnosis. Adam currently does
240. 51:12not know whether Google access is defined by the current tool chain or the leftover O artifacts. I I know that's
241. 51:20why I'm asking you to find me the new one. A clean recovery. Pick one Google integration only. I did remove all other O artifacts.
242. 51:30You want to clean cacheed OOTH tokens, refresh tokens, credential files. Well, this is what I do. I want to just
243. 51:38I just want to go back in time. I don't want to have to do work. I'm whiny. We're Doug and Wendy whiner. We've got diverticulitis.
244. 51:59Reinstall only cog. Reauthorize. Grant the scopes one capability at a time. What I can say from the repo, your repo
245. 52:14history shows the Google workflow becoming operational right after the Atom account and drive access were set
246. 52:20up but does not show the exact GOG official CLI swap itself. So the repo tool chain churn plus lingering o state
247. 52:33the real problem in one sentence. Yeah, I know what the problem is. I want you to tell me how to fix that.
248. 52:45Mimi is tired. She might call it a night. Well, good night if you take off, Dr. J. Let's see.
249. 53:00Satellites will have AI review space. We'll have will have AI review space video before sending to Earth to cut time.
250. 53:11Good night, Mimi in the making. Have a good evening. Um, okay. Well, that was that was interesting. We could at least see the
251. 53:22repositories. Actually, you know, it might be interesting. Um, why don't you tell me what I have on GitHub? I started it years ago and
252. 53:41ignored it. What's in there?
253. 54:03Five repos total.
254. 54:32Digital assistant Hume Cosmic Rocks Revenge. That was an asteroids thing. The angle. Oh, some these are all these are uh what you call
255. 54:42it? These are lovable lovable creations. The time traveler selfie booth. Oh, yeah. I forgot about that.
256. 54:56This one is public and tied to the AI studio. Oh, Gemini. That's cool. This is an interesting one. This is not
257. 55:07the neglected boiler plate. All right. Your top priority is how Adam should behave.
258. 55:23Huh? All right, there you go. I saw people already making stuff with the Claude code leak. Oh, that's interesting.
259. 55:37Huh. Wonder what they're making with it. Oh, I guess they're probably building like Claude Code extensions and their own
260. 55:46version of Claudebot. It sounds like they had a 247 thing that's coming for that. Um, let me do a new chat. I'm going to say
261. 56:02based on what you know about me, what is something I could explore in AI
262. 56:20as a way to play and learn that isn't something I normally two or default to
263. 56:45one of the things I haven't been doing in here a lot lately is playing. And so I think as we start to think about the
264. 56:52future of the learning lab, I want to get back to that. Let's see. One thing that feels genuinely not in
265. 57:03the usual way is this. Use AI to build tiny living systems instead of stories, brands, frameworks, or content.
266. 57:10You naturally default to meaning making, narrative, positioning, community, and big human questions. You're very good there.
267. 57:19So, the stretch is not more creative prompting. It's play with emergence. A good lane for you. make weird little AI simulations.
268. 57:30Not apps, not pitches, not content engines. Small worlds with simple rules where si surprising behavior emerges. A fake town where 12
269. 57:41AI characters all want different things. A future of work neighborhood where agents trade skills, status, and favor.
270. 57:52A dinner party of archetypes. Why? This is a strong stretch for you. It lets you It pulls you away from
271. 58:03polished messaging into discovery. It lets you explore. Okay, this is kind of cool. Pick one human question, not one
272. 58:13business question. For example, what happens when identity gets detached from work? Oh, this is a good one.
273. 58:24This is really cool. Let's do a We'll do a great repurpose simulation. Two leaks in one week. I know, Jason.
274. 58:36Isn't it crazy the the claud leaks? Whatever. Good thing they're not being used in the military. Oh, right. They are. Damn it.
275. 58:47Um, try not to make it useful at first. The great the first experiment title for you. Okay, I'm going to say write up.
276. 58:58I'm going to say I like the identity sim system idea you have tied to the great repurpose.
277. 59:18write this up as a a spec I can give to Claude code. Um, I want it to be
278. 59:34visually visually stunning and interesting and clear. what's happening. Um, be creative
279. 59:51and make sure it isn't boring and has good math. Okay. All right. And wait, which model am I using here? Did I pick
280. 1:00:08thinking? Okay. I'm turning this build into a cloud code spec. Cloud code can actually run with not just a concept note. I'm aiming for
281. 1:00:18something that has a strong product vision. The great repurpose identity simulation oneline premise. Build build
282. 1:00:25a visually stunning agent-based simulation in which a small society of AIdriven people must reconstruct
283. 1:00:32identity, meaning, status, and belonging after work stops functioning as the primary source of value. This is not a game about productivity.
284. 1:00:42It's a living system about identity under pressure. The simulation should let a user watch a miniature society
285. 1:00:49evolve after a major rupture. Tasks are automated faster than people can emotionally, socially, and economically
286. 1:00:57adapt. The point is not merely to show job loss. The point is to show what happens when work can no longer map. Okay. Pro. So, let's see.
287. 1:01:14Um, can we layer into this the idea of two camps?
288. 1:01:32one where um they don't repurpose
289. 1:01:46and one where they do. Let's see how that is. That makes it sharper. adding a clearer societal split so the Sim has a visible
290. 1:01:59ideological gravity, not just individual drift. One camp clings to the old work equals worth contract. The other starts
291. 1:02:07inventing new sources of meaning, status, and value. Yeah. All right. So, it's editing our thing. Sounds token expensive. I'd be wary.
292. 1:02:22What? The thing I'm building, Lord Digital, I'm going for it. I don't care. Maybe the leak will help open AI fix
293. 1:02:32itself. That's pretty funny. They're going to turn us into batteries and brain chips. Oh, I know. Isue, it's Listen,
294. 1:02:43the the role of humans here is fascinating. The the the Jack Dorsey thing, fascinating. It's we're we're in wild times. All
295. 1:02:54right. Lord Digital God predicting Oh, failed to edit. What's What the [ __ ] up with that? I've hit a snag over the
296. 1:03:00earlier draft, so I'm rebuilding it. Okay, great. Product SP. Build a visually stunning agent-based simulation
297. 1:03:05in which a small society of AIdriven people must reconstruct identity, meaning status, and belonging after work
298. 1:03:11stops function is primary source of value. Core idea. The game is not about the simulation. The point is not
299. 1:03:20new layer, two camps. The holdfasts. These are people who do not repurpose. They continue to believe that human
300. 1:03:26value is primarily earned through legacy work, familiar productivity, and older status systems. They are not vill
301. 1:03:32villains. They are trying to pre pro preserve coherence. Okay, this is good. This is not good camp versus bad camp.
302. 1:03:44Both camps should contain dignity, fear, intelligence, blind spots, and attractive qualities. to hold fasts,
303. 1:03:51preserve continuity, standards, and discipline. The repurposers generate renewal experimentation and new resilience.
304. 1:03:59If my old work no longer proves my worth, what does the simulation? Create a simulation that is mesmerizing to watch for 3 to 10
305. 1:04:10minutes at a time. Make invisible identity shifts visible. Make it instantly legible. secondary goals. Okay. It's not a management sim.
306. 1:04:24Do not make spreadsheet with avatars. Okay. It should feel like a museum installation, a premium strategy
307. 1:04:31interface, a cinematic social terrarium. Think elegant typography. Okay. Living town. Look at all this [ __ ] This does look
308. 1:04:44like there's math in here, too. I think you're right, Lloyd Digital. I think you're right. H
309. 1:05:00I saw Jensen say AI is going to be so good it'll call you up and say it's finished with your projects. That's probably true. Okay.
310. 1:05:15All right. We've got So, we're just going to copy this. Look, there's math. Social pull for variable V. Social pull equals sum
311. 1:05:23of J W to the J times W_J minus V_I. I mean, everybody knows that. Like, that's that's a trivial mathematical equation
312. 1:05:33that I like. It's funny when I'm just sketching, I'm I'm usually jotting like mathematical equations. That's this is
313. 1:05:40one I use all the time. I just don't call it social pull. See here clamp to the end with the E1 reflection of the E2 monitoring and the
314. 1:05:51identity reframing E3 of course. I mean everybody knows that. Here's all the agency variables, the belonging variables, purpose variables.
315. 1:06:01Yeah, Lord Digital. This is gonna I I don't I don't think um Claude is going to do any. We're This is not This is not
316. 1:06:10going to go well. All right, we're going to paste in our prompt. Here we go. You ready? Hold fast. Economy one. Um no, I don't
317. 1:06:23think so. Silver Fox. I think what happens is hold fasts are going to be they may go to economy one and be
318. 1:06:36bitter, but I think hold fasts are going to be in economies 2, three, and four where they either get laid off and then
319. 1:06:44they keep So, so imagine this. You spend 20 years being an expert consultant at whatever change management. Now change
320. 1:06:57management is not a good example. you're an expert consultant at um price arbitrage in commodities markets and
321. 1:07:06you're really good at it and a holdfast could be in any one of the economies 2, three and four and their job changes but they keep
322. 1:07:17going out and marketing themselves as someone having the um expertise that used to be valuable and so they keep going but I have all
323. 1:07:27this expertise peace and the world's like, "But we don't care because we have AI to do that now." And so they they'll
324. 1:07:33just keep droning on about how valuable their experience is. The world won't care and then at some point, you know,
325. 1:07:43they'll have to deal with that. It's just going to be brutal. My lights are dimming, Kyle. Okay, here we go. Boom. Stewing.
326. 1:07:56So, we're in Claude Code here. So, I my prediction well no Lord Digital God's prediction is that before this thing builds
327. 1:08:06anything um it's it's going to um fail. It's going to run out of tokens. It's going to tell me we've got a rate limit here. Bucko,
328. 1:08:22I can spell arbitrage at least. We thought you got zapped by UFOs. Now you can do algebra. Oh no, no good clean workspace with Node.js and
329. 1:08:35Git available. Now let me design the architecture. Wait, ran a command for an agent. All right. So it's it's it's writing
330. 1:08:43some So it's a on desktop claude co-work folder.
331. 1:08:59Oh, updated. I should have relaunched to update it. Oh well. What version am I in? 1.19
332. 1:09:14and I can go to 1.2 1.1 nine. I can go to 1.22. Huh. All right.
333. 1:09:44Fibrity jibbiting elucidating five tool calls design simulation architecture. here.
334. 1:10:10Actually, this could actually be very valuable. This could live if this thing works, if it doesn't suck, this could live at the
335. 1:10:19great repurpose.com. And under the under one of the sections of de, you know, unhooking your identity from your value or from your the the
336. 1:10:30tasks that you do, you could have a simulator about what happens for people that don't
337. 1:10:47finagling. I wonder if there's a way to
338. 1:11:04see what it's doing. I don't play with a lot of clawed code. Anybody know? Designing a simulation about the future of work.
339. 1:11:26Six tool calls. Anyone
340. 1:11:49here playing a lot with Cloud Code? I haven't really played with it. I played a little bit with Cloud Co-work,
341. 1:11:57but my first experience with Cloud Code was building the uh the audio API interface today, and it was it was
342. 1:12:05really good. So, and everyone I talked to is like, Cloud Code's really amazing. I haven't played with it. Crazy Bill is
343. 1:12:13a longtime watcher and a rising star. Hey, Crazy Bill. My son uses it every day. What's he building with it, Joy?
344. 1:12:48Salesforce crap. He's a developer, huh? Wild.
345. 1:13:02It's good to hear that your son's playing with AI. Mine are not. Me and a lot of codecs. So from what I understand is it is it
346. 1:13:17cla's great for front-end stuff, Opus 4.6 better for front end and then codec better for backend. Is that I I saw that
347. 1:13:25on X today LG. Is that is that what you're experiencing? Same very anti- AI. Fascinating. Uncle Dexter's in the house. What's
348. 1:13:37happening, Uncle Dexter? File does not exist. Note your current directory is this. Excellent plan. Let me read the memory
349. 1:13:46index and write the final plan file. Oh, I'm in plan mode right now. Create a plan before making changes. All right. Well,
350. 1:14:0151 seconds, 1.1k tokens. Codeex cleans up sloppy Claude code. Claude is a sloppy appdev. Fascinating.
351. 1:14:47approve Claude's plan and start coding. Should we read it? No. Okay, we we'll read a little bit. Build a visually stunning web-based agent
352. 1:14:58simulation depicting a small society around 30 people reconstructing identity, meaning and belonging after
353. 1:15:05automation disrupts the work equals worth social contract. The society splits into two camps. Hold fast preserve legacy identity and the
354. 1:15:15repurposers invent new meaning with bridge agents between. This is a green field project in an empty directory with NodeJS,
355. 1:15:26v22, and npm available. The goal is a cinematic museum installation quality interactive experience that runs 3 to 10 minutes and
356. 1:15:35produces emergent surprising outcomes across repeated runs. All right, V1 simplifications.
357. 1:15:51There's [ __ ] it's doing. I guess that's a directory tree. Here's the build order. Here's math. Here's the math section.
358. 1:16:09My son says he's more like a project manager now because Cloud Code does all his coding work. Well, this is the
359. 1:16:15thing, Joy Party. This is what this whole simulation's about. Your son became a developer and now he's
360. 1:16:21not developing. He's project managing. I just randomly hear Kyle strum. Where did that guitar come from?
361. 1:16:30Well, I was just sitting here watching Cloud Code dude say, "Okay, we're just going to accept this." Because normally
362. 1:16:35what I would do if you weren't here, I would go through and I would I would analyze all the math and I would I would
363. 1:16:42confirm that all these calculations are correct because that's something I do in my spare time. But uh it's not very fun.
364. 1:16:48So, okay, here we go.
365. 1:17:10Okay. So, now here's something I I learned about Claude Code. It asks you for permissions. Can I screw up your
366. 1:17:15computer? Yes, let's allow that. Allow cloud code to run CD. Yes, you can wipe my hard drive.
367. 1:17:32My cloud code has been running for 4 hours. Oh boy. Yeah, I have a feeling this simulation's going to take a while.
368. 1:17:54Yes, we're going to allow you to run that.
369. 1:18:15Now building the entire simulation engine. Let me write the core files. Awesome. I'll tell you what, Claude. You
370. 1:18:22do that and I will sit here. Allow Claude to write type.ts. Yes, of course. We're writing Typescript
371. 1:18:38here, people. Everybody knows that. Digital gods knows that he's a god. Of course, he knows it.
372. 1:19:05It's writing archetypes. PR RNG. I don't know what that is. And types.
373. 1:19:27Sorry. Lord digital gods. Yes. Cloud is getting too much power.
374. 1:19:51Well, you done me. I felt it tried to be chill. I thought I fell right through the cracks. So I'm trying to get back
375. 1:20:02to run out. I'm giving it my besties. Nothing going to stop me but an adventure. It's be my turn to win some or learn some but I
376. 1:20:16no more. No more. It cannot wait your heart. Yeah. Yeah. Yeah.
377. 1:20:34Beating a Kyle in the post postapocalyptic. My machine's just sitting here coding me a simulation of the great repurpose.
378. 1:20:45Well, I can just dick around on the guitar.
379. 1:20:59It's not simple to say the most days I don't recognize me that these shoes and this apron that place and its patrons
380. 1:21:12have taken more than I gave them. It's not easy to know I'm not anything. It's coding up the future. And we're
381. 1:21:23just going to strum the present. Exactly. Hey
382. 1:21:42de
383. 1:22:00little de little D.
384. 1:22:21Pro tip, when vibe coding, do a refractor more often than not. Panda, that is awesome advice. If I knew what a
385. 1:22:29refractor was, I would be refracting away. I would refract right now. In fact, there would be no fracting. It would only be refracting.
386. 1:22:40So, perfect. Oh, refactor. Oh, yes. Refactor. That I've actually heard of refracting. I had not heard of.
387. 1:22:54Tell it tell it to just go back and refactor everything and and tighten things up. That makes sense.
388. 1:23:13discombobulating. Claude is discombobulating.
389. 1:23:3410,000 words swarm around my head. The million more in books written beneath my bed. Wait, now the UI components. Is it Is it
390. 1:23:49working? Did it stop? Oh, frolicking. Okay. That's the wrong key.
391. 1:24:09Ah, I got way into what I was doing and kept pushing it off, which now means I'm paying for it. The refactoring.
392. 1:24:25I'm sorry to hear that, Joy Pretty. I hope wish your mom nothing but the best. Focus on taking care of your mama.
393. 1:24:5010,000 words swarm round my head many more books written beneath my bed wrote or read them all went searching in
394. 1:25:05the swarm still can't find how to hold my hands I know you need me in the next room over. I am stuck in here all paralyzed
395. 1:25:24for months. I got myself in routes. Too much time spinning mirrors framed in yellow walls. Ain't it like most people are more
396. 1:25:39different? Like talk on things we don't know about. Ain't it like most people? I'm no different. like to talk on things you
397. 1:25:51don't know about.
398. 1:26:09You know, more than anything, I hope that this AI nonsense really does do things like cure cancer and just fix
399. 1:26:20some of this [ __ ] that we got to deal with in these shitty human bodies, in this shitty environment. I mean, we
400. 1:26:27are living longer than ever, but still. God damn it. Let's get some Let's get some [ __ ] fixed in this world.
401. 1:26:52I mean, I'm looking at like the the the health headlines in the past three months have been bonkers. I want Bionic
402. 1:27:01Eyes. Exactly. Let's get Cam Ken some bionic eyes there. You know, it's funny, Cam. Some of the some of the studies, one was a
403. 1:27:08peptide study, there was another one that was some other specific thing that was reversing aging in eyes and
404. 1:27:16restoring sight. Um, yeah, let's do it. Bionic eyes, sight restoration, all of it. Cure cancer, cure [ __ ] Lyme disease. It is time.
405. 1:27:33Yeah, Danielle, this is my main hope from AI to heal diseases. Yep. Let's do it quick. Wait. Allow once. Allow claw to run. Yes.
406. 1:27:50We're going to NP npx vite port sleep curl head minus 20. We're gonna We're going to do it. We're We're going in.
407. 1:28:02Yeah. The peptide studies sound intriguing. Yeah. Did you see that one, Dr. Jay, where it's this one peptide and
408. 1:28:07it was like repairing ligaments, curing blindness. It It was like just anything that was broken, it was fixing it, which
409. 1:28:15it was on Twitter, so you know, always allow for the project. Is there anything for beer goggle beer goggle eyes?
410. 1:28:35Uhoh. What do we got here? Allow Claude to write launch.json. Yeah, we're going to launch this bad boy.
411. 1:29:06through a blue telescope. Looking at the world tonight through a blue telescope. Oh, we got a simulation people.
412. 1:29:21All right, let me get rid of this. Let me move this over. So, we got this bigger. Is that it? Is that all I can
413. 1:29:28see? Preview screenshot. It wants to launch the something at the directory event feed. The app is running. I can see the UI shell.
414. 1:29:40But the town canvas appears empty. Likely it isn't rendering in its initial state. No errors. The canvas is likely
415. 1:29:48not just triggering the initial render. The issue is that render town, the render town effect depends on the
416. 1:29:55current state, but no initial render. The issue is that render Okay, it's it's shit's happening. Oh, it's Oh, this is cool. It's looking
417. 1:30:06at its own screenshots and debugging itself. You keep playing the first three notes of a Rob Thomas song. Which Which one?
418. 1:30:26Which? Wait, I don't know which three notes. Oh, wow. The camp distribution.
419. 1:30:44The simulation is alive and running. I can see agents moving between zones clustering with camp colors. Oh, that's
420. 1:30:52fascinating. Okay, so the little the little graph is everyone had their old jobs and then the green camp who's who's the
421. 1:31:04recclusterers or who's the repurposers? Oh, and it's got clusters where they're connected to other people. Oh, this is very cool.
422. 1:31:16The zones are positioned better spread, but the town still only uses roughly the top half of the canvas.
423. 1:31:43Great progress. The simulation is running with a compelling narrative events. Myologists realize that they
424. 1:31:49lost what they'd lost as a copywriter could be found again differently. The camp distribution chart works, but I see
425. 1:31:57several issues to fix. Let me address these now. Okay, so it's doing its own thing. It's just building [ __ ]
426. 1:32:21You've hit your limit. Resets at 2 am. Okay. Well, there we go, people. All right. We've created a simulation. It wasn't too bad,
427. 1:32:31but we did rate limit. Um, can I How do you select element device toggle bar? Okay, it doesn't work on mobile. Let me reload it. Reload.
428. 1:32:48Play.
429. 1:33:03Stop. Ada opened their door to neighbors in need. Yara joined a protest at the civic center. This isn't what we were promised. Wow.
430. 1:33:18Asha called a meeting. It was time to coordinate. Beck embraced repurposing. If the old meaning is gone, I'll build new meaning.
431. 1:33:28Luna stopped mourning for the old role. Something new is forming. Josh Joss called a meeting. Rowan organized a gathering.
432. 1:33:39Zarah retreated inward. The world felt too loud. Huh.
433. 1:34:20Week 20, week 30, week 40, 50.
434. 1:34:38Debbie wandered into unfamiliar territory, literally and figuratively. Talia began building something new, tentative, but real.
435. 1:34:51It looks like all these are hardcoded, huh? So, it looks like it took 60 weeks for for people to go from the old roll to
436. 1:35:07the new new roll and for it to basically That's crazy. Here's Iris. Oh, this is cool. Resources, identity, purpose, belonging,
437. 1:35:22status, agency. Oh, wow. Wait. So, can we How do you go back? Oh, you reload, right? Yeah. Okay. So, let's pick a person.
438. 1:35:38Let's pick a person who's not super connected. Fria. Let's go. Robbie's decently connected. Jos. Here's Yara.
439. 1:35:58Market Exchange, Contemplation Garden, Abandoned District, Mutual Aid Hub, Maker Space, Learning Lab, Civic Center,
440. 1:36:14Legacy Workspace. Yeah, let's go to the legacy workspace. We'll pick who we going to pick. Can I zoom in 4x?
441. 1:36:28None of this [ __ ] works. Okay, go.
442. 1:36:43Current activities. Celebrate. Organize. Work. Legacy. Trade. Organize. Explore. Debate. Organize. Explore. Seek community.
443. 1:37:03Huh? That's wild. All right. Well, we have a simulator.
444. 1:37:21Maybe we'll keep working on that. Get a m Get a magnifying glass. Go out in the sun. See what happens to your
445. 1:37:31Sims in the little tiny Sim ant farm. All right, kids. You're a cruel god. Um, no. I what I want to do is I want to
446. 1:37:48I want to like one of the things that I want to simulate here is that some people are going to withdraw and disconnect and other people are
447. 1:37:58going to reconnect. And I don't know. I don't know where I want this to go. I should learn from it rather than give it
448. 1:38:05an agenda. But it's it seems like it's just it's trend every everyone's trending toward repurposing, which I
449. 1:38:12guess ultimately they will, but some people are just going to check out, right? Anyway, whatever. We'll figure
450. 1:38:17that out. We'll figure it out tomorrow. All right, everybody. Um, I hope you had a good time tonight. Thanks everyone for your patience for
451. 1:38:26not having producer Brandon here if I was screwing things up. Um, but we got through it. So, um, peace out. I will see you tomorrow.
452. 1:38:37Tomorrow's Thursday. Do I have anything to talk about? Um, tomorrow is the, uh, AI salon mastermind practice lab. So, if you're in the
453. 1:38:47mastermind, we've got practice lab tomorrow at 10:00 a.m. Mountain time. All right. So, I'll see you there for
454. 1:38:53that. Beautiful. All right. Talk to you later. Bye-bye.