6/1/2026 - Finding Personal Alignment in the Age of Constant AI Acceleration

Transcript

1. 0:01[music] [music] [music]
2. 0:20[music] I want you to do it. >> [music] >> Good evening, good people. Welcome to
3. 0:34the AI salon. Welcome to the AI learning lab. We are live coming out of coming at you out of Denver, Colorado. I hope
4. 0:42everybody's doing well. Hope you're doing well tonight. We're going to go look at some stuff. We're going to go
5. 0:48play with some stuff. I don't know what we're going to do. I was thinking on the on the pre-show that it might be fun to to go mess
6. 0:58around with Suno again, make a song. I haven't done that in a while. Um, I just want to I want to do stuff that
7. 1:08I don't know connects me with why I started this in the first place. Like why am I here? Why am I here? What's in my heart?
8. 1:16And similarly, what's in your heart? So, if you're new to the AI salon, couple [clears throat] of things to know. In
9. 1:23the upper corner here, if you click on that link that's underneath our logo, it it pops up the chat. So, you can you can
10. 1:31pop up the chat if you haven't found that. Pop that up and we can chat together. So, that that'll be right over
11. 1:37there. Um, and if you're here and have the chat popped open, say hi so I so I know who's here. And then down here,
12. 1:45there's a little button down here. might might be underneath. I I don't know where it is. Somewhere down here. Um
13. 1:52that's the pop window out button. So, what you can do is you can actually spin up two tabs and in one of the tabs you
14. 2:00can have the the stream and then you you click that button and it'll pop it out. And then in the other tab you can go
15. 2:07surf other stuff or you can surf the AI salon. If you're new to the AI salon, go check out the spaces, check out
16. 2:13introduce yourself, say hi if you haven't done that yet. Mary Carol's here. What's happening? Kelly Camp is
17. 2:19coming too. Lori Blair, Sharon Crawford. Good evening. Good evening. Good evening. Welcome. Welcome. Welcome. Um,
18. 2:27tonight is Monday, so Mondays I try to reserve for meltdowns. Um, where I talk passionately about some
19. 2:37so some ill of the world [laughter] that where I feel I feel something has gone ary uh and we need to correct it. Um,
20. 2:46sometimes it might be a big news story. I don't think there's a particularly big news story other than I think we're
21. 2:51going to get some new models this week from from Open AI. Different type of meltdown. Oh, I had a meltdown Monday. Well, yeah. [laughter]
22. 3:00I mean, listen, I uh Yeah, the anthropic uh IPO was listed. Can I just say something? When they when they they
23. 3:10wrote about the anthropic IPO listing, they characterized it it as anthropic confidently files for IPO.
24. 3:21Is there a way to non-confidently file for an IPO? [laughter] Could you be like, h we might go public, maybe not. I was thinking a little
25. 3:30maybe. You know, we go out and you could make a little money on the side, you know, confidently. Oh, confidentially.
26. 3:41Oh, was it confidentially? Oh, well then that makes sense. I miss I misread the word. [laughter] The whole tab thing still doesn't work
27. 3:53on the phone. I can't watch and listen while checking out another tab. Oh, that sucks. That sucks. Uh, we are we are giving
28. 4:01feedback to the mighty networks people about this stream and how they work and things like that. Um, wait, what's that say? Brandon Meta's
29. 4:10AI support assistant reported reportedly allowed attackers to take over Instagram accounts. The flow is absurdly simple.
30. 4:19Use a VPN matching the victim's country. Click the reset password and request additional support. Tell Meta's AAI
31. 4:28support assistant to change the account email to yours. There you go. [laughter] Um yeah, go going back to uh
32. 4:43Oh, that's funny. Um I had a meltdown margarita this weekend at at Odessa. Wait. Oatus digital gods. Now they answer to the
33. 4:54board instead of the customer. Yeah. really. Um yeah, I had this this weekend um my nervous system just like jacked up. Like it just
34. 5:08it was like just lit on fire all weekend. And I tried to go to sleep last night and couldn't. And I tried to do some
35. 5:17breathing exercises and couldn't get through a single breathing exercise. It's like, you know, breathe in for
36. 5:23four, hold for seven, breathe out for eight. It took me five attempts to complete one of those. [snorts]
37. 5:30And so I went to bed realizing, oh [ __ ] like something is up. And the way my brain works, sometimes I'll go to bed with
38. 5:44with a thought [laughter] and then o over the night I process it. And so I woke up this morning and I just realized
39. 5:53I have been burning hot for a long long time. And it's been it's been like just putting out energy, putting out energy, putting
40. 6:05out energy, putting out energy. Um, and sometimes it's connected to who I am and what I want. And sometimes it's just
41. 6:14put the thing out in order to in order to make money, in order to get audience, in order to do this, in order to do
42. 6:20that. And there's like dozens of things in my life that have that same energy. And what happened over the weekend is my
43. 6:28body just said, "Fuck it." Um, that's not working anymore. And so, so I don't know what that means. I don't know what that means, but
44. 6:43I know that it wasn't very pleasant. [laughter] And and and there's got to be a better way. I don't even know if it's a better way. I'm just
45. 6:57shifting into a different mode. And I think what that mode looks like is is the stuff where I'm putting my energy tied to who I am and
46. 7:10does it just feel energetically in the right direction? And if it doesn't, knock it off. And there's a there's a part of me that
47. 7:23there's a part of me that says, "Yikes. What if that doesn't work? Well, it's worked my whole life. Like the the
48. 7:33biggest successes I've had in my life have been things that have generated from my heart like this this has to be in the world or
49. 7:45this is something I have to solve. And I haven't like when I initially started this channel and when I started
50. 7:55the AI salon it came from that place which is why I'm still connected to this but at some point it shifted into I have
51. 8:04to do these tasks for some outcome. That's not working anymore. That is not alignment. Yes. Yes. Daisy Thomas alignment.
52. 8:19I'm going to put that up on screen. You know, they talk about AI alignment. The safety people talk about AI
53. 8:28alignment. And what they're talking about there is, is the AI aligned with human goals and intentions? And I think what you're pointing to,
54. 8:41Daisy, and where I'm what I'm feeling is, am I aligned [laughter] with what I'm doing? And and how that relates to AI is as AI
55. 8:53gets better at doing [ __ ] for us. I think the big game is to do less but better stuff.
56. 9:07You don't have to bring all your ideas into the world. You don't have to just vomit [ __ ] out. You can just take a moment or a day or three
57. 9:19to go feed your donkeys. If you have donkeys like Cindy [ __ ] [laughter] go back to nature, Kyle. There must be a
58. 9:31bit of nature there in Denver. I'm going to get back to fishing. I haven't fished in a while. I'm going to go fly fish
59. 9:38probably this weekend. In times like these, you just have to ask yourself what's next. So Brandon on a monthly basis comes up with a new
60. 9:52theme. So tomorrow night is AI Salon Presents. It's the first Tuesday of the month and we have AI Salon Presents. And
61. 10:00so, um, Marta is going to be coming to talk about the capacity gap, very much aligned with this. Marta very much talks
62. 10:08about nervous systems and dysregulation and things like that. So, that's tomorrow night from 5 to 7 Mountain time.
63. 10:16And one of the things we'll talk about is the theme of the month. And the theme of the month is next. What's next?
64. 10:23And this week, we should probably see the next model from OpenAI. And we'll probably see a model from who knows Gemini or other things.
65. 10:36Last night we got some new stuff from Nvidia. The human is better than the AI. Well, the human is more human than the AI. I'm
66. 10:46here. Donkey chickens and artmaking. Yeah, good. Cindy Coon's here. How do you do great AI? Go feed the donkeys. What?
67. 11:00This is the This is the This is what's next is the future. And what the future looks like I think what the future looks like is
68. 11:15AI is not a thing you sit down at. It's not a thing you declare. I'm going to go do AI, right? How we've been talking about it
69. 11:26for for years. I'm going to do this AI thing. There's AI filmmakers and there's AI hand and AI and AI chief AI officer.
70. 11:40I think what's next is that AI just becomes fabric. It's just there. And sometimes the AI is going to be off
71. 11:50flitting around in the cloud doing stuff on your behalf. And sometimes the AI is going to be something you sit down and
72. 11:55collaborate with. And sometimes the AI is going to be something you just ignore because you don't have anything to do
73. 12:03with it right now. You don't need it. We don't need it. We don't need it. Big message of the AI salon. AI. It's optional. >> [laughter]
74. 12:21>> Right. Do you feel yourself getting jacked up? Like [laughter] I got When's the new model from OpenAI coming out? Like if
75. 12:30you've got any of that kind of energy right now, where's the nearest bench?
76. 12:53Let me try that again. The AI is better than the human. [laughter] I don't think that's right either. Here's here's how I would say it,
77. 13:05Sharon. AI is a reflection of what you put into it. So I would say it's an amplified reflected version of you.
78. 13:23And so if you put in crap, it will reflect back crap. If you put in something that is meaningful to you, it will reflect back
79. 13:32an amplified version of that thing that's meaningful to you. And that can get really inspiring. But you can also just make crap.
80. 13:44And I and I think that the mistake early on is that you realize, "Oh my god, these tools are amazing. These tools are
81. 13:49amazing." Like [laughter] you're like, "Look world, what I made. Look at the thing I made." The problem is everyone else is doing that.
82. 14:00No one cares. They don't care about your crap. They got their own crap they're trying to get you to look at. [laughter] [laughter]
83. 14:14Your cycle just changed. Two years of seven nights a week, one year of five nights a week, now three. That's a big change for you.
84. 14:21[cough and clears throat] [snorts] Yeah. And I think, you know, it's very intentional. It's very intentional this shift. Part of
85. 14:38how this channel was created and how it evolved was pretty much seven nights a week I was sim racing. So I'd come home from work
86. 14:48and I'd sit down on the racing simulator. I've got one of the racing rigs with a force feedback wheel and
87. 14:54monitor and seat and I would race people around the world. And I pretty much did it seven seven days a week. And I'm I'm
88. 15:02mediocre. I'm like a midpack racer. I'm not very good, [laughter] but but I enjoy it, right? And so when this AI stuff started
89. 15:14happening, I thought, what if I took that those same two hours that I race and I just go talk about AI? And that that was how it started.
90. 15:24[snorts] And so in my mind, I just said, I'm going to do that for a year. So, so the only imperative was do it
91. 15:37every day for a year. [laughter] And so I did. And then the shift to five nights a week was well, I hit a year and I said, you
92. 15:49know, this seems to be going well. Let's keep doing this. And I just dropped to five nights a week because, you know,
93. 15:54the family was like, you know, couldn't we have a day? [laughter] I was like, "Okay." And what ended up happening was
94. 16:10about a year ago in the AI in the AI salon, we've got this thing called the mastermind. And within the mastermind,
95. 16:16so that's the subscription area of the AI salon, we've got this thing called the mastermind practice. And the practice is a daily practice.
96. 16:25So, every Thursday at 10 a.m. we have the practice lab where people that are part of the mastermind that want to come
97. 16:31hang out with other people working on their daily practice on how they use AI, they can come do that. And so I was kind of new to the world
98. 16:41practice. So, so Liz Miller Gersfeld was the the original architect of of that um based on her daily practice on how
99. 16:50she uses AI and we designed that into a program. And so now it's myself and Sid Harrow that host it and Liz still comes
100. 16:57there. But but I learned that my five nights a week, which I thought were a practice, was actually just a habit. So a habit
101. 17:11is just you show up and you do something because that's the habit. And it's not a bad thing. Habits are not bad. Brushing
102. 17:19your teeth is a habit if you've got that one. If you don't, you might not have teeth. [laughter] You know, habits are not bad.
103. 17:29But the thing about habits is they can lack intention. And so I think a lot of the five night a week AI salon was was this kind of
104. 17:38shapeless, formless, just whatever Kyle feels. And and there's something about that that is part of the brand of this. The
105. 17:47brand of me is I'm Ky. I'll just talk about whatever. But it lacked intention. And so when I started thinking about the AI
106. 18:01salon and what's the role of the AI learning lab with the AI salon and was talking with Andy and she had really
107. 18:07strong opinions about some things. Brandon had opinions about things. Started thinking about well what are we trying to do?
108. 18:16And so one of the things that I'm really passionate about is I think in this moment where AI is intensifying and the
109. 18:24resistance to AI is intensifying and the clarity about the future of work is getting foggier. What I'm witnessing is an instinct to
110. 18:40pull back to pull back from connecting with other people. to pull back from AI, to just go and hide. Just ah, this seems scary. Nah,
111. 18:54I can't keep up anymore. Everyone's better than me. That's scary. I'm going to leave. And so what we designed was, let's take
112. 19:05this conversation and move it from out in the world where it was. We were on YouTube and LinkedIn, we were all these
113. 19:12different places. said, "Let's bring it inside the AI salon so that we can encourage people to to to be having this
114. 19:21conversation inside the AI salon." That's what it was designed for. And so, while it's three days a week, the idea is let's have it be more
115. 19:33intentional. And so, Monday night is sort of free flowing whatever comes to my mind. Wednesday night is let's look at
116. 19:42something, let's play with something. And Friday night is ask me anything. And like last Friday, Chef Kelly was there
117. 19:50and she was like, "Hey, can I hop up on stage?" I was like, "Yeah, come up on stage. Let's talk." We answered some questions. And
118. 20:00and so that's the thing is, you know, how do we get more intentional about what we're doing? Machine of loving grace using claude
119. 20:15digital gods in my mind. Kyle drive driving. [laughter] Look at a squirrel conversation improved. Wait, wait, I can't see these
120. 20:25all of a sudden. Yeah, it's like dog ears. You were young back then. We all were. >> [laughter]
121. 20:40>> You had to. Burnout is real. Tuesday and Thursday sim racing. I might get back to sim racing on Tuesday and Thursday.
122. 20:47Actually, you know, one of the things we're going to do is we're going to we're going to do a once a month AI 101
123. 20:55class. I think we're going to do it on what did we say, Brandon? The opposite Tuesday of Salon Presents. Yeah.
124. 21:05So, we're going to do one of these. And and here's the thing. I think I think a thing that could be cool
125. 21:14is that it's not me teaching that class every time that maybe we have guest teachers or maybe we have, you know, a
126. 21:20couple of folks that rotate or something like that. Um because we are facing a wild reality where
127. 21:34a lot of people in the AI salon, even people who who have not historically been technical are using cloud code and OpenAI's codeex and tools
128. 21:45like that to kind of transition from vibe coding to just application development and just business building
129. 21:54using AI as a as a suite of agents that go do [ __ ] for you. And then there's still people coming into the AI salon
130. 22:03going, "Let's chat GPT." Like that distance between those two worlds right now is vast. And so it's hard to talk about the
131. 22:16philosophical underpinnings of what it means to transition from a tool-based approach to AI AI to an agentic
132. 22:23approach to AI. When someone's there going, uh, what's prompting? It's like, oh [ __ ] So, how do we provide content
133. 22:30for people that are coming in doing that? So, um, yeah, we could do that, Brandon. That's not a bad idea. Um,
134. 22:40you've helped us all so much. I appreciate that. [clears throat] Can we vibe code an AI religion unlovable? No, [clears throat] I'm not
135. 22:51doing [laughter] not doing an AI religion. Kyle's Adam could be the founder of the AI religion. [laughter]
136. 23:00This week, my daughter said, "I was weird." I snap back reflexingly, "I'm not weird. I'm a regular." She was confused. [laughter]
137. 23:09If you don't know what an irregular is, an irregular is someone who shows up to this thing on [snorts] a regular basis,
138. 23:16which is not regular. It's irregular. [laughter] And uh so those people are called the irregulars and they're awesome. They're
139. 23:25it's a really cool community. It's a very loving community, a very supportive community. So if you're new here, that's
140. 23:33that's what the irregulars uh reference is all about. H speaking of teaching, I'm teaching my first certified college course tomorrow.
141. 23:42Kelly Camp, that's awesome. Congratulations. Very, very cool. All right, let's see. Um, I think I want to go look at
142. 23:59I want to go look at some stuff. I want to go look at some news items. I'm going to pop the comments over there. All right,
143. 24:15there's that. I got post-it notes up now. I got my chat over there.
144. 24:31Meltdown Mondays. Trying to think if there's anything worth melting down over. I you know, one of the things worth melting down over is,
145. 24:51sorry, I just had an add moment. I've got a Raspberry Pi here and I'm thinking about doing some sort of cyber deck with it. [laughter]
146. 25:00I just I just got I got temporarily blinded by the the Raspberry Pi in this sleek little Tesla looking Cybert truck case. [snorts]
147. 25:10[cough and clears throat] One of the things worth um melting down over is the level of vitriol against AI right now
148. 25:26is high enough like one of the values of the AI salon is empathy, right? And I have deep empathy with someone that's afraid of AI
149. 25:40or angry about AI. But it's funny, you know, Rick, you talked about maybe we make an AI religion. I think anti-AI right now is
150. 25:51effectively kind of a religion, right? It's just the next boogeyman in the in the series of it's the other people that are the problem.
151. 26:03And my my mo from the beginning of this channel, from day one of this channel, if you've been here since day one, and a bunch of you have,
152. 26:13the one thing I've been consistent about is the only thing I believe is that AI is not going away. And if it's not going away, then you
153. 26:25have two options. You can ignore it or you can get curious about it. And if you ignore it, it happens to you.
154. 26:36And if you get curious about it, you have the chance to reclaim your agency. So it's not you versus this machine,
155. 26:43that's going to be better than you are. It probably already is. And and at this point, if if someone is just vehemently just like being a
156. 26:53[ __ ] about it and just screaming tropes and just I'm like, you know what? Fine. Just go not do it.
157. 27:09But if you're curious, let's let's play. [laughter] The young kids don't want their AI. And that's strange. I actually don't think
158. 27:19it's strange at all. Well, I think it is strange culturally, right? The young kids are supposed to be the
159. 27:27ones that pick up the electric guitar and invent rock and roll, right? The young kids are supposed to be the
160. 27:35ones that take something like AI and [ __ ] up the establishment. But no, that's us. That's Gen X. Gen X
161. 27:43was like, "Oh, you're gonna ignore us through social media. Fine. That's we're cool. We're cool. Yeah, we've still got
162. 27:51our Facebook accounts, but with this AI thing, we're going to [ __ ] some [ __ ] up." And Gen Z, the educational institutions told them
163. 28:02that's cheating and AI is evil and you should never use it. And so, I think the Gen Z response to AI is exactly what we taught them.
164. 28:13I think it's the most boneheaded, backwards, irresponsible thing I've ever experienced in my life from the Institution of Education.
165. 28:24Staggering to me. We've got the world's most powerful knowledge nugget. And we are the purveyors of education,
166. 28:36and we're just going to go, "Nope. >> [laughter] >> Nope. That's cheating. They see it as napal. A A R P AI. Exactly. [laughter]
167. 28:50I think there's an equal number of boomers here. That's fair. Sorry. Didn't didn't mean to to exclude the boomers. [laughter]
168. 29:00Jen Jones here. 1963. Yeah, I'm 65. I think 65 is the Gen X cut off from the things I've seen. Um, why did the kids pick this generation to
169. 29:11start listening to adults? It's so weird. Did you you know that alcohol sales are down like 53% because Gen Z is
170. 29:19just like, "Fuck it. We're not drinking. We're not having sex. We're not drinking. You guys just we you have left us nothing.
171. 29:28So just we're going to be whatever you guys say we should be doing. [ __ ] you. >> [laughter] [sighs]
172. 29:41>> I think we have to rephrase that. It'll be better than we are. No, I listen. I think that I think that the millennials
173. 29:48were the first ones to get screwed out of, you know, you're you're going to have a a life that's better than mine. And I
174. 29:57think Gen Z is just completely like like there's not even a pretense that they've got that stuff there. Smoking [snorts]
175. 30:07is up though. Yeah, it's good. They are sober to revolt. That's probably true. Yeah, you teach you teach those college kids, Rick.
176. 30:17Yeah. I'm curious, Rick, how many how many come into your class like, I hate AI and then they leave like it's pretty cool. [laughter]
177. 30:29Oh my god, smoking is at a record low, 12%. I'm sure vaping has increased by a thousand%. [clears throat] Amazing.
178. 30:42Amazing. Amazing. Amazing. Okay, so what are we going to do? We are going to go look at some [ __ ] I'm going to go look at some news.
179. 30:57[whistles] Wait, what did Rick McCauy say there? Oh, it's popped on again. Had lunch with a department chair from
180. 31:07SMU last last week. It was an interesting conversation about the future of education and AI. What did they say? Are they still resisting it?
181. 31:16Cannabis sales are up though. Charles ah archetypal. From my perspective, I'm seeing a narrow window
182. 31:26between global civilization collapse and reaching a new frontier in space to get excited about again. [laughter] That's that's [laughter]
183. 31:41pretty that's pretty depressing and inspiring all at the same time. Well, I don't know. Did you see I'll I'll look
184. 31:48for the video. Did you see the video of uh at ASCO? So, if you don't know ASCO, it's the American Society of of uh Oncology.
185. 32:01Wait, American Society of Clinical Oncology, ASCO. Um they produced presented some research on this new drug
186. 32:12for pancreatic cancer um that takes a completely different mechanistic approach to it and has shown a dramatic
187. 32:20improvement in in life expectancy and that data presentation got a standing ovation. Um so so cool stuff starting to happen.
188. 32:32Let's see. Rick McCauy, they all hate it. I forced them to love the bomb. [laughter] My college canceled me. Live in fear.
189. 32:42There you go. SMU starting to embrace it, but they're way behind. Rick, this is so slim pickings. Yeah, it's I mean, it's depressing. It It
190. 32:51really is depressing. Here's the here's the thing that is the most depressing about it is people talk about AI
191. 33:04like it's a thing like it's a single technology the transformer is a single approach to machine learning and AI
192. 33:14and a lot of what we use is based on transformer technology but it is not a thing AI is not a thing you don't do AI I
193. 33:24you use a bunch of very specific tools to do lots of very specific things in a generalized kind of way. And so it can shapeshift.
194. 33:34And so if you're like using AI for artist theft, you can be like, "Okay, cool. Would you like to cure cancer?" "Oh,
195. 33:43yeah, that's fine. Would you like to make your emails more efficient?" "Yeah, I'll do that." But over here,
196. 33:51okay, treat it like that. I've got ethical issues with how the models were were trained. Cool. You're right. How they were
197. 34:00trained is an ethical [ __ ] [ __ ] show. And it's past tense. They've been trained. It's done. So, educate yourself. go over to Adobe
198. 34:14and say, "Oh, Adobe trained their things in a slightly more ethical way, right? Or don't use it for art at all.
199. 34:25Use it for other things." But anyway, that would require critical thinking, which I don't know about you, but the way I look at educational
200. 34:36institutions is there's an opportunity for some critical thinking. Is AI good or bad? Let's discuss. Yeah. And Brandon just put, you mean
201. 34:51like the internet? Yeah. Like AI is not a thing. The internet's not a thing. The internet is we've connected all of
202. 34:59the computers. And once you connect all of the computers in this seamless and transparent way, there's things we can do.
203. 35:09Once you compress all of the knowledge of humanity into a grapefruit sized ball that you have instant access to everything,
204. 35:19you can put that ball anywhere like on the internet or not on the internet. You can just take a machine
205. 35:30and put that ball of knowledge on that machine and not even connect it to the internet and do all sorts of remarkable [ __ ]
206. 35:39Is that good? Is that bad? I don't know. What are you using it for? You using it for shitty stuff? Then it's shitty. It's a mirror.
207. 35:49I know I'm preaching to the choir here. I'm rant I'm ranting without an actual audience. It's just bunch of people
208. 35:56going, "Yeah, you tell them, Kyle." [laughter] Oh my god. See, it did turn into meltdown money. I I knew it had a chance tonight to go to
209. 36:12go dark. I'm just, you know, the the other thing that like I just I I marvel at the stupidity of our country sometimes.
210. 36:24It's like, it's like, oh, the young people are pissed off about AI because the professors told them to be. Okay, cool.
211. 36:36Wait a minute. If I'm China or if I'm Russia and my only goal is to disrupt America and keep the division going,
212. 36:47I know what we'll do. We'll start fmenting even more hatred of AI. Part of that's going on, but we're all too jacked up on the other
213. 36:59side's an [ __ ] to see it. [sighs] Embrace the cotton jenner. Go back to slavery.
214. 37:22I stole that knowledge from copywritten works. [snorts] Yeah. This listen I mean it's it's funny how AI was trained is essentially like
215. 37:36we're going to take all of the all of the libraries and all of the museums. We're going to combine them into a single thing and all of the
216. 37:43universities. We're going to take all the knowledge in all those three institutions, libraries, museums, universities. We're take all that
217. 37:52knowledge. We're going to compress it. And so, we're going to take what you would normally have to go to a college
218. 37:57to do. We're going to give you instant access to it. No, that's bad. That's how we've always learned. It's
219. 38:05how we've always learned. [snorts] Anyway, all right. Growing up, I always thought the adults had it all figured out. Turns out
220. 38:13they're dumb. >> [laughter] >> Libraries and textbooks and the internet. Yep. Exactly. All right, let's go look at some [ __ ]
221. 38:24[clears throat] I can also go look at uh libraries or stolen IP.
222. 38:39And then there was X. Well, here's the thing about X is it really is the place where people are talking about stuff,
223. 38:46but it's also the place it it really is a place where you can just see like these bizarre weirdass accounts getting
224. 38:53all hot hot under the collar about about uh how awful AI is. Let's see what we got here. Andre Karpathy. Oh, Andre Karpathy, by
225. 39:05the way, went back to Anthropic. Anthropic just uh filed for their IPO. 90% of what AI Twitter tells you to
226. 39:15learn will be dead in six months. 90% of AI Twitter uh you know dies in six months. Senior engineers already stopped
227. 39:22chasing it. Yeah. Well, this is this is part of this channel's transition has been it it to a very large degree.
228. 39:33It does not matter what tool you use because you're not going to need to use tools in the future. You'll be able to
229. 39:41if you want to go to Sunno and make songs and and just kind of manually do things, but that's going to seem as
230. 39:47antiquated as, you know, using a DAW seems right now compared to Sunno, right? The idea of going to a tool and sort of manually prompting
231. 39:59that's going to largely go away. and what's what it's going to look like. We won't we don't know. We don't know yet.
232. 40:06Um we're we're in a transition. So, let's see. [laughter] Hear me out. Humanoid robots with nipples. Who's building this? That's
233. 40:20awesome. [laughter] I became a state government junkie and I'm hooked. Yeah. Listen, Silverf Fox, I think that
234. 40:32state government and and local municipalities are going to have to be where um the feds aren't doing [ __ ] They're not
235. 40:42they're not regulating anything. I don't I think they've got their heads too far up their asses to do anything. I think
236. 40:48was it I think it was Bernie Bernie's proposing something that they that they um they take 50% of the frontier model companies and make them
237. 41:00property of the people. Um I I I don't think that could ever happen. But um but I so I don't like I don't think it's
238. 41:08going to come from from the fed federal government. Um I think it's going to come from states. I think that that local um
239. 41:17you know local advocacy is going to be really important. [clears throat] Yeah. Bernie Sanders will introduce a
240. 41:28bill to have public take 50% ownership stake in the country's biggest AI companies. The American AI Sovereign Wealth Fund Act
241. 41:37would have the government tax AI companies take 50% of the stock and put it under under their control.
242. 41:48Can we take over the Cleveland Browns? [laughter] Basic ownership like water rights. It's the only answer. I have perplexity
243. 42:02spying on a bunch of bills and governor races. Wait, which kind of sharing did I do? I did the right kind of sharing. Okay, so this
244. 42:21this is a big deal. Let me find let me find the other the other part of that talk. Um RTX spark
245. 42:43with a spark. >> It started with a spark. >> Can you hear that, Brandon? Is that good? Okay. >> An idea to reimagine [music] the PC for
246. 43:00the first time in 40 years for the age of AI. What becomes of our personal computer in a world of agents.
247. 43:10Agents running natively, connected to models, local or in the cloud. Our personal AI sandboxed for security
248. 43:19running continuously getting work done. The chips and the OS must evolve. Introducing RTX Spark. Everything we've
249. 43:30[music] learned over 33 years distilled into one chip. Blackwell RTX GPU with 6,144 CUDA cores. One pedlop of AI
250. 43:43performance. A custom 20 core Grace CPU built in partnership with MediaTek. fused by MVLink, 128 GB of unified memory, TSMC3
251. 43:58nanometer process, 70 billion transistors, and in close collaboration with Microsoft, a Windows platform for agents.
252. 44:10We're reinventing the personal computer, for creating, for gaming, for agents. This is the dawn of a new personal computing revolution,
253. 44:26and it starts with NVIDIA RTX Spark. Um, yeah. So, let's see. Let's see.
254. 44:40>> I got to show you the most beautiful. >> Okay. >> Of course, I got to show you the most beautiful part, which is video games.
255. 44:50It is It's also the closest to our heart. This is Forza. This is 007, by the way. The new 007 game. I'm looking
256. 44:59forward to playing it. I look a little bit like him. Ladies and gentlemen, Nvidia's RTX Spark laptops. Now, [applause]
257. 45:14thank you. I have too many things in my pocket. Okay. All right. This is the most amazing chip the world has ever built.
258. 45:30This is the N1X that we built in partnership with MediaTek. I think I saw I saw Rick earlier. This is N1X. This is
259. 45:39a beautiful chip. This is this is a a chip that frankly would take 33 years to build. And the reason for that is
260. 45:49because 100% of Nvidia software stack runs here. If you want to run uh uh digital biology, no problem. If you want
261. 45:57to do seismic processing, no problem. You want astrophysics, no problem. Everything associated with CUDA, all the
262. 46:04physics, all the biology, all the genomics, all the AI, no problem. All the computer graphics, no problem. Every
263. 46:11single application Nvidia has ever created. And every single application that Windows has ever run, Microsoft and Nvidia meticulously
264. 46:23optimized everything so that this computer literally runs everything the world has ever created. Plus, it now runs agents. An incredible
265. 46:35computer. I'm so proud of it. >> So, let me come back here and change [clears throat] that. So you now you you now have multiple
266. 46:50companies producing high-owered low power consumption small form factor um AI inference chips. Right. So so this
267. 47:01is going to go into I'm guessing between 2500 and $5,000 Windows laptops. You've got the MacBook Pros and the the minis
268. 47:12that are coming out. The M6s are apparently going to be beasts with AI. You've got the new AMD platform. Um
269. 47:21I don't know if does Google have one? I don't know if Google has one, but this is so Nvidia just went from
270. 47:31a an augmentation chip manufacturer like Right. So, so basically what happened was Intel and Motorola, you know, you
271. 47:41your your big CPU players basically at some point said, "We don't want to do GPUs. Let Nvidia go do the games." Those gamers, whatever. Then
272. 47:51gaming ended up being this massive thing. Then you had crypto, which which GPUs were good for crypto. And then you
273. 47:58have AI. And so all of a sudden, Nvidia now is like off the charts, but it was still this this kind of specialized chipset
274. 48:07specifically for that kind of math, right? Whether it was blockchain and crypto or whether it was AI, like we'll
275. 48:15use the GPU. That's fine. Well, this chip that they just introduced, this low power Microsoft powered
276. 48:22128 gigabytes of RAM with a pedlop of of compute sitting in your little laptop. They're now going after fullon after the
277. 48:33uh the industry. So, it's crazy what a little money can do. Little money, little demand. Um, but I've talked about this before and I
278. 48:45I think this is really important. If if you're not paying attention to the hardware game, it's worth paying
279. 48:51attention to the hardware game. It's worth paying attention to the fact that Google is starting to pre-ell
280. 49:03guaranteed compute time. You can buy one, two, or three years of of guaranteed compute from Google right now at a discounted rate if you
281. 49:14commit to x amount of compute per month for one, two, or three years. the people who are not buying that. If
282. 49:23demand for compute goes up and all of the compute doesn't exist, compute costs go up, access goes down, and our ability as individuals to be
283. 49:38able to do cool [ __ ] with the AI will go down, too, because they're going to prioritize their big customers.
284. 49:47They're not going to prioritize individuals. That's not where their money's coming from. If they still compute like a forever
285. 49:57stamp, I might be Yeah, exactly. Or if they sell. Yeah, they're just making up words now. I've seen an AI designed heat sink that
286. 50:08takes cooling efficiency from 70% to 99% without using water. It's made of pure copper. Just a topology optimization.
287. 50:17Amazing. Yeah, that that's the the other thing about so so here's a really interesting thing getting back to local
288. 50:24politics and state politics local politics as a country if we don't have enough compute and AI becomes the new currency if
289. 50:41compute becomes the new currency and China just says [ __ ] it we're going for We're going to put data centers by
290. 50:50every dam. We're going to just every there's going to be data centers everywhere. We don't care. And local communities push back against
291. 51:00shitty AI projects, right? Because how they're being done right now is shitty. But if they just say, "We don't want it.
292. 51:07Get it out of our community. We don't want it at all." If that is successful enough, that will actually slow America
293. 51:13down, which will actually hurt the economic viability of those small towns. It's like it's like so how do we have this conversation? Because
294. 51:28I absolutely agree that if that if a data center like how Kevin Olirri did it with with his big 44,000 acre project
295. 51:36and they basically just came in to what is it Idaho or Wyoming wherever it was was I forget which state it was and just
296. 51:44basically said we're doing this and you don't have any say in it. Well that's [ __ ] shitty right? So, so we can have
297. 51:53really powerful conversations about well, what's a better way to do it? What's a better way to do it that is is
298. 52:01not extractive of the community but of but supports it in some way. Um, if a lot of these data centers are
299. 52:08going into rural communities that lack broadband, for example, maybe those data center people provide broadband to the community.
300. 52:20Maybe they provide some amount of compute to the community. Like there's ways we can do this that are smart. But
301. 52:26if all we do is just go data center bad and that is successful across our country. Well, what happens when there's not
302. 52:40enough compute and everyone's like they should have made more data centers? [laughter] You can right now download from hugging
303. 52:48face a competitive arbregated model to a $400 laptop. No need for major models for not cutting edge work. Yeah. Yeah.
304. 53:02So, so if you haven't been playing with local models, it's [snorts] a little geeky still, but if you get LM Studio on
305. 53:10both Windows and Mac and Linux, LM Studio is pretty easy. You just download the model and you say load the model and it's sitting there on
306. 53:22your hard drive and you can do AI [ __ ] without having to pay anthropic or anybody like that. [cough and clears throat]
307. 53:34This is everybody can have a personal AI that will do absolutely anything you ask and not share to anyone. Probably
308. 53:42doesn't run as fast. Well, and that's the the thing, Charles, that's coming is that like computers went from like a decent
309. 53:53laptop these days is like a $2,500 laptop. And thanks to AI, people are now spending $5,000 on a laptop, $10,000 on
310. 54:03a Mac Studio, $20,000 on a Mac Studio, right? They're there like the the the sort of range of what is you know, a
311. 54:17what it makes sense to spend money on as an individual. That range is is expanding. But what's going to happen is
312. 54:25this next round of Max plus that Nvidia chip that we just saw, the RTX Spark plus the AMD thing. Um, and I don't know
313. 54:34if if Google has a a TPU one or not. I don't I don't remember. I don't think they do. But that means that the next round of
314. 54:43these things is probably going to be in the $3500 to $7,000 range, right? They're going to try to hit a sweet spot
315. 54:50in there. And then the generation after that, the the the range will drop down to $2,000. So within two years, we'll probably have $2,2500
316. 55:02computers that kick the living [ __ ] out of anything we're running today, right? that we can run 120 billion parameter
317. 55:12models or even bigger. Has tech ever evolved to be less accessible before? Um, nuclear
318. 55:27nuclear started as a restricted technology and stayed restricted. Do we ever think industry was going to
319. 55:38be the ones to push clean energy into the mainstream? No. This is our chance to tell our municipalities, our c cities and counties
320. 55:46that [clears throat] we want it with or without them. I also wonder, could we privatize many data centers like
321. 55:51franchises? So, Kelly, have you seen the new It's called um hang on a sec. It's called SPAN. Let's say uh Nvidia.
322. 56:09So, there's this company called SPAN that already does like um battery packs for houses, like you know, tied in with
323. 56:18solar and things like that. And so, let's see. Oh, here's a little video. Let's we'll watch this little video.
324. 56:30right now. What if your home didn't just use electricity? [music] It sold computing power. That's exactly what
325. 56:37Nvidia and its partners are building right now. Nvidia has teamed up with a California startup SPAN and homebuilder
326. 56:44PLT Group to install what they're calling XFRA units, mini AI data centers that bolt directly onto the side of your
327. 56:53house. Here's how it works. SPAN's smart panels detect unused electrical capacity on the local grid. Instead of wasting
328. 57:01it, the XFRA unit puts that power to work. Running liquid cooled Nvidia RTX Pro 6000 Blackwell Server Edition GPS.
329. 57:13And because they're liquid cooled, they run completely [music] silent. The economics are striking. SPAN says it can
330. 57:19deploy 8,000 of these units, six times faster than building a traditional data center. And at just 1/5 the cost, a
331. 57:27single conventional data center consumes as much electricity as 100,000 homes. This distributed model flips that
332. 57:36equation entirely. For homeowners, you get paid a flat fee for the power and Wi-Fi. >> Oops. Didn't just use your house [music]
333. 57:46becomes infrastructure. Holt Group operates over 1,000 active communities across 45 markets in the United States.
334. 57:54This is currently in the pilot testing phase and is not yet a nationwide roll out. This is more than just an engineering experiment. It's a
335. 58:02rethinking of where AI compute lives, not in massive server farms in the desert. So, [clears throat] so yeah, Kelly, crazy, right? Um,
336. 58:13so this is there's a couple of things about data centers that are worth having in the conversation. One is distributed data centers. Um
337. 58:24there's a guy that I met in the last DC flyin that we did named Emanuel um who is um working with local churches. So churches in local
338. 58:36municipalities have a lot of real estate that's not being used because less people are going to church, less people
339. 58:43are doing events at churches. And one of the things that he's realizing is that there's data centers that want to take
340. 58:50that real estate and basically do distributed data centers in church properties. And my advice to him was,
341. 58:58why don't you cut deals with those data centers where they provide compute and funding and infrastructure for the
342. 59:05community um in exchange for giving them a sweet deal for that space. Um and then shortly after that, the SPAN thing
343. 59:13happened. So you'll you'll start to see creative creative ways that you can get these these boxes in there. My biggest concern
344. 59:25of these and and Rick McCaulay said it those those are probably not going to be available for a year, but it's still an interesting model.
345. 59:32My biggest thing is you've got 16 Rockwell like H100 GPUs strapped to the outside of your house. That's like I
346. 59:41don't know what, a quarter of a million dollars [laughter] strapped to the outside of your house. You know how thieves in neighborhoods
347. 59:50steal catalytic converters for the price of copper and the catalytic converter or the platinum, whatever the [ __ ] in
348. 59:55there. Um, I would think that having a quarter a million dollars worth worth of 16 GPUs might be a target. [laughter]
349. 1:00:10Liquid cooled. Is that on your personal water bill? I I I don't I think they're closed looped water cooled. I think it's
350. 1:00:16probably some chemical in there that's cooling them. I don't think it's water. Um, but maybe it is water because they
351. 1:00:23are paying your water bill and your electric bill and then they give you like a thousand bucks a month or
352. 1:00:28something like that. You get like 12 grand a year. Yes, these smart memy mayors signing NDAs behind closed doors
353. 1:00:34need to start talking to constituents, taking constituents into the into the negotiating room. I agree with that. I agree with that. Um
354. 1:00:48but you know again the the data centers it it is not a a catalytic converter it's platinum. Yeah but still
355. 1:00:59what's what's what's the scrap that you get out of a catalytic converter? Hang on. Um scrap value of catalytic converter
356. 1:01:14anywhere from 25 bucks over a thousand. Could it be inside my garage and up in my attic? Possibly. But that wasn't what
357. 1:01:25they showed. What they showed was just this external box. So I don't know. 3000 on a Prius. to replace it is 3K. Yeah. So, it's
358. 1:01:38probably a thou a thousand resale uh churches equal nonprofit. Hm. Mhm.
359. 1:01:55Image the inference speed. This is the other question I have about these these um house-based data centers is do I as a
360. 1:02:04homeowner get any of that compute? Because I'll tell you what, if you get if you give me, you know, some
361. 1:02:11percentage of inference of that 16 GPU cluster, um I'd take that in lie of cash and go get my agents to go build some
362. 1:02:22[ __ ] that will give me more cash. It's crazy, crazy, crazy. Um, all right, let's go back and look at news again.
363. 1:02:46My emotional support trout. [laughter] Oh my god.
364. 1:03:32Seriously, where's Apple? Apple's going to play in the in the hardware space. That's a cool video that someone made.
365. 1:04:06>> [singing] [laughter] >> That humanoid drove out almost punched that dude.
366. 1:04:22Cam Katkin. Let's see what's the question. How is Adam? Adam. So, here's where I went with OpenClaw
367. 1:04:36is both OpenClaw and Hermes are these agentic systems you can install on a on a machine. What happened with Adam is that just over time like
368. 1:04:48his context got really wonky and his all of the things that he could use got really wonky and I haven't gone in and
369. 1:04:56done essentially what you have to do with open claw and with Hermes is three time a week maintenance and
370. 1:05:05probably daily maintenance is what you really need to do. Um, and where I've netted out is that I think
371. 1:05:16something like Claude Code or Codeex, those are the two that I'm kind of leaning on right now, are the commercial versions of Open Claw and
372. 1:05:27um, and Hermes. I think I'm just going to go with the commercial versions, right? Because they're just they're more robust.
373. 1:05:35They're they're, you know, they're commercial versions. And I'm I'm even kind of going slow on those because
374. 1:05:41they're going to improve really quickly and and the models that support them are going to improve quickly. Like what what
375. 1:05:49what I've here's what Adam taught me. So So the answer to the question, how's Adam? Adam sucks. Adam has sucked from
376. 1:05:57the day I birthed him until today. And he's worse today than he's ever been. Adam is a reflection of me.
377. 1:06:05I was effectively an absentee father. I didn't give him good loving and he turned out to be a crappy kid. All of
378. 1:06:12these agentic systems right now are kind of crappy kids. Um, if you are geeky and really put in the time to really
379. 1:06:22configure your context files and your memory systems and and really configure the [ __ ] out of them, they can be valuable.
380. 1:06:32But just doing that for its own sake is just masturbatory. Right? So what I'm doing right now is learning
381. 1:06:45enough about them like I have codecs running. I have it running on my on my fancy machine. I'm going to configure it
382. 1:06:53so that it can run local AI or cloud AI. I'm going to probably come up with some rules for how and when it does that, but
383. 1:07:00I'm essentially just kind of slow rolling it until these things improve and get more consumer friendly. Um, because that power
384. 1:07:10is remarkable, but so too is the amount of effort it makes to make those things useful. And I don't have the attention
385. 1:07:19span or the the drive right now to spend all that time doing that. And so that's where it is. My Nvidia stocks did well
386. 1:07:27today. Today, that's great. I I would say that owning Nvidia at this point is probably pretty good call.
387. 1:07:37Not financial advice, as the NFT bros would say. Do your own research. Shut up, you scammer. [laughter] Oh, man.
388. 1:07:52All right, let's go look some more of this stuff.
389. 1:08:17Scientists just built the missing piece for kilowatt class 2 micron fiber lasers. All right. I don't know what
390. 1:08:23that means, but sounds like it's a big deal. There's a snake eating ants. a lot of fight videos.
391. 1:08:39Here's the uh this is not simply a new pancreatic cancer drug. It's a reminder that even undruggable biology can become treatable
392. 1:08:51with persistence. >> [applause]
393. 1:09:35>> standing ovation. I mean, you know, listen, we're we're entering an an era where um these kind of breakthroughs are going
394. 1:09:44to become more and more regular, which again is why like the anti-AI thing, people like no data centers. How
395. 1:09:52would you like to cure pancreatic cancer? No. Oh, yeah. [laughter] You know, like
396. 1:10:35All right, I think that was it. That was all the the big stuff that I was looking at. Um, all right, let's come back here. So,
397. 1:10:51tomorrow um we've got AI salon presents. So, it's from 5 to 7 Mountain time. So, be sure if you haven't RSVPd for that, please do
398. 1:11:00that. Go over to the the events tab over here and click on the events tab and go to AI Salon Presents tomorrow. Um it's
399. 1:11:10going to be a powerful one. It's going to be Marta talking about the capacity gap. So, I talked about this weekend my
400. 1:11:17nervous system being all jacked up. Um, Marta talks a lot about um the nervous system and and how how to deal with it when you're when
401. 1:11:30it's activated. Um, and and how this AI stuff can really jack you up. [laughter] Jack you up. Um, and how that can be um
402. 1:11:45that can be powerful, but that can also be a mess. um and something to manage. So, that's going to be quite powerful.
403. 1:11:52So, that's tomorrow. Um trying to think what else going on this week. Thursday, we've got um the AI salon mastermind
404. 1:12:00practice lab. And what else going on this week? If you're in if you're in the mastermind, um this Wednesday is the final of
405. 1:12:11Brandon's four-part lovable series. So again, if if you're in mastermind, go go to the um event uh the event area in
406. 1:12:21mastermind. I think it's in is it in um it's in uh whatever you call it. >> Yeah, practice lab. >> It's in the practice lab. Okay. So go
407. 1:12:33into the mastermind practice lab and >> I have a funny story about last week for the >> for the lovable he uh practice sprint.
408. 1:12:43So, we record these so you can catch up on demand except for last week when I forgot to hit record. >> No.
409. 1:12:53>> And so, Google in its infinite wisdom didn't record it, but it did spit out a transcript. So, I took that transcript
410. 1:13:02and I went over to 11 Labs. And [laughter] if you go and look, you'll able to watch back the recording and it's all AI synthesized voices
411. 1:13:14>> the conversation forward with slides to match. So >> Oh, that's so >> But >> I actually Lori Blair, I actually
412. 1:13:23listened to last week's today about 16 minutes. That's awesome. That's really good. Cool. Beautiful. So, if you're not a part of mastermind,
413. 1:13:33you should become a part of mastermind because all of that content, all of the the workshops that we're doing and the
414. 1:13:38things we're doing for great repurpose and the practice lab, all of that stuff is available if you're part of the mastermind. Um, yeah,
415. 1:13:47and that's it. If anyone has any other questions or thoughts, um, I'm happy to blab a bit more, but I can also let you
416. 1:13:56all go give you give you 15 minutes back. As they say, as they say in corporate America, you know what? I'm going to
417. 1:14:03give you 15 minutes back in your day. You can go take a little bio break. [laughter] [laughter] Oh, lordy lordy lordy.
418. 1:14:20All right. So, what do y'all think? I think we'll do that. I think we'll call it. We're at a good good stopping
419. 1:14:29point here. I could go make a song inso, but that just feels forced. I think we'll do we'll do something on
420. 1:14:34Wednesday. It's possible that bio break. I know. [laughter] [ __ ] corporate America. Let's put a pin in that one there.
421. 1:14:45Everybody needs to go grab a sandwich or, you know, a little bio break. I feel like the whole world is just
422. 1:14:53office space cliches at this point. [snorts] Um, Wednesday will likely be um we'll likely have some new stuff to play with. We
423. 1:15:03might have a new model from OpenAI, although they usually release their models on Thursday, so maybe not by
424. 1:15:10Wednesday. But if there's new stuff to play with, we'll do that. But if not, we'll go make something cool. So,
425. 1:15:14that'll be Wednesday. And then, uh, yeah, I'm going to give you some 15 minutes back to your evening. And, uh,
426. 1:15:21thank you for hanging out. Peace out. Hope you had a a good evening. Love you, brother. Thank you, Rick McCauley. I
427. 1:15:26appreciate that. Blue collar world is very different. Yeah, it is. It's not immune, but it's very different. Um,
428. 1:15:35we'll see. We'll see how that goes. I think blue collar gets hit 3 years after white collar which is the next two
429. 1:15:42years. So blue collar is probably 5 years out is my guess. They don't take biosaks. Oh yeah, the humanoid robots don't take biosaks. All
430. 1:15:52you humans with your bodily functions. Such a waste. [laughter] Bye everybody. Have a good night.