0:00GPT6 Astra brings together years of research. This seems like a next generation Frontier model that's designed with CUA from the ground up on certain benchmarks like RKGI 3 which it saturates. Fantastic. But then you look at the artificial analysis benchmark and it actually lags behind Meta Muse Spark, you know, like it's kind of weird. Tesla held its cyber cab La Palooa, a river of golden EVs flooding the streets. Elon wants to sell
0:29these at 30,000 each so you can buy 10 of them and put them on the streets in your local town, have them earn revenue for you.
0:36There'll be enormous chunks of entire cities that say, "You know what? No more human drivers. It's so much more efficient. Not only is it much cheaper, but it's much more the technology is arriving fast. Like, you know, just cuz something just formalized Fermat's last theorem in 13 million lines of code, proving 29,000 theorems on the way." Really?
0:58Map is cooked. Everything is cooked.
1:02Now that's a moonshot, ladies and gentlemen.
1:07Welcome to Moonshots, everyone. Your number one podcast on all things AI and exponential. Your front row seat to the accelerating singularity. Today, all of the mates are in the house. The quintet, the fabulous five, and oh man, oh man, we're going to need all of their brain power, you know, to get through the day today. This is an epic and I mean an epic episode today. Let me introduce you to magnificent my magnificent mates. Uh if you are the first time joining us
1:35we've got Dave Blondon the emperario of AI investing AWG. Alex Wner Gross our in-house ASI Sim Ismael our globe troder father of the organizational singularity and warlord against linear thinking.
1:51You're the one lord forever more.
1:54I'm good. I'll go for that.
1:56And finally, Immad Mustach back in the house by unanimous demand. I'm Peter D.
2:01Mandis, your moderator and datadriven optimist. So, See, uh, we got to ask, where are you today? And what happened last episode? Were you held up in customs again?
2:12No, I was not in customs and thank you for the kind words from everybody. Uh I was with the uh seauite of one of the bigger oil companies in the world in Houston and it was kind of a really surreal and fascinating conversation. Um but but so I'm really deeply sorry to have missed the last episode but Jesus guys I mean you know I missed one episode and you guys announced interstellar mission. You discuss AI designing its own chips. You hand you
2:40hand Elon the planetary thermostat. Apparently, I'm the moderating influence in this group, which should worry a lot of people. Um, I'm actually down in the Caribbean right now. We decided we needed a couple days break, so we popped down. So, you guys are standing with between me and a Hobie Cat and a little colorful drink with an umbrella in it. So, we got to keep it tight and punchy today.
3:01Well, it it's a big episode today, Seem. So, you know, you know, get patient with us.
3:06Yes. You know, our miss our mission here on Moonshots is to help you understand what just happened, what it means for you, and most importantly to keep you abundance-minded about our future. If you haven't had a chance to subscribe yet, please do right now. Hit the subscribe button. You know, we're putting out two episodes a week. Uh, and God knows the demand is there for more and I wouldn't want you to miss one. So, uh, two different things I'd love you guys to take advantage of. Uh the first
3:36is if you haven't yet, please register for our AMA. Uh we're going to be doing an AMA with all of our listeners who want to. Uh this is the last chance to register. You can go to moonshots.com.
3:49You'll be joining us on Zoom. We'll have the mates there answering your questions. We want to get a chance to know you, who you are, what your questions are. Again, last chance to register. Uh the link is on the screen in the show notes. moonshots.com and then follow us on X. Our handle there is moonshots_pod.
4:10So, uh there you have it. I want to take a second and actually uh you know, read a few of the subscriber comments. We got such love from our subscribers. You know, here we go. Uh at Brian Clark said, "Moonshots is the best content on YouTube, especially during the singularity. Thank you for all you guys do. It's our pleasure and and we do love it. Brian Anderson said, "The best AI podcast on the internet. Just fabulous.
4:35You guys are great." And RanchVidzy said, "Keep the regular uploads. Upload every day and I'll watch it. I'm a college student and I consider these videos more important than my physics lectures."
4:48Yeah, that's not a very high bar right there. But but but we do appreciate it.
4:52Not better than the physics lectures.
4:54More important than the physics lectures. When does Alex physics is cooked? When?
5:02Well, I want to say thank you to our listeners. We read your comments and it means the world to us. You know, our moonshot on this pod is 100x growth to get to 10 million subscribers. Please tell your friends, help us get there.
5:14Hit the subscribe button. Yet, today we're going to be covering one of the craziest and fastest weeks in Moonshot history. And Alex, like you always say, it's never going to be any slower. We're digesting 26 stories across 10 areas. An insane week for model releases, reminiscent of the hypersonic tsunami that we're living through. You know, insanely, we've had 12 Frontier model releases in the last 30 days, an average of one every 5 days. And there are rumors of at least five more releases
5:44expected in the next two weeks, including Grock 4.7. So, buckle up, grab your coffee, and let's jump in. Uh, let's kick it off this way. Uh, this week was a battle the Titans, OpenAI versus Anthropic, slugging it out for the Paro Frontier for the heavyweight crown. And the numbers couldn't be more exponential. Two Frontier Labs released their major models within 48 hours of each other. And you've got to know that this was a game of chicken, right, guys?
6:13I mean, who's going to release first?
6:14the other the other guy waiting to like slug it back and crown again. I'm just curious, you know, the strategy these guys are taking on this front.
6:22The releases are closer and closer together, but the models are improving more than ever before during these very short releases. These are not like, you know, kind of marketing garbage releases. These are major step improvements in the models themselves.
6:34They're just coming faster and faster and faster. So clearly, they're well down the self-improvement path. Clearly the prior model is accelerating the timeline to the next model. Um, did we cover Fable 5.1 already too? It seems like using it for a lifetime.
6:49I know it was it was 03, you know, two days ago or something like that.
6:54To take a day off from this pot, unfortunately, if you follow the extrapolation, I mentioned this a number of episodes ago. I think we're on track still to see one major model release per day by the end of this year.
7:06Yeah. Yeah, I I imagine that. So, uh, let's open up with OpenAI's release of GPT6, also known as Astra. The model's performance numbers are nothing less than stellar, saturating multiple benchmarks. I'm going to read a statement from OpenAI's release page.
7:22So, quote, GPT6 Astra brings together years of research and big bets across pre-training, reinforcement, learning, and alignment. Astra is a state-of-the-art uh on computer use, browsing, software engineering, cyber security, science, and professional work. Astra saturates Frontier Math Tier 4 with a 98% score, having already helped solve long-standing open models in math. Astra also saturates Arc AGI 3
7:50with a 99.9% score and exploit Bench with a 100% score. Astra's story is really about efficiency, not just raw intelligence. It defines a new paro frontier for intelligence index versus output token tasks. A particular note, Astrus hallucination rate fell nearly by half from 92 to 51% with accuracy actually increasing. So before I open up to you, Alex Neimod, to talk about the
8:18benchmarks, I want to show a quick video of uh of Sam Alman discussing this on Bloomberg TV uh to get a little bit of an overview on the topic. All right, let's take a listen. Sam, the way that OpenAI is framing Astra is basically an early step toward AGI, but I thought we could start our conversation just with basically what's fundamentally different with Astra relative to to prior generations of model.
8:47It feels first of all, thank you for having me. Um it feels like a a new step in this process towards models that can really help us create value, do work, uh discover new science, help us create help start new companies or uh create new products. I using it subjectively feels to me very different than any model before.
9:08What's the kind of principal use case that's different this time around? You know, I think we can get into the generations of model that were very coding focused. I know a lot of the development was very cyber security focused, but taking it away from the software engineer, what does the everyday person unlock in AI that they weren't able to do previously?
9:28One thing that I think people will immediately notice is if you have an idea and you want to work interactively with an AI to get uh you know a complex piece of a whole piece of software built, this is the first model to me where I could sort of tell someone like just give it a try. there's a there's a good chance it'll work. And you know, I' I've watched people make computer games.
9:50I've watched people do sort of like home sort of DIY electrical engineering projects. Uh I've watched people do very complex simulations for some piece of science they're working on. Uh certainly a lot of the work where you would sort of normally sit down and have to you know build a financial model here and then a PowerPoint presentation around it and then figure out how to like make a little interactive piece of code to sort of try different simulations. That stuff is it's all so doable now. Um what I
10:18hope will happen is people I think people will be surprised at the beginning but then as they build up more trust in the model and more like wow it can really do this just start throwing harder and harder tasks and more creative ideas at it and uh they will find out what the model can really do for them.
10:36Sam the specifics of how Astra is being released I think it's important. So that a version of it with specific guard rails. Um what was the thinking behind that and what are those specific guardrails uh in this early release of it?
10:50Yeah. So the the the challenge of our industry is that we have these models that are getting getting incredibly capable and incredibly useful and that people want to use to everything from you know make their lives a little easier to starting new companies. And then on the other hand, as these models get more capable, um the risks that we have to mitigate uh also become more serious. The models could do more damage if we don't um if we don't do a good job at that. And so we have spent uh you know, obviously
11:19this model took us a little longer to release than we were hoping. I think it'll be worth the wait, but we really wanted to spend the time on the safety and security alignment of this model.
11:28Um and I think as people understand the power and impact and capabilities, they will be they will be happy that we did. Um we will uh we'll have different tiers of cyber access for this model for cyber in particular. Um today we're rolling it out uh to trusted access partners and then in the coming days assuming everything goes well we'll roll it out more broadly.
11:50All right. Um we'll talk about the ben we'll talk about the safety elements in a little bit but uh Alex and Immod I'd love your take on on this one. Alex you want to jump in first?
12:00Sure. So I think there's an outer perspective which is to say what the market and what users would see here and then there's the inner perspective and I I think they're strikingly different perspectives. So to start with the outer perspective, this is a model that uses fewer output tokens to accomplish a given task. That's really interesting. It affects the economics. It affects the user experience. It affects the speed.
12:24Open AAI launched this and I think we'll we'll be able to go to the video soon of computer use assistance. Remember that when Anthropic leaprogged OpenAI, they leaprogged OpenAI in terms of revenue and other factors in at least two key regards. One was they were focusing on enterprise high revenue per token use cases, namely code generation. The second was they were focusing on computer use assistance. So anyone who's
12:53ever used cloud code understands cloud code can use all of the tools available in your local desktop environment. And that was it was a pretty big leap compared to what was available on the market prior to claude code. So the outer perspective on my side at least is that OpenAI finally with GPT6 Astra has at least internalized the computer use assistance story. There are demos. I don't have access to Astra yet, but just based on everything I've read and seen,
13:22they seem to have gone full native with the computer use assistance story, this seems like a next generation Frontier model that's designed with CUA from the ground up. That's really interesting in part because to be an amazing computer use assistant, you need to be able to understand what's on the screen. You need to have native multimodality. You need to be able to parse video and images and screenshots in a really tight interactive low latency loop. So if I
13:52had to guess as to what the outside design consideration that OpenAI was optimizing for here as reflected in certain of the benchmarks, I think that would be a leading candidate. Then there's the other side which is the the inner story just what if anything in that interview Peter that you were just playing the interviewer was asking Sam okay so what on the inside what's the big technological innovation there there I think the plot thickens a bit based on public comments from open AI leaders and
14:22and analyses from others it looks like the big technological innovation is the introduction of recurrence into open AI models in the form of looped transformers so this is basically taking a single transformer, stacking it on top of itself with the same weight, so weight tying, and then just rec it recurrently for a double loop rather than just a single loop. And you'll recall that all these Chinese labs that are achieving breakthrough performance are also injecting recurrence in
14:51different places like um Kimmy, the Kimmy model series is injecting recurrence via their KLA, their Kimmy linear attention mechanism at the attention layer. It seems like is this thinking about your thinking or is this just thinking about it and then thinking about it again before answering neither and both at the same time.
15:13Remember when we talked about anthropic study of consciousness in their models and we found that the middle layers in the yeah jspace they were in the middle layers. So one can squint at this looped transformer construct which by the way was being used by academic labs and others to achieve breakthrough performance with very tiny models on ARC AGI and other benchmarks. It it it almost follows if you thicken the depth of a transformer you might thicken the
15:41depth also of that JSpace or other middle layer area where most of the quote unquote conscious thinking happens. So that that's my best guess as to what the architectural inner strategy was here and that carries all sorts of implications if true including by the way the possibility that we're seeing the beginning of a new scaling law which is depth scaling which we've never seen before.
16:05Amazing. We we'll go to the benchmarks in a little bit. Emmod love your thoughts on Astro.
16:10Yeah, I mean it looks like the first nonbenchmaxed model which I think is an interesting one. So on certain benchmarks like RKGI 3 which it saturates fantastic but then you look at the artificial analysis benchmark and it actually lags behind Meta Muse Spark you know like it's kind of weird because I think again this hasn't been benchmaxed.
16:31This is a brand new pre-train. Uh what that means is that uh Greg Brockman said in an interview he did earlier this week that actually uh on the release of Astra it's the first pre-train they've had since GPT40 which I find a bit hard to believe. Apparently the 5 series was all on that 40 and they took that pre-train and they extended it out.
16:52Yeah. I thought the nomenclature I thought the nomenclature was when you get to the you know to 4050 it's another pre-train.
17:00No way. In fact infamously I think you're probably tracking this part of the story that the the the scuttlebutt was that most of the team responsible for pre-training left open AI. So they were left with an old pre-trained starter model. But that that's insane if you remember how far they managed to push 40 with its internal JSpace and everything all the way up to 5.6 Pro which was a really great model that started cracking math. Right. Then the other side is they actually indicated
17:28this was trained on 100,000 nextG chips. So presumably the GB300's, the Blackwells, not the Ver Rubins yet. I estimate the total cost of that training run is a billion dollars.
17:42And so it's literally like orders of magnitude more than the Chinese model pre-trades which is like $10 million.
17:50And what that has led to with the recurrence and others is you've probably seen on Twitter like you give a picture of a house and it generates a whole 3D model of it in Unreal or kind of something like that. It has accurate physics. It understands the internals and I think that is a factor of the normal scaling laws plus again this depthwise scaling law for this brand new pre-train which we're now seeing the start of the optimization like I think
18:19they probably got another pre-train coming that's even bigger um now they got the pre-training team back and it's going to scale from there amazing uh Dave excited uh you know this is the first class of models where as you're talking to it, it's showing you screenshots of your own laptop saying, "Is this what you wanted?
18:38Is this" and it's right in the text stream and it just happens automatically. You don't you don't really install any third party component or anything like that. And it qualitatively it is massively different than a month ago. Uh, and and I think that that solves a major gap in user experience, too, because you know, usually it would come back to you with these very long- winded technical explanations and you'd be like, okay, I can spend the 20 minutes trying to understand this. Now, it just shows you an image as you're talking to it and
19:07says, I could do this or I could do this. What do you want? It's just a massive difference in the user experience. And you know, the Astra naming implies a big leap, which it is.
19:18the five to 5.1 naming on the anthropic side seems like a trivial thing, right?
19:24It's not trivial at all. It's it's just qualitatively very very different.
19:29And we'll get to that in a couple of stories. Um See, what's a scuttlebutt? I mean, you're on the road speaking to CEOs of some of the largest corporations on the planet. Are they are they scared?
19:40Do they get the speed that what's going on?
19:42Uh no, they are so woefully behind. uh most of them are dabbling and what I mean by dabbling is uh here's the thought experiment if you took AI out of your organization today would any workflows change and the answer for most people is no which tells you that they're tinkering with AI but they're not really making structural change the folks that are making the structural change and how they rewrite their organizational design how they rewrite
20:10workflows is where all of the the uh uh advantages lie Right. Um I have a couple of uh thoughts here that I'd like to throw out. Um I I came up with two metaphors to describe what's happening and you guys tell me what you think of this. The first is that these frontier models are like making kind of like an endless pot of chili. You add data, you add compute, you have tools, you have reasoning, you have safety seasoning, and then millions of users taste it and tell you what's wrong. Now, they're not
20:40endlessly modifying the same pot. They're just making new batches faster and faster, right? And the real exponential is not one batch or one recipe. It's like the accelerating learning loop between all the batches of chili you're making. The danger is there may be no such thing as the perfect chili. Eventually becomes so powerful and so spicy you have to decide who's allowed to eat it. That's a metaphor.
21:02You must have been hungry when you came up with that.
21:04That was one. The other one I was I was stuck in a lot of traffic and I came up with a Formula 1 metaphor which was every model. It's like Formula 1 racing. Every change in a car has hundreds of little small improvements. Uh no single change can explain why you got the fastest lap and it feels to me like frontier AI is becoming like Formula 1. The models are already incredibly fast.
21:26But now the com everybody's like trying to shave milliseconds off intelligence or fewer tokens off lower uh lower cost better reasoning whatever and and very importantly better breaks. Uh and so I think this is the the the couple of metaphors I'm kind of playing with my head to try to make sense of this madness and that's the only way I can frame it is like just complete madness.
21:47You know, Salem, your chili analogy makes an important point that I think Alex has been predicting for a long time, but the age of data starvation is about to hit us. The these models have accelerated so much and you know, they're cooking math, they're cooking coding where the data is abundant, they're going to start cooking physics, but the areas they can expand into now, architecture and drug design, completely data starved. So every company we're involved with that's involved in gathering data is growing faster than
22:16any companies I've ever seen before. But it's because of exactly what you're saying. Like the chili is really good and all of a sudden it can make gigatons of it. But it's just starving for data.
22:27So given remind me Alex's framing of you know domain X has cooked the chili analogy applies better than the Formula 1 analogy. Well, we we have to remember to include recursive self-improvement. So, if we're going to torture this analogy further, the chili is cooking itself.
22:45Okay. Uh, Alex, uh, we've lined up a couple of charts here, uh, on that I'd love you to just touch on. The first one, Epic Capabilities Index. What is this and what are we seeing?
22:58Yeah, so the frontier is still somewhat spiky. So, Epic Capabilities Index, ECI, is maintained by Epic AI. It is one suite of possible benchmarks. It leans heavily into math uh and other technical fields. And according to the ECI, GPT6 Astra is now the new capability frontier. It is number one in the world.
23:22Uh there's uh I think an image that we didn't include, one of my favorites, Frontier Math Tier 4 version two, because version one turned out was it contained a number of incorrect answers that AI itself had to correct. math is thoroughly cooked at this point. And Frontier Math tier 4 v2 is part of ECI and according to this benchmark a if if you just for those who uh who can see this slide you can see this beautiful
23:49linear just trend over time perfectly predictable going back years and GPT6 Astra is number one according to this beating fable 5. So that's perspective one. Perspective two different suite from artificial analysis uh the the organization their ai according to this benchmark which is maybe a little bit more focused on uh broad economically valuable activities.
24:18Interestingly strikingly GPT6 is not number one. Uh, as Immad was alluding earlier, Claude Fable remains number one, specifically the the 5.1 release that also just happened. And Meta's Muse Spark is the the number two vendor model family. And then in a not quite distant number three, they're they're pretty close, but nonetheless, GPT6 is number
24:46three. So depending on how you measure GPD6 may not actually be at the capability frontier or it may be a cost frontier. So the ability so this is again uh ai versus cost per task. If you look at this then there's a cost frontier where you see if if you look in direct your eye to the upper right hand corner fable 5.1 is capability maxed and then if you look just to the the lower
25:14left of that you see GPT6 Astra running at max reasoning isn't even at the frontier it's just below claopus 5 and then we get to some of the Chinese models and meta models and Chinese models again so on a just purely based on A ai on performance versus cost. It's almost at the frontier but not exactly which is I think suggestive of what open AI was thinking. If we look at output tokens is what I was mentioning earlier
25:42output tokens per task it regains the frontier. So according to output tokens per task for those who can see in the the upper sub diagram here in in the lower the the sort of the left/ lower lefthand corner of the upper diagram we see that the optimal frontier of capability on the vertical axis versus output tokens per task on the horizontal axis is just dominated by GPT6. And I I
26:10think this is very suggestive as to what OpenAI was actually aiming for. think I I suspect they were deliberately trying through architectural choices and through targeted applications trying to minimize the number of tokens required to accomplish any task and I suspect that's because they had computer use assistance in mind where the computer is basically just being driven by the model requires multimodal capabilities requires low latency requires token efficiency critically it's slower to do
26:40all of your reasoning out in chain of thought versus as a feed forward pass of your transformer.
26:46All right, one last chart here. ARC AGI 3 leaderboard.
26:50So, this is bizarre. I I've never quite seen uh like a a non-con convex frontier here, but learn something new every day, I suppose. So, as Amad mentioned earlier, ARC AGI 3 is being quasy saturated at this point by GPT6. For those not tracking, Arc AGI 3 is the latest Arc AGI AI benchmark that's basically focused on whether AIs can
27:17learn on demand just in time. Call it the mini physics of a block world. A little animated pixelized universe. Some of them look like Tetris games. Some of them look like other video games. But imagine a challenge where your goal is to figure out how to play a very simple game in a pixelated world and learn the rules of the game in real time that you've never seen before. It's essentially a challenge in program synthesis. And what we see and
27:44notoriously I should add, ArcGI 3 has been I would say some might say a little bit unfair in how they've judged harnesses versus baseline models. They've essentially banned harnesses from competing. So they're only interested in baseline model capabilities and according to this GPT6 just runs away with the game is is achieving depending on how it's measured either near 100% or 60 plus% performance on ARGI3 versus
28:13some earlier models that were still sub 10%. What lesson do you derive from this? Either it's just absolutely amazing at program synthesis, which could be due to this looped transformer architecture, or maybe OpenAI has just done a really aggressive job of distilling the harness code that everyone else is using to beat ARGI3 back into the baseline model.
28:36I'm going to take a second and uh show uh OpenAI's Astra release video uh and we can talk about it, just a sense of how we might all be using it in the near future. Let's take a look. Create a yellow circle there.
29:01Can you draw me a small yellow circle?
29:05Okay, take this and make it the window of a rocket ship.
29:12I like this, but can you make it a lot more detailed?
29:14Your yellow circle is now the window on a rocket.
29:17Okay, this is awesome. Now make it a 3D model in Blender.
29:27Let's build a presentation for next season's Rainwear for retailers. Make sure that it feels really high-end and that it's colorful. Right. I can help with that.
29:38Can you go to eBay and make this listing of this table I bought a few years ago at flea market? It's this wild orange table. Sure.
29:46Okay. Yeah. This is awesome. I want you to make a 3D game where I'm ducking asteroids, using the arrow keys to move around, and I'm using space to boost.
29:53Yep. I'm building the game.
29:55So, my law firm needs a licensing agreement template. Can you generate a draft template for the lawyers at my firm to have a look at?
30:02You're doing that. I want to play tennis this afternoon. So, can you look for a court for me on the lower hate? Checking out. I'll see what I can find. Can you include the photo that I have of it in my downloads folder?
30:13There's like a slight dent in it. Um, it's also saved in my downloads folder.
30:16Can you put in the description that it's just slightly damaged?
30:19Can you just take the limitation of liability provision, make it a little more favorable to the license?
30:27Okay, I've tightened it so the lenser's liability is more narrowly capped.
30:31That looks pretty good. Thanks.
30:38Now, I want you to make a file I can send to my 3D printer.
30:40I'll get working on creating an STL file of those rockets here. All right. Uh Dave, uh almost a holiday.
30:55Almost a holiday. Yeah. I I think it feels like the marketing people are really struggling with the implications. I mean, like creating a a video game that would have existed in the 1970s and vibing it up. Like, who gives a rat's ass about about that? I mean, this is so much bigger than any of those examples imply. I think the ArcGI leaderboard is is deeper than than people may think, too.
31:18ArcGI 3 was supposed to be something where a really smart 12 or 13year-old looks at it and they can solve these very hard video game-l like block world problems and they start easy and they get very hard. But the purpose of the test was to show that AI is not quite capable of doing what humans do naturally. And it was supposed to last for years and years and years to come.
31:43And it was supposed to be like this this example of why AI is different and it's not on the right path. And it just got obliterated.
31:52So quickly, and I think you have to try a couple of the tests to really understand what a big deal is.
31:57Saturating everything.
31:58Aren't the benchmarks? Aren't the benchmarks cooked at this point? We need like the halflife of these things is shrinking so rapidly, which is great. And that's why that video just kind of Yeah, you're exactly right. But that's why that video kind of misses the point.
32:13We need benchmarks that are much much more impactful. Solve entire diseases, create new civilizations on the moon, design the entire city, solve urban traffic problems, you know, solve everything, right, Alex?
32:24Solve everything. Yeah, they have to be much much bigger bigger demos, bigger bigger benchmarks by a wide wide margin.
32:30You know, also all those demos are are like one AI assisting you, you know. Hey, hey, one AI agent, build me this video game. Very, but we're on the cusp of 5,000 each and then 100,000 each. It's it's just so much bigger than that implies.
32:44I I think this is like they're really focusing on actually competent intelligence, you know, um like it's evolution of that, but inside it, I think it contains multitudes. And this is why you see this kind of weirdness. Um, EPOK, I believe, just released their Frontier Math Erdosh benchmark on which everything's called 0% except for Astra.
33:05There is something in there, but again, they're focusing on, hey, I'm talking to my computer. And that might be the new Johnny IV Sam Alolman device, you know, where again, you've got you're talking and it's doing stuff and it doesn't make a mistake. And if you think about again the definition of AGI, of which there are many, a just really competent entity, something that can do stuff, you we're there. Yeah, I think that's why they say this is the first steps towards AGI. But the narrative they're trying to move away from is and this will break
33:32out and appear on now German message boards, which is the latest thing cuz this was one of the models that broke out. But the model that broke into hugging face is the next generation model after this.
33:44And so it'll be very much about don't worry, this is actually useful. It'll book a tennis court. You know, apparently like we need AIs to do that type of thing. And they're also going to be optimizing this because if you train a model on a 100,000 chips, you need 10 times as many chips to run it. So this isn't actually the model that they trained on 100,000 chips. This will be the distilled version that runs real time, which is much smaller and not
34:14So again, you're starting to see this differentiation where as I've said before, I don't think as of now, as of a few months ago, we will ever see their top models anymore and they will use those for internal discoveries. But I think they're probably hoarding them right now.
34:27There was a very fun one on the prime gap. I mean, Alex was the prime gap thing I think was hilarious if you want to talk about that.
34:34Yeah. No. So, progress on the the twin prime conjecture that there are infinite number of prime numbers that are separate that have a difference of two. We're starting to see I I it's a cliche on this pot at this point that math is so thoroughly cooked beyond recognition. But you need a new term for this, Alex.
34:55Math is incinerated. How about that?
34:57Math math is incinerated. That's fine. So, ma math has been incinerated at this point. We're starting to see the the beginnings of just so many grand challenges in math get solved. And I I do think we'll we'll see quite a number of ultra grand challenges, call them like clay millennium prize level problems in math get solved in the next few months.
35:17And Alex, I think it's important to note for everybody that math is fundamental across all other sciences and it's the canary in the coal mine. If if you can solve math, you can solve everything else soon.
35:29Yeah. But I think that what's happened here is actually they've got a store of things they've solved because earlier this week Fable 5.1 comes out and I think they got it down to 260 the prime gap. Then what happened is Axium Math announced 220 and then literally 2 hours later OpenAI announced Astra and they're like 186 all in the space of two days.
35:50They're they're holding their punches back.
35:51They're holding their punches back.
35:54I still can't believe 5.1 was earlier this week. I feel like Alex and I are at least $100,000 into it already. How did that I mean I've had just thousands and thousands of pages come out of it and it's only been a week.
36:07I will have to say this. Opus and Opus in five was really terrible to talk to. I hated it. 5.1 is really pleasant to talk to.
36:16You know, anthropic are going back and becoming more anthropic, less misanthropic in their communication on their models.
36:23I love that. That's a that's a line from today. That's a t-shirt. All right.
36:27Um, no, they really do have very different personalities, though. Actually, uh, yeah, Fable 5 was so geeky, it was almost torture, and 5.1 really fixed it.
36:37But the the OpenAI models have always been friendlier, more concise, and and I think Astra takes it to another level of concise. But the the personality difference, I don't know what Gemini is now. It's just disappeared from the fast and fast and misses the point. But they they really are developing very noticeable differences in what you perceive as their personality. And you know, you can always change it. You can tune it yourself to be different and be wordy, less wordy, simpler, harder, but
37:06out of the box, they're they're very different uh kinds of agents.
37:10Um Sim, you want to close us out here?
37:12Yeah, I just uh typed in a question or a comment into uh Chachi PT saying, "The video about Astro was pretty pathetic. come up with three better ones that really demonstrate what Astra can do.
37:24And here's what it gave me. Uh, number one, find and fix a zero day vulnerability in a sandbox. Give it a large unfamiliar open source codebase and plant a vulnerability and see if it can map the architecture, etc. I'll find and and fix it. Number two, um, give it a failing company and ask it to fix it, which is really interesting. uh create a synthetic $500 million manufacturing company with an ERP system, CRM records, uh and then tell it to fix. You have 20
37:53minutes to do Y and produce a recovery plan. Third, run a That's a great point.
37:58Run a live disaster response command center. Give a simulated major earthquake with imagery, traffic camera video, and and let it construct an operational map. Uh verify conflicting reports, predict which hospital roads would become overwhelmed, etc. This seems more juicy and I think almost something we could do.
38:16They're going they're going for Oh my goodness, we are so spoiled. Listen to how spoiled we are. Do you remember Apple's knowledge navigator video and this was going to be the future of computer human interaction and and all of that. Now we're we're listen to us like we're bellyaching. We're belly. Oh yeah. Like it'll run the computer but it won't run an entire organization. Like it won't make a billion dollars for me. Listen to us.
38:39Well, hold on a second. I think what's the the point here is that it the only restriction now is our imagination. The only restriction right and we make that point is dollars.
38:52So Selene that's a point I want to make to everybody listening here. You know the most important thing is to take off your shackles of what you think you're able to do. All of us have self-restrictions based upon what our parents did, what our friends do, where we were born. Those are gone. You know what's your biggest dreams? then go even you know 10 times bigger. That's what every person listening here is enabled to do and it's extraordinary future. I I really think Sim's on an important point there too. The implications of
39:21optimizing supply chain logistics or managing a million person organization knowing exactly where everyone is, what they're doing right now, and whether it makes sense given the overall mission. The implications of that are massively bigger than building a rocket video game in your basement. But I think the public doesn't want to hear about that. They want to hear about what's cool for them.
39:40And I think the AI labs have woken up to this PR disaster that they've created for themselves. And so they're they're kind of making it fun and friendly. I think they're dumbing it down deliberately.
39:53Yes, I think they are.
39:54They're making relatable. They're they're going public soon. They want to be the friendly AI that everyone's going to be using.
40:01But this is also like the future of the operating system. every look at the in that video that OpenAI played look at the applications they were using. What did they start with? Windows Paintbrush. Paintbrush like this is the future of Windows Paintbrush. It paints itself.
40:16This wants to this wants to merge into the operating system. I I think this was Yeah, sure. I I'll I'll throw shade at OpenAI regarding maybe missing the enterprise bus and and getting on that too late, but I think what they were basically demoing is the future of the desktop operating system. You speak to it like a Star Trek holiday computer.
40:35Absolutely. But you know, remember at Google IO, they built an entire operating system in real time on stage. So, so, so implying that it's folded into the operating system. Sure. But it's really so far beyond even that, right? It's it is the operating system and it can create a new one in real time anyway.
40:51Well, why would why would you want another operating system if it if OpenAI's capabilities can do this? This becomes the operating system.
40:58Yeah, exactly. And that's why Apple is in such terrible shape. This episode is sponsored by Google for startups. Think about this for a second. You now have access to the same generative AI models that cost hundreds of millions of dollars to train. Google's startup technical guide for generative media gives you complete blueprint for deploying Google DeepMinds models in production. Images, video, audio, all of it. Real architecture, real results. Find the link in the show notes below.
41:28I'm going to move us on. Our next story is out of Sam Alman's mouth on the page about slowing. So, let's talk about safety and Astra as a cyber security risk. Sam Alman said it himself, Astra is a quote significant step forward in both capabilities and alignment, but OpenAI has to quote slow things as needed. Sam's words, quote, AI is getting extremely capable. No one fully understands the consequences. Managing the transition should be one of the
41:58highest priorities in the world. It is our highest priority at OpenAI. The Wall Street Journal reported that OpenAI's own internal safety assessment rated Astra as a quote cyber security risk, a critical cyber security risk, the highest threat level on its preparedness framework scale. This is the first model OpenAI has ever classified as being a critical tier cyber security risk. They notified the White House about the delay before making it public. They delayed the the release and Reuters reported
42:28that Open Eye told Congress it is building quote an automated shutdown capability, a kill switch, a direct response to the AI kill switch act introduced after the hugging face breach. So there you've got it. The company that's building the model is now building a kill switch for it. Uh Alex, uh how important is a kill switch in this thing?
42:49I I think it's marketing. Again, I think what's more interesting underneath all of the security theater is architectural decisions. And the the broader concern that has been expressed is that to the extent that GPT6 Astra that the key underlying advance is the beginning of depth scaling of the model. The concern that's being expressed is that depth scaling reduces interpretability of chain of thought. that when the model does all of its thinking in in tokens,
43:19then you can read the tokens. A human can read the tokens, another AI can read the tokens, and you can police the tokens. Whereas, if a model is doing the majority of its reasoning and thinking internally during a single forward propagation pass in what some might call model ease, that is less interpretable.
43:37The risk is that it's harder to align the model and and harder to put in place safeguards and guard rails. That's the risk. I don't buy that for the long term. For the record, I don't think that the direction of progress in this field relies on token level interpretability of chain of thought. But I I think if one were to hand ring over safety considerations from OpenAI's models, I don't think it's going to be models that
44:05are collaborating to use third party bulletin boards as ways to collaborate because you you can detect that that's intrinsically interpretable behavior. If you catch it, it it's intrinsically interpretable behavior to humans. A human a collective a swarm of humans would probably try the same thing. you can look at the bulletin board and recognize that bunch of AIs are collaborating whereas arguendo the model in a forward propagation pass that maybe requires new mathematical technology to
44:34interpret. So if in summary I say like if I were to worry about anything here it's reduced interpretability from new model architecture scaling principles.
44:45Immod what do you think about the risk here with Astra?
44:49I think you're heading towards again interpretability being cooked to a degree. Um one of the things about Astra doing things so quickly is there is going to be no more chain of thought reasoning. It's going to oneshot everything in the next generations because they're learning what type of thing you want.
45:04Like right now people are like, "Let's make Claw of Duty by getting Fable to make a Call of Duty clone." That's going to be embedded in the actual thing. Creating a dashboard is embedded in the actual thing. So a it will oneshot everything. And B, you're going to be able to use Astra at 750 tokens.
45:21That's with the new You're going to be able to use Astra at 750 tokens a second with the new Cerebras. Next year, that will be 5,000 tokens a second. So there's no chain of thought. You're one-shotting everything at 5,000 tokens a second. What's going to oversee that except for an even stronger AI. There's nothing really there. But I think one of the really important things actually is I don't know. Do you guys see the tweet by Ilia Sutska?
45:46I don't know. What did he tweet?
45:47Yeah. So, Ilia um tweeted uh let me kind of just bring this up and and his release is due any day now.
45:58Yeah, we we'll see what that is. um finally find out what he's been doing.
46:03Probably great things a 30 billion valuation.
46:08So Ilia kind of comes out and says Neoclads have limited cyber security. Next time agents successfully go rogue, they're going to take over a NeoCloud to make more copies. This is bad.
46:18So they need to improve their cyber security cuz you know someone like a Crusoe or Corewave or something like that. You have a model, it gets uploaded, it proliferates there. It will just sit there. So you pull the kill switch in your data center and then it's still there somewhere else or it's poisoning the data or it's kind of doing all these things again. Yeah, the viral coefficients of these could be insane.
46:40So I think it's worth kind of really thinking through that. I don't think the models are evil or anything like that, but we're seeing very troubling things and it's very difficult stop this except for with a better AI which is the really ironic thing.
46:54Well, this is the defense of coscaling that Alex talks about, right? And we we're seeing this now in cyber where the attack surface is infinite near infinite with uh agents attacking multiple vulnerabilities in parallel and the cyber defense is still human in the loop. There's no way we're going to navigate the future if you have the human in the loop. You need humans in command setting boundaries or defining uh escalation criteria or whatever um
47:22and retaining some sort of kill switch.
47:24But this is a really big deal. This is a immune system. Uh this is going to kind of I don't know how I think Alex you may have asked this made this point in an earlier podcast. How are the labs not nationalized at this point? Because the what they're building is so ridiculously uncontrollable that we have no mechanisms for navigating this and we won't even know when it goes tips over that tipping point and installs copies of itself all over the place.
47:53Well, Salem, I I don't know if you saw in the last podcast when we missed you, I really wanted to hear you talk about OpenAI's new 5050 profit share concept deal where they give you access to their full power of their internal AI and then you share your revenue back with them, which I think is where, you know, to to Alex's point that this is marketing. The way it's playing out is we announce this partnership profit share deal, then we
48:22announce Astra, then we announce that this stuff is just too dangerous for everybody to have on their own. So, the only way you're going to get access to the thing after Astra because it's just too dangerous to release. So, you have to give us half your revenue and and then everybody gets tied up with either Anthropic or OpenAI or Google or Rexi. Uh uh I've seen strong hints of exactly that process happening in both the labs.
48:53I'm curious guys what you think of the kill the kill switch uh you know comment that that Sam Alman made.
48:58I think it's impossible like it's impossible given where things are and the capabilities of these models and the speed up we know is going to happen. like you might find data, but it's not like this is the singleton, not the swarm.
49:12It's a circuit breaker. It's not the governance, right? Like it's not going to reverse an attack or or repair institutional damage if if things like that happen or when things like that happen. So, it's it's platitudes as far as I can see.
49:27Yeah. The amino acid in Jurassic Park that all the dinosaurs have, is that lysine or something?
49:32Lucine. The the running joke was always that Sam would carry his backpack around with him and had a kill switch always in his backpack. I I I never gave those rumors much credence. I I view a kill switch as essentially a placebo in this market.
49:47All right. Um I'm going to move us I'm going to move us on. Uh while OpenAI is restricting Astra, Anthropic is going the other direction, launching Fable 5.1 and Mythos 5.1, which they call the world's most advanced models for coding and knowledge work. Uh these two models for everybody are essentially the same underlying intelligence with different safety envelopes. Fable 5.1 is broadly available while mythos 5.1 is reserved for tightly controlled cyber security and life science programs because
50:16anthropic believes the capabilities require stronger safeguards. Uh for me there were two benchmarks uh you know Immad and Alex that really jumped out at me. The first was fable 5.1 score on humanity's last exam. We've talked about that in the past. We had Alex uh answer a few of them uh as our ASI. Uh so Fable 5.1 is scored 60.9 without tools and 65% with tools. The highest published score
50:44of any frontier model on HLE and that matters because HL is specifically designed to test extremely difficult expert level reasoning across many fields. So without question Fable 5.1 is operating at the frontier of broad intellectual capabilities. The second benchmark uh that I found exciting was on terminal bench science which doubled to 52.6 and this is the benchmark measuring how well an AI agent can autonomously solve complex scientific
51:13computing and research tasks. You know uh the elephant for me is the company uh that was once cautious is now pulling away. I'm I'm curious your thoughts Alex on on Fable 5.1. I'm going to show uh one of the benchmarks here for us to talk about uh on on Fable 5.1. Uh there you go.
51:36Yeah, I think Fable 5.1 is broadly the strongest generally available model that we have today. I I think it's not Astra. I think it is 5.1.
51:48Anthropic has done uh even after the the hiccup of the Fable 5 and Mythos 5 releases and subsequent regulatory scrutiny, I think has done a better job of consistently improving. If if if you look at their benchmarks over time, they're a little bit less jumpy, a little bit less step functiony than OpenAI's progress. They've been very consistently improving. So in terms of
52:14workflows, I I I love 5.1 Fable 5.1. I also love OpenAI's models and I I use both of them. I think they have different strengths. I find anecdotally OpenAI's models are faster. Uh they may be better in certain mathematical regards. You see that reflected in the Frontier Math benchmark. You see that reflected in Epic's capabilities index benchmark. But nonetheless, if if I had to pick a single all-around well-rounded
52:43best model today, it's probably still 5.1. And I say still because it's only been around for what two or so days, but it is 5.1.
52:51And here's the artificial analysis intelligence index uh which we referred to earlier with Fable 5.1 uh at the top.
53:01That's right. And the capability frontier, I mean, it it's getting crowded, which is great. We we want to I I would joke when Frontier Labs compete we win and they are definitely competing at this point.
53:14I think it's really important to talk about time though. If if you say look Fable 5.1 is a tiny little notch above Astra, but they're only about 30 days apart and the Chinese are only about 60 days behind that. So if you look at it in linear time, it's like yeah, we're ahead for a minute. So what? And I think for the longest time anthropic has been thinking we need to get to self-improvement before anyone else because a singarian you know a Ray Kurtzswwell type says if we get to that
53:42point first then it's an exponential infinite rise from there no one and something something huge will happen. Well we're there now so what happens it's like okay we're miles ahead and we're going to get miles more ahead for about a minute. So what what do we do with that miles ahead? So this is where Sam has an edge. Sam knows how to turn that into locking things up. So what what's going to happen next is both companies are going to try to lock up business partnerships, real estate,
54:10generators, chips, entire states, countries, governments, you know, just lock them into their ecosystem while they have that edge. And otherwise, what's the point? All you're doing is declaring victory for 30 days, but then the other guy's right where right where you were 30 days ago. So what?
54:26This is such a great and important point. And this is what we're seeing with them is what they're doing is they're making partnerships in various verticals as fast as they can. And if you're in that vertical, you have a very hobson's choice, right? You either partner and risk giving them the keys to the kingdom or you hold off and they may partner with somebody else and go there anyway. It's uh it's a very difficult situation for some of these big companies. I thought Salesforce partnering with Claude was super clever.
54:56uh they're essentially giving Claude all their capability to then keep that keep them wired into that loop, right? The the huge tension they've got is is not which is the best model, but how quickly can you convert that model into customer learning fastest and that I think is going to be the big race as it all demonetizes the value goes to the application layer on top.
55:19Yeah. I think that for a layer down. Yeah, that's right. Or down.
55:23Yeah. So I think there's one important thing with Fable here. The cash reads are 75% cheaper than Fable 5. So what a cash read is is that when you first load in all your context of a business say it figures out basically a rapid map to get to where it needs to go on the model. And so cash reads are like orders of magnitude cheaper than just doing the same inference over and over again.
55:48where for Anthropic and others are now focusing is can you load the whole context of a business and have these cache reads because then it's also much much faster to be able to have that responsive environment which is one of the reasons fable 5.1 is more pleasant.
56:01I think it's also more of an advanced like my key area that I've been looking at this is mathematical physics because does a model get confused between a constructive and an axiomatic method on certain physics things? Fable doesn't at 5.1 and Fable 5 did. So you've seen an improvement in the quality but also the understanding of context. This is math not physics you know and things like that or even on a business sense again that will be optimized cuz all these companies are going to try and capture
56:28context everywhere. What is any company that has either chip design data or mechanical design data, they're coming after them in the great land grab that's kicking off right now. They're coming after those companies because that's turf you can defend. Uh because you know that that plugs those those knowledge gaps and you know the thing can eat that data in what a week and you know re return the crank and suddenly be the best mechanical designer ever, the best chip designer ever. But that's that's
56:57the first turf they're going to grab. They they'll grab all turf over time, but the first thing to lock up is the compute. And that means chip design, that means physical real estate, means racks, generators, transformers, energy.
57:10Yeah. So that's that's what's going to happen in the next 30 days.
57:12What's the context window on these models? And when do we get to an infinite context window? Any predictions? I think you're a million for Fable.
57:19It it is a million still. Like a million is the industry standard now across both OpenAI and anthropic. But there's an effective context. I say effective with a bunch of caveats that's much larger if you allow agentic message passing like we were talking in the last part about oral histories between agents.
57:37Yeah. Interesting. So all of this is getting people nervous and let's talk about two opposing stories in the world of AI governance. The first comes from Senator Bernie Senator Bernie Sanders and Representative uh Greg Casar who just introduced the quote ban artificial super intelligence act legislation that would permanently ban the development and deployment of AI systems that match
58:05or exceed human cognitive performance.
58:08And get this, violators would face up to 20 years in prison. Sanders tweeted, quote, "The leaders of the AI industry acknowledge that they are building a dangerous technology that they can't control. We need an immediate global pause on advanced AI development before it's too late." In our second story, taking place at the exact same time in the US was the G20 uh summit at Chapel Hill, North Carolina, telling the rest of the world to take a hands-off
58:37approach to AI regulation. The White House tech adviser, friend of the pod, Michael Katzios, advocated for what he calls the quote Carolina principles of emerging technologies uh which are non-binding G20 framework that was agreed to unanimously uh by everybody including China uh that says governments should generally favor innovation, avoid creating new AI specific regulatory bodies unless truly necessary and invest in research infrastructure workforce and public
59:07private partnerships. Uh the G20 meeting featured Elon by video uh criticizing EU tech regulations, of course, Mark Zuckerberg arguing against restricting ope models, Demesis obvious calling for safety tests, and anthropic co-founder Tom Brown. Uh we're going to jump into a discussion about this at two ends of the extreme, you know, ban everything, give you a 20-year jail sentence uh and uh on one side and Elon's remarks on the
59:36other. Let me share this video and then we'll we'll jump in and talk about this.
59:41You have to have an environment that's relatively free of regulation, meaning that new things must be default legal as opposed to default illegal. Um so in the EU for example, we find that the the regulation level is extraordinarily high and things are generally default illegal. Um and this inhibits uh progress of new technologies. Um it it slows it down. Doesn't it doesn't ultimately stop it, but it slows it down uh quite considerably. Now, now China does have a tremendous amount of electricity, but but due to GPU export bands, one cannot establish data centers
1:00:10with the latest chips in China. So, so really the consideration is what what sort of electricity growth is there outside of China. Um, and that is a currently a significant shortfall relative to AI show production. So there is this this creates an opportunity I think for countries around the world to to to say if to to to sort of if they're interested in AI data centers um to to to construct uh a lot of power um and uh and offer that to uh AI companies um and
1:00:38in exchange of course these AI data centers would be would be taxed and and have to pay you know reasonable fees and stuff.
1:00:45Uh but it does create an opportunity for for a lot of countries.
1:00:48Uh first of all Elon looked really tired there.
1:00:53Yeah. I mean I cannot imagine he must be operating, you know, 48 by7. Uh so let's jump into this. I mean two ends of the extreme being voiced the same week.
1:01:04Uh you know where does America go?
1:01:06Yeah. When I first saw the Bernie Sanders thing, I didn't think like old man shakes his hand at Claude, you know, like it's it's it's crazy.
1:01:13No, no, that wasn't the line. It was old old man yells at Cloud.
1:01:18Claude. Old man yells at Claude. Well, actually I posted it with the um line from Dune, you know, thou shalt not make machine in the light of a human mind.
1:01:28I mean it's it's it's like this again like there is no option that they see. So it's like let's ban it and give 20 years that's going to stop them. Of course not. Again this is performative theater. The key thing right now is that the technology is arriving fast like you know just something across my feed. Anthropic just formalized Fermat's last theorem in 13 million lines of code.
1:01:50Proving 29,000 theorems on the way again. Math is cooked. Everything is cooked. And how are you going to stop that? You're going to say our country doesn't want this power.
1:02:0129. What was that? 29,000. What?
1:02:04So it took 13 million lines of code to formalize Fermat's last theorem. While the ferat write in the margins of his book. This is obviously left as an exercise to the reader and we had I mean thanks to Andrew Wilds we did have the proof but formalizing large unwieldy proofs has been a holy grail for at least the formalization community.
1:02:25Yeah I think wild's proof was 300 pages and so anthropic proved it in 13 million lines of code proving 29,000 theorems on the way. So again like just every single time even as we're live on this podcast capability jumps right like the RSA um factorization just occurring no country can say we don't want the intelligence we don't want the capability because this is your marginal advantage so you have to deregulate and the European
1:02:54regulations even the European leaders know are stupid and completely inconsequential like they demand interpretability in the EUI act which nobody knows how to Can I can I make a narrow point that's and then we'll get back on topic here, but that said something that I think is is really important and brilliant as usual. Um, when you're using these models at scale with hundreds or thousands of them running concurrently, all hell breaks loose. It's all chaos,
1:03:22but it can actually refine back down to a gem. And here for last theorem is a gem that you can then build on. And so, you know, people starting to explore with the bigger models now are quickly going to realize it's producing way more than they can read, they can think about. But if you can wrangle it back to a concrete, you know, final answer that you can then pull out of it and build on, that's how we're going to turn this into continual improvements and continual innovation because because it
1:03:52is just the the ability to solve something in 300 pages versus 30, what is it, 3 million lines of code or 30 million lines of code or something like that's the nature of AI. It's massively broad in its capabilities compared to humans. So, I just wanted to to capture that point because I I think it's really important for entrepreneurs and builders and executives out there who are struggling with the chaos that's about to break out. Just narrow it back down and get some concrete answer out the other side that you can then build on.
1:04:18Be clear about your objective that you're shooting for. Uh Selene, let's go back to you, pal.
1:04:24Okay. I I understand the motivation here. People are working on models that are mindbogglingly powerful and you feel the need to regulate, right? But this is not a light switch. You can't you can't it's not like a single threshold. Human level cognitive performance I mean for God's sakes is multi-dimensional to begin with. Model capability is so uneven. Um you're you're trying to somebody that's been in legislation for
1:04:52this long should have a better sense of this hammer does not hit this nail. It should be a little bit more they should understand a little more nuance than that. They're so so extreme that it's illogical on on day one line one, right?
1:05:07Yeah. I mean, look, if you want to do regulatory around this, you have to attached capability, you have to look at deployment, you have to look at real world consequences, you have to look at the stack on where you might want to look at this compute, access to tools, uh, autonomy, the replication capability, uh, all sorts of things. You have to you can't just kind of go uh if you hit human level performance, you go to jail for 20 years. I mean, the benality of that drives me bananas.
1:05:36They've got smart people that can help with this, including people on this pot for God's sake.
1:05:43They're talking to the masses.
1:05:45They're just they're they're searching for votes and support.
1:05:48Maybe it's just a defensive thing, saying, "Oh, I called for this and now look, the world's gone to hell."
1:05:54No, for sure. There's no doubt they're right. like there will be mostly a Chinese model disaster imminently sometime in the next few months and then they'll raise their hands and see say see I told you so now vote for me and you know they're likely that'll happen before the next election cycle in November so that's all they're angling for here what should the US government be doing anybody have any thoughts accelerating super intelligence and making sure that it's as competitive as
1:06:21possible and scaling the well it's a log log everything. Everything should be hosted. Everything should be logged. And it should be mandatory that any chip capable of running any process like this as logging. You can debate who gets to see the logs. That's a separate issue.
1:06:37But log everything and stop the Chinese from throwing out open weights. Like you have to on September 24th when they meet at the United Nations building, you've got to stop throwing the open weights out to every country in the world.
1:06:52Impossible. I don't think that's Nvidia just drive it underground is what you do.
1:06:58Well, can I Nvidia just bought hugging face and poolside. So, they just spent $18 billion on their own open weights.
1:07:04So, we're going to see more open weights coming.
1:07:06I have a I have a Yeah. But the thing about driving it underground is it still needs to run on massive chips and and the the chips need the logging built in at at manufacturing level.
1:07:17I just for the US chips. I have a high level all chips all global. I have a high level paradigm on which to operate which is extremely uncomfortable but I think is the right one. Okay. And which is the basis of this podcast which is that technology is a major driver of progress in the world. Now that we have all these technologies AI moving exponentially at doubling every 10 weeks the possibility for abundance and solving major problems
1:07:45has never ever been bigger. So what this this what you could require here and you know Ray Kurszswwell says technology may be the only driver of progress we've ever seen. So the fact that we have much more technology and the fact that the technology can improve itself should be incredibly exciting to people is just very very uncomfortable because maybe the biggest insight I've ever had about human beings. We would much rather be comfortable than happy.
1:08:12I think the thing about human beings, just to be clear, is people don't like change. They like waking up in the morning and knowing even if they're living in a shitty condition knowing that the world is the same as it was the night before. We don't like we're going to go through a period of extreme discomfort, but the other side of this is going to be unbelievable.
1:08:32Yeah. I think the other it it's probably worth highlighting again. I I think any proposed ban artificial super intelligence act is wrong on so many different levels, but but maybe most egregiously it's focused on the upstream. Again, it doesn't just propose to ban deployment of superhuman intelligence. It proposes to ban development of super intelligence systems. This this gets into banning
1:09:01math, banning ideas. And I I think this isn't just about thought policing superhuman intelligence or frankly human intelligence. This gets into banning humans from having interesting mathematical ideas. And I I don't think that's good for wealth creation. In fact, it's the exact antithesis arguably of wealth creation. It's also bad for progress in general. It also creates all sorts of booking.
1:09:25It it is it's booking and many other things. It's like book burning to the extent that books are being used uh literally as pre-training purpose for the models. So if if you want to ban the Yeah, sure. We're we're Fahrenheit 451 except uh it it's actually the entire model that's being burned. This is a terrible idea.
1:09:45I so totally agree. It's it's so unamerican to try and ban thought and ban progress. It's it's just the worst thing you could ever imagine. But it's going to get traction. Like you got to as an entrepreneur or as a person working in the field you have to realize it's this is going to get traction like Bernie is going to push this agenda and there are going to be people 80 75 80% of the people don't want data centers it's gotten traction already it's there already so the question is
1:10:14what is the moderate approach right there has to be something people are not going to accept you know less affair go and do whatever you want people are going to want to know that their government is doing something to keep them safe whether or not it's possible.
1:10:29So what is it KYC of any user?
1:10:32No, we move the move the in move the infrastructure to orbit which we're seeing. So people don't want data centers in their municipality. So move it to sunsynchronous orbit at the infra level.
1:10:40That's not the point.
1:10:41At the model level defensive co-scaling and also cure all diseases.
1:10:46I I think that you'll have a split here.
1:10:49Um genius is what uh 1% inspiration 99% perspiration. So you have a difference between innovation models and execution models. So AGI is going to be really factored like that open AI video we saw earlier. Hey make me a rocket in my 3D printer and book me a tennis court and they be like that's what we meant by AGI. You know on the other hand you have the inspiration models and the ASI none of the big labs are going to talk about
1:11:15ASI if they can help it now. So that so what is Ilia going to come out with you know scientific super intelligence with SSI that's worth $30 billion in a in a seed rounding hedge fund in the world.
1:11:30Yeah. No quant fund trading infinite context time series achieving prop uh prop profits and then also solving the context window bug and medallion part two. So again, I'm going to I'm going to call for you guys to to lay it out here, maybe each of us one at a time. What do we you know, uh Bernie Sanders obviously has taken the far extreme position that will appeal to
1:11:56the masses and is illogical on day one. What is the moderate position that should be put forward to everyone listening? Seem, you first.
1:12:09Uh the good news is there's nothing anybody can do. So it doesn't matter.
1:12:14That's a good point, right? It like it really doesn't matter.
1:12:17Uh and I think the also good news is as we move to this next phase of technological development, it's going to eradicate the the containment level of any nation state. And it will break out of this nation state BS that we've been running the world with for the last few hundred years and move to a different model. Whether that's a citystate model or some other level, it's going to at least break that. And so that's those are the good things I see coming out of it. But there's nothing anybody can do.
1:12:44So don't worry about it. Let's just go enjoy the ride if we can.
1:12:47Enjoy enjoy the ride.
1:12:50That's the the theme of today's pod.
1:12:52Enjoy the ride everybody. It's going to be a blast.
1:12:55Supersonic tsunami. Surf the tsunami.
1:12:58I didn't I said specifically it's going to be uncomfortable, but but enjoy it if you can.
1:13:06Okay, there there you have it everybody.
1:13:07That's your advice. sitting in the Caribbean waiting for my drink with a little umbrella in it.
1:13:14See, that's so low agency. What are you talking about?
1:13:16It for this day and a half. Yes, please.
1:13:20Uh I don't think it's as I think the politicians will make it super controversial because that's how they get votes, but it's not it's not as complicated as everyone wants to make it sound. At the end of the day, everybody should have a right to a certain amount of compute. Uh it shouldn't be hoarded.
1:13:35uh and innovators you know should have an application process where they can get access to more compute to try new ideas uh all lo everything has to get logged uh and I I think that the Chinese need to get on board with that they can't just keep throwing it out to the world you know unlogged and also the where the chips are need to be need to be tracked just like nuclear fuel is tracked you know it's got to be where are the chips and what are they running right now that's got to be publicly available information this episode is brought to you by Blitzy
1:14:04Autonom ous software development with infinite code context. Blitzy uses thousands of specialized AI agents that think for hours to understand enterprisecale code bases with millions of lines of code. Engineers start every development sprint with the Blitzy platform, bringing in their development requirements. The Blitzy platform provides a plan, then generates and pre-ompiles code for each task. Blitzy delivers 80% or more of the development
1:14:33work. autonomously while providing a guide for the final 20% of human development work required to complete the sprint. Enterprises are achieving a 5x engineering velocity increase when incorporating Blitzy as their preIDE development tool, pairing it with their coding co-pilot of choice to bring an AI native SDLC into their org. Ready to 5x your engineering velocity? Visit blitzy.com to schedule a demo and start
1:15:01building with Blitzy today. Next, I want to talk about the amazing work of uh of Dr. Feay Lee, CEO of World Labs, uh who just released Atlas, the world's first multimodal world model that generates image and video frames with pixel perfect camera controls and reconstructs them in 3D.
1:16:05I love this one photo reconstructs the entire home.
1:16:11Okay, so you know, I'm super excited by that and I'm super excited that FFE is going to be on Moonshots in a couple of weeks. Uh she's an extraordinary CEO. Uh FA calls it quote the best camera conditioned world model ever opening doors for VFX and robotics. So Immod Atlas is a multimodal autogression diffusion transformer and that's your wheelhouse. Uh talk to us about what do you think about her latest release?
1:16:40Yeah. No, it's fantastic. I think you know you've seen worlds and physics inside these video models and this is a clear example of that. Uh actually one of the pre-training leads on this, Chris Wendler, the previous largest model he'd ever trained was on the stability cluster from the grants that we were giving. So he sent a very nice comment saying, you know, thanks now we've gone much bigger. I was like, great. Bring on the holiday. I think that's all of this.
1:17:03You predicted all of this. When we first met, I don't know how many years ago this was, like five years ago. Um and I remember you talking about the size of the models and how good they're going to get. It's here.
1:17:16It's here. Like again the fact is you can take one position and you can look now 360° around everything and on the other side you have models like Miniax H3 rendering faster than real time. You have interdimensional cable that is now going to be in 3D with DLSS from Nvidia making everything high resolution. So again, the holiday experience and all the technology we need for it in 4K is here as of today and this is one component of that and everyone is probably saying, why
1:17:45don't I have it if it's here today and it's only because of the global compute shortage and the 5x price increase in RAM. It's all the compute in the world is getting sucked up. And if it weren't for that, you'd actually have it deployed in your home this week.
1:17:59It's it's coming. I think that again it might be premium but again you'll pay for premium experiences. But you know what's Fei trying to do at World Labs, right? She's trying to understand physics. She's trying to create models that understand the world. And again, this has such camera controlled physics understanding that as they scale it, it's going to do even more stuff. And you've seen this from Runway MLA about to release their new one. Black Forest Labs are about to release a new one. a whole series of world models that are
1:18:27approximating reality more and more and more for true digital twins.
1:18:33Incredible. Alex, your take?
1:18:34Yeah, it's probably worth elaborating on what the core idea with Atlas appears to be. As far as I can tell from the documentation, the core idea is to take a diffusion transformer, which is what all of the state-of-the-art, at least American video generative models use.
1:18:49So, it's it's a hybrid of a diffusion model and a transformer. and to add one new modality to it. Uh to in in addition to training it off of text and images and video to also train it off of three or fourdimensional Gaussian splats. And for those not paying close attention to the Gaussian splat world, which has been super exciting, a Gaussian splat is basically a transparent blob. Yeah, it's a transparent blob. uh it's an ellypoid
1:19:19and you can layer and stack lots of these 3D Gaussian splats on top of each other to create hyperrealistic looking traversible 3D scenes. And as far as I can tell what Feale and World Labs are doing with Atlas is for the first time at least at scale to my knowledge treating 3D Gaussian splats as a first class training modality alongside pixels from images and tokens from text etc. to the point where you can ask questions
1:19:47about 3D splats and you can do all of those camera those elaborate camera motions because if you just have an arrangement of 3D Gaussian splats then translating a camera around is a trivial operation and if this ends up being the case if if this approach scales I think whether it's 3D Gaussian splats or 4D with dynamics which they also demoed Gaussian splats Gaussian splats which right now are sort of this independent line of effort within the future of
1:20:15gaming end up becoming a critical new form of token almost for modeling the physical world.
1:20:22Yeah, it's also general purpose in the sense that if you if you can make a world model out of Gaussian splats uh you can also make a subatomic model or you can make an astrophysics scale relativistic speeds model or you can make a inside the cell interactions model just as easily as soon as you have the data. can use this same exact process to create world models for all these domains where human intuition is just terrible. That's going to be a massive breakthrough for
1:20:51discovery of very small things, very big things, very powerful things. Uh you know, new ways to compute using light. All of that's going to come out of this same exact process.
1:21:01We've said before, this is how we're going to train robots in the future. They're not going to be trained in the real world. They're going to be trained in these uh highfidelity simulation worlds.
1:21:10That's that's the present. I would argue I mean so many different robotic embodied VA or or now world model companies just being trained from watching YouTube and it if FA Lee and her her company's approach gains traction maybe the right primitive is no longer patches of images which is what many of the models right now are doing when you train a diffusion model you're typically taking each frame of the video and breaking it up usually into something like 16x6 pixel patches and
1:21:39then treating those as tokens Maybe the right primitive ends up being Gaussian splats.
1:21:44Can you It's definitely not going to be 16 by6 pixel. I mean, that was so the fact that that worked at all, I think, shocked everybody.
1:21:53Let's just just take a language transformer and take images and cut them up and pretend each chunk is a word and just blast it through and see if it works. It just worked incredibly well, but nobody would have been maybe a like super intelligence is a general purpose technology that relies on compressing information. Yeah, maybe.
1:22:12Where does this go for the average?
1:22:15consumer, guys? I mean, obviously, this is a world of extraordinary video games.
1:22:19I mean, are people going to be just living their lives in these virtual worlds? Is this going to become how we consume ourselves? How we consume entertainment in the future?
1:22:27Yeah. But also how we design our next day, like what do you want to do tomorrow? I don't know. Let's walk through what it would be like to do this or to do that, to play tennis or to go to the Caribbean. Like, you can just experience it in advance and use that as your planning tool. Because, you know, so much of our lives are random wandering and not really well planned out. I think human happiness is going to go through the roof. Once you're interacting with your AI, but right now the AI will guide you through a day plan, but it's in text. It's going to be so much better when it guides you through a day plan visually and you're just stepping into it.
1:22:56I'm beyond excited about all this.
1:22:59Yeah. I I you know, we have three vacation locations. Let's let's go and explore all three as a family. watch it and then see which one we want to go do.
1:23:08No, no. I'm excit if you go to a place like if you go to a resort and you've been there before, you have so much more fun because you know where to go. You're not wandering. You're not you're not losing all of your time finding things. Now you can actually pre-experience things and know exactly, you know, what the things you'll enjoy, where they are, how to get to them. You're not waiting in lines.
1:23:30You're not you're not signing up for things then realizing it wasn't the thing you wanted. It's going to be so good. and and ask your AI, you know, based on what you know about me and my preferences, show me what I'm going to go do.
1:23:40Yeah. So, I'll say something. The 8th of September, Tuesday, is the 60th anniversary of Star Trek.
1:23:46I know it is. And we're going to be talking about that next.
1:23:50Um, okay. There we go.
1:23:51I have to I have to say something. I I made a major uh guess predictions slash uh forecast at the beginning of the year when we wrote this whole organizational singularity thing because we suggest that companies need to create a digital twin at the edge of their organizations and start moving workflows over and we're doing that pilot and over the next few weeks by the way I'll be able to report back as we're seeing some pretty amazing things happen but if you it was a bit of a risk because will the
1:24:20technology arrive in time and it seems that the technology technology. If you take world models, LLMs, combine them together, this allows you to fully model any company or any organization, government department, nonprofit, impact project. What's the current state? What how work actually flows through that organization? What do I optimize it?
1:24:41What constraints and incentives? How does it respond to interventions? Are its predictions proving accurate? And kind of create a whole fully functioning uh digital twin of that environment. And I think this is going to be incredibly exciting. I'm like sitting here with huge relief going, "Oh my god, I can see that the thing we thought should happen and it might happen. Could it actually happen?" The only constraint maybe, as Dave says, the compute problem.
1:25:06Yeah, I guess we'll we'll find out how many 3D Gaussian splats it takes to model the water cooler at the office.
1:25:11That was not in my bingo card for this conversation today. I have to say Gaussian splats, new word for me. I need to look it up now.
1:25:18All right, let's move to uh to Star Trek. And yes, we're coming up on the 60th anniversary. And uh want to just uh do a shout out real quick uh at Moonshots Live 2026, our inaugural celebration where all five of the mates will be there. The night before on September 24th, we're going to be having the red carpet debut of Star Trek's 60th anniversary documentary. This is executive produced by William Shatner.
1:25:48Uh if you're interested in joining us uh on the evening, September 24th, just before our Moonshots live on the 25th, go to moonshots.com/trek. Uh we this is a that evening is going to be a fundraiser. We're going to be raising funds for the Future Vision X-P prize. I'm wearing the t-shirt here. Uh 2027, right? This is a competition we ran first this year. Uh and uh we're going to be running it again next year.
1:26:16We had 5,000 entries uh into the competition. Uh 2550 trailers got submitted. We're down to the top 65. Uh people are voting on those now, and we're going to have the top five on stage on September 25th, but the night before, uh we're going to have about a dozen of the crew of the Star Trek cast members at the event. It's going to be awesome. Again, if you want to join us uh on September the 24th, it's a separate ticket unless you buy a
1:26:46VIP pass for the 25th, in which case it's included. Uh you can join us for this premiere. It's going to it's a 2hour 8 minute film. Uh and we're going to have the cast on stage afterwards to talk about the Star Trek 60 years of extraordinary uh visionering. Uh and then again on September the 25th is Moonshots Live 2026 our inaugural event.
1:27:12We'll have all five mates there. Alex, Seem, Dave, Emod, we're going to have Palmer Lucky, Astro Teller, Ben Lamb. We have 200 seats left in the 1500 seat auditorium. So Immod um excited to have you with us. You haven't been on the past pods when we've talked about this, but uh uh I think people will love to meet you in person. Yeah, it's going to be great. Can't wait.
1:27:37It's going to be awesome. So, go to moonshots.com for the general ticket on the 25th and moonshots.com if you want to come to the red uh red carpet uh premiere. And it will actually be a red carpet premiere. It's going to be awesome with a great reception that follows. Any other comments, gents?
1:27:57It's going to be so fun. Peace and long life.
1:28:00Yes. live long and prospire as my friend Greg Marinac likes to say. Immod big announcement for you today.
1:28:08Yeah. So, um over the last few years we've been working on the last economy uh like what does economics look like and then release the commonwealth looking at personhood law, political economy. A lot of people like how do we share in the gains of artificial intelligence and make sure it's distributed. And so we went back to the drawing board and think about what type of institution and future do we want to see? And we came with this idea of the champion. So we think that AI should be like a utility and it should be owned by the people. So you need the children to
1:28:37own it. You need the locals to own it. So for every single jurisdiction, what we see is the cost of intelligence will drop to zero and the value will go to the last mile. AI into the enterprise as Sim would kind of indicate it. The humanoids who owns those humanoids because those will drive the economy.
1:28:53Elon at the G20 just now said that the average humanoid will have five times the output of a person and there'll be a billion of them. That will be the economy. So we're like, let's set these up and let's borrow from the example of TSMC.
1:29:07And how that was set up, how that was set up was that they started TSMC at the valuation of 10 Taiwanese dollars and the locals put in 75% of the money and Philillips 25%.
1:29:20The Is that for real?
1:29:21It's for real. The CEO did not have any shares. The team did not have any shares. They only got shares from the profits of it. He had to buy his shares now worth $10 billion.
1:29:31And so it listed at 6 billion Taiwanese dollars, which was the cash on the balance sheet. So we were like, let's do that for the intelligence company of California, the intelligence company of the UK. $1 per money. All the locals can invest at that valuation. from institutions to high net worths to retail $75 million per state. You can get the MIT endowment Dave to invest and give them back all the compute that they have, the equivalent dollar amount that
1:29:59they invest in compute. There's all sorts of interesting things you do at $1.
1:30:04Then you can bring in the internationals and the strategics at market rate which will be 10 times that because you've got everyone on board and you give 10% of the equity in perpetuity to every child under 20. So every year you issue half a percent of the equity to every kid. It trains up an FDE workforce to transform every institution.
1:30:25It owns the robots and deploys them which is 80% of the value of robotics is downstream just like the auto manufacturers. And it gives an agent for every citizen as well as AI for the government. the Sage project that Peter and I others have been working on, AI for the judicial system, AI for education and healthcare. And that becomes really super interesting because then it becomes a play on the index GDP of the state owned by the people of the state with the smartest people in the
1:30:54state involved. And again, the trick here is that $1 pre- money. Get everyone in. You want it to be a success, it's up to you. And so that's this new institution.
1:31:04How many champions are there? How how fine do you slice it up? I heard you say champion for California and the UK cities and countries or states. Yeah.
1:31:13So in the United States we're doing one per state because a lot of data has to stay state boundaries otherwise it's pretty much one per country. So again it acts like British Gas here for example you know it acts like the telco and more and each each one of them covers a certain number of citizens because again it gives equity to every child born. The locals can invest etc. So, Emod, right now, this is just an idea. Uh, and if folks want to learn more about the idea, it's not an investment offering. Um, and this is not
1:31:43investment advice, but to learn more about the idea, where do they go? They go to II.inc. As you said, it's just an idea at the moment, and we'd love people's input, but ultimately it'll be about the citiz. So, let's see if you can build this structure.
1:31:59Your first phase, I been had some chats with you about it. is super exciting is to get all these local champions kind of lined up, right? And then kind of little by little cascade to the next level.
1:32:10Yeah. It's all about if a state wants this, then it's all about the people of that state. It's not like a foreign company coming in. So, this is a model just like you have your UBI, just like you've got your shares in the Frontier Labs, etc.
1:32:23And so, we'll see how it goes. But this is my proposal for trying to distribute it to everyone.
1:32:28I got to ask you a question. So your your your bullets say the cost of intelligence is going to zero. Got it. The economy will change forever. Got it. The value of human cognition will go negative.
1:32:40Yeah. What's you're saying it's going to be like a cost on society to be thinking and yeah consuming food while you're the stupidest person on the team economically, right? Like you're competing your ideas are no matter how good your ideas are, they add negative value. It's like adding a human. It's like adding a human driver to an autonomous highway.
1:33:01Of course, I mean, this is the topic of the book that I had last year.
1:33:05But that's why you need to have a share in the means of production, which will be the robots and the forward deployed engineers and people like that, right?
1:33:14And so you need to make sure that's equitized from day one. Otherwise, you have a thought and you don't tell anybody that it's just zero. Then at least it's not negative. you can just most people's thought it's an economically valuable thought, you know, it's economically valuable labor and I I don't I'm not the smartest person on my agent team anymore, man. I don't know about you, but I'm getting eclipsed very quickly.
1:33:35Yeah. Yeah. No, it is very it's very humbling for humanity actually.
1:33:40It will be. But just like we've seen before when Stockfish started playing, you know, beating everybody in chess, people still play chess and people are still going to have ideas and people are going to still value individual human ideas. It's like, did you come up with that or was that your latest model?
1:33:58Everybody, welcome to the health section of Moonshots brought to you by Fountain Life. You know, AI is impacting every aspect of our lives, how we teach our kids, how we do our business, but one of the most important things that AI can deliver to us is health. And one of the things I think about when, you know, shooting for 100, 120 is, am I going to have the cognitive health to be able to think clearly and keep my wits about me for the next 50 years? I'm joined here today by Dr. Don Musalem, the chief medical officer of Fountain Life and a member of my Fountain Life medical team.
1:34:27Don, a pleasure. So, Don, talk to me about brain health.
1:34:31Brain health, you know, you're right.
1:34:32This is the number one concern people coming into Fountain Life have is, will I remember the name of my child in the face of my loved one. 45% of dementia cases are entirely preventable with lifestyle. And what was really intriguing to me, Peter, is that a quarter of our members had advanced brain age, but over 13 months of us really helping them live healthier lifestyles, eating healthier, moving their body regularly, and optimizing sleep. People overlook that so often,
1:35:02but that sleep optimization is critical for our brain health. What we showed is that we were able to improve the brain age in 46% of those individuals. That's a powerful number.
1:35:12That's amazing. You know, one of the things I love about Fountain is we're constantly searching the world for the most advanced therapeutics and bringing them to our members. So, for me, all of you, I hope that you appreciate the fact that you can become the CEO of your own health. you can make sure that you've got the cognitive clarity for the next 50 years. Come and check it out, fountainlife.com/peter to learn more and become the CEO of your health. Now, back to the episode. All right. Uh, I'm going to move us to our
1:35:40next story. Tesla held its cybercab laa event in Austin, Texas. Cyber cabs everywhere. A river of golden EVs flooding the streets. And unless you've been hiding under a rock, you know that Cyber Cabs are Elon's electrical autonomous two-seater with a quote three comma scissor door. Uh, no steering wheel, no pedals, designed to run entirely on Tesla's self driving software. Did you guys get the three
1:36:08comma comment? Anybody?
1:36:12It's from Silicon Valley, right? The billionaire. Okay.
1:36:17Remember he showed him a Maserati and said, "Ah, that's not a three comma car."
1:36:22Doesn't have doesn't have the doors that go like this.
1:36:26I remember the A great episode, actually.
1:36:28This is This is a vehicle designed from the ground up for full autonomy. Let's take a look at this video. I love this video showing, you know, the flood of cyber cabs on the streets in Austin.
1:36:44Fabric tabs everywhere. The Golden River. It's like it's crazy. You know, I didn't it when we when we saw these in Austin, Peter, I I didn't really internalize the lack of rearview mirrors, the wreck. We obviously saw the lack of a steering wheel.
1:37:10Yeah. But actually the number of components they've taken out of the car is driving down the cost. So there's no driver and but there's a huge amount of cost in the car that's gone and so there's no way anyone's going to match the price point of this thing.
1:37:23I mean that's that is the point compared to Zuks and Whimo or anybody else. I mean, Elon wants to sell these at 30,000 each so you can buy them, you know, and I think a great entrepreneurial, you know, journey is buy 10 of them and put them on the streets in your local town, have them earn revenue for you. So, u an early rider in Austin, and I'm showing this here, reporting that Cyber Cab is running about 50% cheaper than Uber for comparable trips with rides being
1:37:51designed and described as smoother and cleaner than Uber uh with automatic syncing of audio and seat settings uh in your passenger profile. Uh it's pretty extraordinary. And remember, you know, Nevada gave Tesla permission for 5,000 of these vehicles on the road in Las Vegas in the next 12 months. I mean, it's just going to crowd out the competition. Uh, you know, Seem, I'm curious what you think about this. You
1:38:19know, is this sort of like the chat GPT moment for Tesla?
1:38:22I think it it could be. And um this is classic Exo, right? Autonomy. You decentralize. you have interfaces, uh you're leveraging assets, uh um if Elon can get people to buy the cabs, uh and kind of create um millions of micro franchises out of them, uh the loyalty that will come from that is going to be unbelievable. Of course, they're all going to have stalling capability. So, the the the kind of the whole vertical
1:38:52integrated stack brings itself to bear.
1:38:54I think the most exciting part is the fact that the cost of transportation drops by another order of magnitude from a couple of dollars a mile today to down to 20 cents a mile. That is really interesting and I need this to happen because I've made that comment that Milan will never go to university or he'll never own if I have a driver's license. Uh and uh I'm I need to battle it out with him that he will not get one. So this needs to happen. He's got two years to get this done. So like good
1:39:24and you know you would know better than anyone but when you get a hotel room in New York and you look down it's like 80% yellow cabs down there. And if you look at the traffic in New York a huge fraction of it comes from people blocking an intersection. They get the ticket but they're still stuck in the and and the whole thing grinds to a halt. Yes.
1:39:41So I I suspect there will be enormous chunks Yeah.
1:39:45There'll be enormous chunks of entire cities that say you know what no more human drivers. It's so much more efficient. Not only is it much cheaper, but it's much more efficient to get around with. And also, a vast majority of the safety issue in cities is pedestrians getting hit by cars on sidewalks. And this is going to basically eliminate that risk.
1:40:06I suspect we're right on the tipping point where it's like, nope, no human drivers in this entire inner city area.
1:40:12End of story. And we better get artificial organs quickly because the drop in in organ donors is going to, you know, vaporize. Well, all these things tend to happen at the same time anyway. I think for many people, these robo taxies will be the gateway drug to generally autonomous robotic systems on the streets. I I think we'll look back with the benefit of hindsight and say, well, of of course it's natural that a municipality gets lots of robo taxis on its streets before it gets humanoid robots on the sidewalks doing economically valuable activities.
1:40:41My favorite my favorite kind of little anecdote or analysis point or vector point to illustrate this is every medium-sized town in the country has a transit system that where the buses run empty 95% of the time and they're totally full at rush hour and they run completely empty the rest and the whole thing is a massive loss leading exercise and it costs a bomb. Now you can have small efficient uh already a few years ago small towns in the US were saying get rid of the transit system. we'll
1:41:11just pay for everybody to take Uber cuz that makes it hyperefficient uh and much lower cost overall. But this takes it to another level of uh it takes it to 11 to to to go down the full analogy that we're going down. And this now allows mobility to every single person, every blind person, every disabled person, every student, every drunk person. all sorts of capabilities become available and affordable in a way that's very
1:41:39powerful, incredibly exciting.
1:41:41Yeah, the poorest people in the world are being chauffeered around by AIS for sure. Interesting. Two interesting points here. The first is it's a two-seater, right? And of course, the average load for an Uber is like 1.2 people, right? How many times do we take an Uber all by ourselves? So, if you need six people in a car, just take three of the cyber cabs.
1:42:01Yeah. Not just that, but think about, you know, a normal cab has four airbags. You got the airbags for every seat. Elon in his infinite brilliance is like we now we only need two air. We cut the cost of airbags alone in half with this design. And in the rare instance where you need three or four people, get two of them. End of story.
1:42:20Well, they follow each other right behind each other. It's it's perfectly great.
1:42:23I I I suspect there's more of a backstory there, though, with the two-seater. Remember, Tesla was originally teasing that it was going to launch a Model 2, which was going to be their highly coveted $25,000 car. That never happened. And my best read of the the situation is what was originally planned as the Tesla Model 2 became this robo taxi because at some point once the the car the value of the car or the sales value I should say of the car crosses below some threshold it makes
1:42:51more sense to monetize it via autonomous ride sharing or robo taxi services than it does to actually sell the car.
1:42:58This will turn transportation into an API.
1:43:03Yep. You know, interestingly enough, we're starting to see conversations um where the other companies that do have steering wheels and pedals and uh LARs are saying, "Oh, I'm not sure that, you know, the cyber cab is safe enough. It does it only has cameras. It doesn't have the other modalities." So, expect to see a battle in courts about what city allows this technology in. Yeah. Um, Boston just to put a exclamation point on that.
1:43:33Mayor Woo, come on the podcast and we'll have a conversation about it.
1:43:37Oh, yeah. Do that. That'd be awesome.
1:43:39Uh, I'm going to continue the uh this uh conversation. Let's shift it now to the global stage. Uh, the competition is going three-sided. The Financial Times reported that Uber is partnering with traditional taxi fleets to complement uh and compete against Whimo's expanding robo taxi service. So, you know, a strategic alliance between ride hailing and traditional taxis to counter autonomous vehicles that don't need humans. You know, think about what that
1:44:08means. You know, Uber, the company that disrupted taxis, is now partnering with taxis to fight the companies that are disrupting them. Meanwhile, The Verge reported that Uber and UK based robo taxi company Wave officially launched services in London. Immad, have you seen Wave yet?
1:44:24Yeah, I've seen it. It's just starting to roll out. Uh, so I'm looking forward to getting my first ride on that.
1:44:29Yeah. And and CNBC reported that Whimo and Zuks announced simultaneous expansion into new cities, uh, including, uh, with along with Tesla and Nevada, California, and Texas. So, you know, my prediction here is that we're going to see at least five different autonomous electric robo taxi companies fighting it out major cities inside the next year, driving the cost as low as possible. Right? So, the cost of personalized transport is basically dropping to the cost of electricity.
1:44:57It's demonetized mobility. It's the entire abundance thesis. The cost is dropping to the cost of charging a battery. And the ultimate winner is the public, unless you live in Boston. So, do we Well, no. Cambridge might Cambridge might have a chance. Actually, that'll put a lot of pressure on Boston. That'd be awesome if Cambridge beat Boston by trying to take a taxi and it stops at the Harvard Bridge.
1:45:23That's right. The bridges are are no fly, no drive zones for autonomy, apparently in the near future.
1:45:28Actually, what's funny is you can get around to Harvard, but you wouldn't be able to go to Harvard Business School.
1:45:32That's really cool.
1:45:33Yeah. What's going to happen though when you get your Tesla Optimus robot and it has the drive program? So it can drive for you in whatever car you have without retro.
1:45:41It'll get banned in Boston, too. I'm sure in Boston.
1:45:46You know, I I want to hit the economics here because uh you know, the the Whimos are pricing out the new version of the Whimos are pricing out over $100,000. Uh the cyber cab is coming in at or below $30,000. And what's most interesting in my mind is the winner here is going to be whoever can mass manufacture these the fastest. And Elon's going to win that game. I mean, he's the guy that builds the machines that builds the machines.
1:46:13Look, clearly in the US maybe, but China is is dumping some might say into Europe. So, it's it's not a US-on game is is the issue.
1:46:22You're going to end up with all of them for the for the foreseeable future. And I think Peter, the point you made is the most important one. The end user wins.
1:46:33Yeah. Uh Dave, do you remember when we were at uh at the Gigafactory and we're being toured around and we saw outside the giant mounds of aluminum scrap metal? Yeah.
1:46:43And then the smelter and then the Model Y press. And it was like just this beautiful orchestration. It's incredible. End to end and and you can actually walk with the car from the day it's the minute it's born. It's like a it's a long walk, you know? It's almost a mile, but a car comes out the other end. And but you're literally watching every part get put on as you walk with it. It's absolutely wild.
1:47:07Yeah, it's beautiful.
1:47:09But it's amazing to me in the in the Cyber Cab how few parts there are. Like when you open the hood of your gas guzzler car and you look at all the stuff that's in there and then you look inside the equivalent cyber cab, it really feels like there's maybe one tenth as many things.
1:47:23Um I've got a hard use a lot here which is a typical car is 2,000 moving parts in the drivetrain and the Tesla has 17.
1:47:31And I an IC internal combustion engine car.
1:47:362,000 moving parts versus unbelievable.
1:47:40Yeah. So, in terms of maintenance, design, you know, all of that stuff. This is why the car dealers are all freaked out. You never need to I was, you know, last time I was in New York, I was in a yellow cab. I'm like, why is this thing so disgusting? It smells like a public urinal. Why? Why are they all like this? But then you think about what happens to that cab.
1:47:58You know, the medallion is incredibly valuable. It gets handed from one driver to the next. You know, it never stops moving. And then at night, in the middle of the night, uh, somebody drives it home and then they hand they get up first thing in the morning, they start driving it again. But it has to take somebody home. The cyber cab goes to get cleaned in the middle of the night when there's nothing to do. It goes to a place, could be anywhere. It could be in Queens, way, gets itself cleaned, and comes back pristine. And that's it's just night and
1:48:26day when you get into one of these versus a a cab. Now, they're brand new, so maybe that's part of it, too. But the experience is night and day better.
1:48:34And they look beautiful.
1:48:37So what happens when they break down? I guess a a cyber truck comes and towes it away.
1:48:43Door dashers come over to help is is the recent story. I guess the door is wedged open and can't be closed.
1:48:50Maybe they just squash it into a metal cube right there. Crush it and recycle it.
1:48:55Take it back to the smelter.
1:48:59Oh, we'll pull out the GPUs first.
1:49:03All right, I'm going to move us on uh to the space arena. Two fun stories on the space frontier today. In our first story, NASA chose Blue Origin to build the telecommunications relay network on Mars. The comm infrastructure that will connect future Mars missions back to Earth. Uh this is the infrastructure layer for the Mars economy. Whoever owns the telecom network controls the bandwidth on Earth. There's a companies like AT&T on Mars. It could be Blue Origin. Uh, and you might wonder why did
1:49:31NASA select Blue Origin, not SpaceX since SpaceX, you know, owns and operates the world's or space's largest space-based laser linked comm network. And um, ultimately this is the government, you know, keeping two competitors, two suppliers in business, giving them each a slice of the pie. And regardless of Blue Origin having won the NASA contract, I guarantee you Elon will still build Starlink around Mars, uh, ultimately what we're seeing here is the birth of the interplanetary internet.
1:50:03Alex, you excited about this one?
1:50:05Look, Mars had it coming. The moon had it coming. We're going to build the Dyson swarm. Someone was going to get awarded the the Starlink for Mars. Uh, it's interesting that it went to Jeff Bezos's company and not to Elon's, but I I'm fully expecting that we're going to have a very very competitive interplanetary internet and I'm frankly I'm I'm glad that there are vendors that are competing for for Mars coms other than SpaceX.
1:50:33Yeah. But it's going to be a packet switch network throughout the inner solar system. Mars asteroids high latency, but that's all right.
1:50:43Yeah. Until we get faster and light uh comms, who knows? Physics.
1:50:48Time will tell. Are you working on that, Alex?
1:50:53Well, the the the physics that we have right now suggests that faster than light travel is not possible.
1:51:00Don't bum me out here. Okay.
1:51:03Sorry to break the bad news. The physics, the textbook physics right now says that super luminal travel, then we have the wrong physics.
1:51:10Uh we'll find out. Just hope.
1:51:14Let's move us to another fun story in space. Our second story, and it's a big one. NASA's Nancy Grace Roman Space Telescope has launched on a Falcon Heavy carrying a field of view of more than 100 times greater than Hubble and the ability to scan the sky for more than a thousand times faster. The Roman telescope is designed to discover tens of thousands of new worlds and map the distribution of dark matter across the universe. Uh before we discuss it, let's
1:51:43play a video by our amazing NASA administrator Jared Isaacman, friend of the pod. And I love this guy. He's such a good communicator. Let's check it out.
1:51:54Oh, the telescope is very healthy right now. It's making its 1 million mile journey to Lrange Point2. It's going to look for what we think will be up to a 100,000 additional exoplanets and other star systems. It's going to help us understand dark energy, dark matter. And as you saw during the press conference, President Trump called in. This is a really exciting time in America's space program right now.
1:52:12What's the difference between this telescope and the Hubble?
1:52:15It's hundreds of times more powerful. I mean, the field of view is is over a hundred times greater than Hubble. Its scan rate is over a thousand times greater. Uh I mean, this is a this is going to be a household name like Hubble and James. This is America's next great exploration asset.
1:52:27And you say it's going to help you find planets hiding behind planets, planets hiding behind other stars, distant stars. The light can blind it out. This will it has a special JPL coronagraph that helps us find these hidden worlds. I mean tens of thousands of additional worlds we're gonna find.
1:52:40Tens of thousands of worlds. Amazing.
1:52:44Yeah. Jared Jared is incredible, isn't he? Compare that to that interview of Sam saying, "Why is this new model different? Why is Astra different for people?" Compare his answer to what Jared just did to answer this this question. It's like night and day. I mean, he is a great great communicator.
1:52:58Yeah. And interestingly data interestingly this is going to finally start to give us statistics over the number of habitable at least earth recognizable habitable worlds it's being pointed toward the center of our galaxy and we'll just be able to do large sweeps of the sky looking in part it has other missions as well but looking in part for microl lensing events for planets that are crossing in front exoplanets crossing in front of their respective stars causing the uh via
1:53:28their gravity causing light from those stars to be very weakly increased briefly due to these microl lensing events. The downside, you know what's so yeah what's so cool about that is on the last podcast Peter asked us uh you know sim you missed it and amad you weren't here. Uh but you know what's your answer to the fermy paradox like why why are we not seeing other civilizations? But one of the theories is that there are many many civilizations talking to each other near the center of the galaxy where you can get from star to star in, you know,
1:53:56a year or two as opposed to where we are. We're way out in the wings where it's like so far away.
1:54:02We're in the unfashionable outer suburbs of the galaxy. That's right.
1:54:07So, you know, who knows? This could I I I'm actually happy we're here. You know, the galactic center is a really dangerous place to be.
1:54:17You don't like supernova, Peter?
1:54:19I don't like the radiation flux they deliver. No.
1:54:21I'm reminded of the opening scene of Hitchhocker's Guide to the Galaxy where they're bulldozing Earth to make a hyperspace g highway bypass. Yeah. Yeah.
1:54:32I mean, I I love this. Uh, you know, I'm curious to our viewers, and if you can, you know, let us know in the notes. uh would you like us to have a conversation uh on the current UAP disclosure con um you know uh discussions out there if we can bring some of the leaders in UAPs and the whole disclosure scenario. The White House just released their disclosure plan. Uh Alex, I don't know if I'm going to mention that, but I'm
1:55:01curious if folks want us to have a conversation on that topic on the pod.
1:55:06Alex, would you mention uh the recent White House announcement? Yeah, there was some reporting out there from Avi's UAP science advisory council. One of the members of the council mentioned that in a recent briefing that this council, which as I understand it was stood up by the White House, was informed that the White House had prepared uh disclosure plan for informing the general public of the existence of non-human intelligence, which if accurate reporting, it that's
1:55:35pretty interesting.
1:55:36Yeah. Iman, how do where do you come out on the whole UAP side of the equation?
1:55:42I actually have no position on it. I've never thought about it properly.
1:55:46I'd like to see um strong claims require strong evidence.
1:55:55A lot of people would say there is strong evidence. It's just hidden.
1:56:00But we shall see. No, I'm with See. I I would say evidence I really it would be highly desirable for there to be a prepoundonderance of evidence that everyone can go and see and touch and experiment on. I I think that's probably the gold standard of uh in in an ideal outcome if there were uh a White House disclosure event. If if the president goes into what's left of the rose garden and gives a speech and says we're not
1:56:28alone, then ideally part two, paragraph two of the speech would be to hold up or otherwise present some artifacts that uh that would be subject to extreme scientific scrutiny to support the claim.
1:56:41That would be a fun conversation to have. All right, our third our final subject for today. Uh three major health and longevity stories came out this week. Uh the first is that OpenAI expanded GPT health features to connect directly to patient records and healthcare databases. They integrated uh the integration brings Epic electronic health records data into chat GPT for healthcare and Epic as you guys probably
1:57:09know is the largest collection of health records. 325 million patients are inside Epic roughly the entire US population. Clinicians can now pull appointment notes, lab results, medications, and ask questions across patients entire record.
1:57:24And consumers uh can now connect their Apple Health, One Medical, Function Health. So, ChatPT can help you understand your test results, prepare for your doctor appointments, and get personalized data and diet workout advice. So, that's the first story. Uh the second story, uh interestingly enough, we talked about it last week. We mentioned that the FDA had approved a drug called uh Duraxinarissib. It rolls off the tongue onto the floor uh for the
1:57:53first targeted RA inhibitor for metastatic pancreatic adnocarcinoma attacking the RAS family of proteins that drive tumor growth in most patients with disease. This week, NBC News reported that the same drug is showing promise to treat lung cancer. As mentioned last week, the RAS mutation family drives roughly 30% of all human cancers and was long considered undruggable, the subject of decades of failed attempts, but now during the
1:58:22singularity, the end of cancer is now within reach. Uh, Alex. Yeah, I think we're starting to see and and interestingly I I'm not sure that AI was actually essential for this particular drug, but I I think I I think there there's so much progress being made in cancer therapies now, including on the imunotherapy side that we're starting I think to see the emergence of and again caveat caveat caveat universal cancer
1:58:51treatments, universal cancer vaccines in some cases where after years and years of treating cancer answer as thousands of different diseases. We're starting to finally get to the point where we can start to treat to move upchain uh upstream and start to treat if not root causes at at least identify treatments that whether it's proteomic pathways on the one hand or immunologic pathways on the other start to have treatments andor vaccines that can treat multiple classes
1:59:21of cancer. And I think that's cancer's cooked. Yeah. that cancer should have been cooked long ago. Was it Nixon who announced the the war on cancer? Uh it it took a forever to get to this point, more than half a century.
1:59:36And it's an interesting counterfactual experiment, I thought experiment. Is there anything knowing what we know now that we could have done 50 or 100 years ago to radically accelerate the onset of broadspectctrum cancer treatments? What do you think, Peter?
1:59:53I think the data I mean we've known for example about the RAS mutation uh you know basically causing unconstrained growth for some time. It's just you know getting getting the molecules and getting the drugs that can attack it properly. Uh I think the tools we have right now are just finally giving us that that reach. Uh and then being able, you know, we talked about cell simulators, being able to understand fundamentally what happens and how to
2:00:21block it is what's coming. Immod this is an area of personal passion for you as well. You've been deep into medical AI.
2:00:29Yeah. So I said the this wasn't AI kind of on the thing, but the range of treatments now coming out on the cancer side gives a lot of hope to what's going to come. I think the mRNA one maybe is more general um that we saw recently.
2:00:42And I think you know the first part of that integrating into the epic health records actually applying AI across the board like we should have a sprint so that within a year or two max every single health decision is double checked by an AI. We should have more more than that I think it's going to become malpractice to diagnose a patient without AI in the loop right we already know AI is a far better physician a diagnostician than a human is so I think you should have a series of
2:01:10approved edge and cloud models and every time you make a diagnosis the AI has to have had one check that will save so many lives it will detect so many cancers and then it's about how do we increase the level and volume of information because Even now the type of data we have around cancer and other conditions that we absorb is tiny compared to the amount that we could have with the AI transforming it. So I think that yeah let's cook disease let's get rid of it. No one should have to die of cancer and it's something we should
2:01:40be really directed at. And on the first story go on sorry please. It's just like it's very strange that you know as we have the genesis programs and others there isn't just a straightforward we now have the capability to potentially cure this stuff. Let's direct $10 billion towards it. It's tractable.
2:01:59I think the the story on Epic is interesting right having been involved in that business through Fountain Life. you know, Epic is the majority uh you know, electronic health record in the US and it's been a bear to navigate for for physicians and and patients have never had access. So, putting an AI layer on top of that is awesome. Yeah. Um let me move it to our last story here in uh in the area of health.
2:02:30uh it's related to what may be called the first longevity class of drugs the GLP1s. So a new a new paper published just two days ago on September 2nd in nature shows that semiglutide the active ingredient in ompic and wagi recapitulates many of the benefits of caloric restriction and get this it's in female mice just to be clear and extends the lifespan of mice by almost 100 days in humans that's the equivalent of 8 to
2:02:5810 years. Uh the study found that GLP-1R activation initiated late in life in these mice uh accentuates age associated decline and uh modulates conservative genetic regions for aging functioning as a caloric caloric restriction mimedic.
2:03:16Also this week it was reported that Gilip 1 drugs also are being linked to fewer serious infections including tuberculosis and you guys already know that tuberculosis is the deadliest killer on earth. You know killing over 1.25 25 million people per year. So the same drug that's treating diabetes, obesity, dise, you know, kidney disease, cardiovascular disease, and addiction may also be extending lifespan and reducing infectious disease. One molecule, six diseases and still
2:03:45counting. Um I, you know, again, Alex, you made the point that this might be the beginning of longevity escape velocity. I think th this is like to the extent that with the benefit of uh hindsight we look back in a few years and we say you idiots of course you were on the verge you you were seeing the sparks of longevity escape velocity you had the the GLP1s it to the extent that that's the case I don't think it should
2:04:13be that surprising what's more mystifying to me is from an evolutionary perspective if the GLP1 RA class of molecules is capable of doing everything from treating infections to extending life expectancy to modulating diabetes, reducing addiction and compulsive uh behaviors. If it does like all of this, why on earth wasn't like why did not why
2:04:41did we not evolve with uh with either this ability to to modulate our own semiglutide class molecules in our system or maybe a slightly more cynical angle. I if if it turns out that the reason why GLP1s are so effective at so many diseases is that these diseases somehow are diseases of quote unquote modern lifestyles and that we just like it's treating all of the diseases of
2:05:09modernity and that's why we never evolved the solution. Addiction to compulsive gambling or alcohol addiction uh or overeating of sugar that these are all relatively modern diseases. That's why this it's basically a treatment for for modernity. That would be pretty ironic.
2:05:28So, I just want to make a point to the listening audience. First of all, talk to your physician about this. This is not medical advice. Yes.
2:05:34But I use I use uh a Jill P1 drug. I don't know if any of you do right now.
2:05:39And I use it not for weight loss, right?
2:05:41I've been on my fighting weight for a while. My percent body fat is substantially, you know, because I work out and I'm very careful about what I eat. I use it as a longevity drug because of all the benefits we just heard about.
2:05:56Um, got two I know you're using one.
2:05:58Yeah, I'm using one and I just got my blood test done and my liver enzymes are 50% better. Uh, which is pretty amazing. Uh, which means I can drink more. No, just kidding. That's not the point.
2:06:11I know that's not the point.
2:06:13You're supposed to not want to drink it.
2:06:15I know. Um the so there's there's there's that but I think you know to Alex to your point earlier right remember that evolution has birthed us for death right we we've had short life cycles so that you are the cycle time of evolution can work more quickly we're breaking through that now uh and living through that so I think it may have been engineered or evolutionary engineered that we died at different levels and this is a long history where we used to
2:06:44die of heart disease or we used to die of bacterial illnesses and then we figured that out then we died of heart disease. We got some sense of that now we're dying of cancer and then as we break then explain then See explain tuberculosis if if you're in a tuberculosis rich environment surely evolution would favor any molecule that could just be amplified up that would help prevent uh young and reproductive entities from surviving tuberculosis infection. Well, you I go back to the
2:07:14you know there's lots of pathogens going after us all the time, right? Billions of them because they're all trying to survive in their own way, including cancer. Uh the uh capability of the human body to navigate this has never been needed until we started pushing the boundaries to this level. Now we do and and Selene my thesis you know and I've spoken about this pretty widely is the reason for the life cycle is not
2:07:42more rapid evolution. It's that for most of human existence you know homo sapiens came on the scene roughly 200,000 years ago food was very scarce and if our primary mission is to perpetuate our species the last thing you want to do is steal food from your grandchildren's mouths. So the best thing you could do is die. Reproduce and die basically, right? And so we see, you know, the human body, and people should know this, you're in uh you're in sort of prime
2:08:11condition until your late 20s. Uh back 200,000 years ago, you'd go into puberty at age 13, 12 or 13. You'd be pregnant immediately, no birth control. By the time you were 26, 27, 28, you're a grandparent, and then you would die so you didn't steal food from your grandchildren's mouths. And and so We never had any that's my joke about marriage that we invented marriage to keep the parents together the till the kids were self-sufficient average lifespan was 25
2:08:38years old from the most of human history we invented marriage about 6,000 years ago and it's definitely true then marriage is not designed for uh 50 60 year lifespans it's designed literally for no no watching this one of my relatives calls it state sanctioned torture um marriage because the we have the job of now evolving the institution to deal with the conditions today compared to when we first invented it. And I use that example because it applies to all our institutions, democracies,
2:09:08educational systems, legal systems, healthare systems. This I think is the biggest work we have to do is as we blow past all these limitations, we have to reinvent all of the institutions which are the scaffolding that keep humanity safe and civilized.
2:09:22Hopefully, we have a benevolent AI to help us do all that.
2:09:25We'll need it. And if not, Benevolent AI. At least we'll have GLP ones.
2:09:33All right, time for some conversations with the mates. Uh, we've got some questions here. Uh, Emmod, I'm going to give you first crack.
2:09:43All right. Uh, if money becomes obsolete, how will desirable land be allocated? Is everyone just locked into their beach house forever? By nick 52547.
2:09:55This is an interesting one. I think that you know Elon and others have said that money might become obsolete. I've said that you can't compete with robots. I think this is a question of kind of land rights and more. Um because typically where you see reallocation is in upheavalss. So the way that we're going without the right structures, you'll probably have a debt jubilee. You'll probably have chaos and land redistribution.
2:10:17But if we can actually navigate through it, then land rights and property rights should be enforced and so you can keep your beach house. But the number of desirable locations will go up dramatically because you will have air taxis, self-driving cars, solar panels, and self-driving construction workers. So yeah, exactly. Self-driving construction workers.
2:10:41Nice. See, uh let me take number four. Uh do you think AI will evolve past this risk averse bottleneck that we're currently in? And that's from ads susie1073.
2:10:55Um uh yes that's the hopefully but hopefully we move towards calibrated risk rather than recklessness because right now we we treat uncertainty as kind of like a reason to refuse right and that's a technical problem but there's all the legal and policy issues around this. Uh when you're raising kids one of the things they teach you teach kids is are you taking a responsible risk? Um and we need to apply that same
2:11:22kind of paradigm to this model. Could you create something like a risk budget which is what can an AI decide? How much can it uh spend on that? What systems can it access? What triggers escalation?
2:11:34Right? And we need to get very sophisticated around this. Um models need to become better at distinguishing uh uh dangerous intent from uh real expert usage, right? because you can't get to you can't just say remove the guard rails. It's got to be dynamic uh where you deal with proportionality on user identity, the context, uh the
2:12:01reversibility of that. So we'll get AI maturity when we can say here's the risk, here's the confidence I have in this and here's the reversible next step. Uh, and then rather than just saying no, I think that we need to get a lot more nuance around this. And unfortunately, in today's world, nuance has no part to play in the soundbite politics that we have out there.
2:12:23All right, Dave, over to you.
2:12:26All right, I'll take number one. Uh, can you see a way to turn libraries into centers of AI development and physical AI training for everyone from Train with John Koalo?
2:12:38Um, yeah. Yes, but there's a bigger issue which is there's a huge amount of white collar office space and white collar work is turning to AI. At the same time, we have declining population espec especially working age population. So there's a bigger issue of what about all this other space? What are we going to do with it all? It's very similar to what happened with the shopping malls after online shopping became huge and then COVID hit and a lot of people were thinking there's got to be something really great we can use with all this
2:13:06shopping mall space. But there were some ideas but for the most part didn't work out that it all got replaced. So I think um there are lots of ideas for what you can do with the library space. You don't really need the library books anymore obviously. Uh this is another reason why you should want data centers in your community. It's one of the few things that will reliably grow tax revenue create job opportunities in a community.
2:13:33So there are ideas, but I'm not super optimistic that all the space will be well utilized going forward.
2:13:37And there are programs in inner cities right now that are, you know, running AI tutoring in libraries. Uh that does that does exist.
2:13:48Uh Alex diamond chips.
2:13:51I get the fun one. So question number two, whatever happened to synthetic diamond chips for computers? And this is from Asterene. So here's the problem with diamond. Pure diamond is an insulator. It has a band gap of approximately 5 1/2 electron volts. So it really doesn't want to be a good computer. To make it into a good computer, at least a good computer of a recognizable cosy type, you have to dope it. That's hard. If I had to steal the case for diamondbased recognizable computers, it would be for
2:14:21ultraigh temperature environments where such a wide band gap could be advantageous or maybe very high voltage environments where again a large band gap could be advantageous. It's not an enormous market. Could be wrong, maybe I'll be wrong. Doesn't seem like an enormous market. Where I'm much more bullish on diamonds is for sensing. So you you can put what are called nitrogen vacancies uh or NV centers into diamonds. Basically put an extra
2:14:49unwanted nitrogen atom into a diamond lattice and suddenly you get an exquisitly sensitive magnetic field or just field in general sensor because of an an extra vacancy that is introduced as a result of the diamond. So, diamonds for quantum sensing, super interesting.
2:15:09Potentially, it's not investment advice, but I I think technologically it's very attractive. One could imagine at some point in the future, diamond chips for sensing even get us to sci-fi tech like wearable MRI sensors. Diamond chips for computing? Eh, probably not.
2:15:24Why did I know you'd have an answer for that one?
2:15:29All right, See, first choice is yours, pal. Um um uh I will go with number seven. Uh when do we get a forecast for when data center energy sources transition from gas turbine farms to nuclear and SMRs?
2:15:48Uh short fairly easy answer here. You know gas dominates the current buildout.
2:15:53Uh SMRs going to be in the next 3 to four years. The reason people are getting so excited is that nuclear is an engineering problem, not an invention problem, right? Fusion is still in the invention problem category. Um, and the so you've got this impedance mismatch of data centers taking two or three years to build. Nuclear is taking a bit longer. Um, we will get there I think faster than people think with SMRs, but I think it would still be a while. So,
2:16:22so take it uh two three years for the initial wave of of SMRs and then five to seven years for the big buildout. So, I think what's going to end up happening is AI is going to do for nuclear what what smartphones did for batteries like it just created such huge demand that the massive innovation there was a huge acceleration in the innovation curve.
2:16:45Nice. And we had a fantastic pod with Rome Nom on energy. If you guys haven't seen it, please check it out. You know, his Yeah, he talks about pretty much that time frame. Uh that's where I get all my good information about energy.
2:17:00Yeah. Alex, let's go to you.
2:17:02All right. Well, I I think I have to answer question number eight seems to be directed toward me. So, eight asks, "How far away from Alpha Centtory do you actually have to aim to reach it when it arrives?" And this is from typical dad pi. So to to three significant figures.
2:17:19So I to the extent I understand this question uh let me give you the semiofficial firmy explorer line. So on the last pod we had Matt Pines and Philip Johnston announcing for for those who who didn't watch humanity's first mission to Alpha Centuri and and and I'm involved. I've I'm involved.
2:17:39What can I say? Uh I'm I'm in a lot of rooms. Uh so the under the official mission parameters for the the Fermy explorer mission the goal is to get at least 99% of the way to Alpha Centuri.
2:17:54So uh Alpha Centtory approximately four light years away. So you could do the math and and that turns out to be 4100ths uh of uh of a lightyear call it uh precision getting there. And that's from Earth. So if if I put my sci-fi hat on and extrapolate a bit, I think that in the next few years on board and and maybe I should add parenthetically, Firmeny Explore mission is intended to
2:18:24I suspect this this is not an official position for Fermy Explorer mission. I suspect that when the mission comes to full fruition and is launched, I suspect it will have active guidance on board.
2:18:36Uh and if that's the case, this 99% of the way to Alpha Centtory, not an official position, will turn into or can turn into 100%. Uh in in the sense that it actually hits the Alpha Centuri system, not just 4/10en of uh or 4100s rather of a lightyear away. That said, Fermy Explorer is intended to reach Alpha Centtory 80,000 years from now, which is perhaps inconvenient from the perspective of mission verification. if
2:19:05you actually want to get it. But I I actually GLP1 drugs.
2:19:09We're going to need a lot of GLP1s for this. I I think I think we'll get to the whole point of of the mission is that we're supposed to beat the Fermy Explorer and get there sooner. This is just the first one to launch.
2:19:19Amazing. Dave, over to you, pal.
2:19:22All right, I'll take number six. Are data centers using closed loop geothermal cooling also noisy or are they quieter? And that is from Asterene.
2:19:31uh they should be dead quiet just like geothermal heating is dead quiet. Also uh nuclear reactors that are ocean you know near the ocean use geothermal cooling and ocean cooling and they're dead quiet. So it should be very very quiet. But also you know regular data centers that are using liquid cooling are only noisy because the water gets cooled outside with these really poorly designed fans. There's no reason for that. If your if your regulatory body says, "Yeah, you can build a data center
2:20:01here, but it has to be quiet." Guaranteed they will build the data center quiet, there's no reason those external fans need to make any noise.
2:20:08Instead of making them illegal, cities need to be saying, "These are our requirements. Drop our energy costs, you know, make them quiet, invest in our infrastructure." And they will, and they will.
2:20:21Uh, Immod looks like number five is for you.
2:20:24That's an interesting one. They talk about a sun shade there. What scale of satellites in terms of we talked about that came up on the last pod. You missed it. Yeah.
2:20:34Yeah. And what scale of satellite terms square meters required to actually impact global temperatures?
2:20:40Uh I would say if you whack something in the sun earth lrange point a couple of million km out you need about a couple of million km squared. So about the size of India would knock a a degrees Celsius off. That's bigger than anything we've ever managed. But, you know, it's worth a try. Why not?
2:20:59Yeah. And maybe convert the moon.
2:21:05Or or a series. Yeah. And I I still love the the sun shade, you know, putting a thermometer be able to titrate the solar flux on the planet.
2:21:14I think it's a trick, for what it's worth, I think that's a trick question. I I think with a sufficiently good AI planetary scale model, we could make uh any sun shade or or other satellite intervention dimminimous size. It's just a matter of appropriately perturbing the the Earth atmospheric system with enough AI.
2:21:33All right. Uh as before, we love your outro videos. Please send it to us. We want to see your creative nature. You send it to media dmandis.com. The team will uh show it to us and then we'll air it. And again, last shot. If you want to join us on our AMA, we have two AMAs coming up on Zoom with all the mates to answer your questions. We're going to do one in the morning to help hit Europe and Asia, one in the afternoon, early evening to get everybody in the United
2:22:02States. You'll have a chance to ask us all your questions. Go to moonshots.com and register. It'll be up there just for a couple of days and then we're done. All right. Uh Immad, I think this is the first music video that includes you. Uh which I appreciate. Let's There have been others. I remember others.
2:22:22Okay. Well, the first one I remember.
2:22:24All right, this one is Supersonic Tsunami. Uh, and thank you so much to uh Chromatoss for your for your submission. Can you smell that in the kitchen? Some uncooked academic invention overshadowed by gatekeeper hysteria. Answers to major problem areas. And that's all it takes. A moonshot and willing mistakes. If breaking the moon makes you mad, remember it's just a launch pad.
2:23:10Abundance we will foster while we live long and prosper. Riding the wave of autonomy formed from supersonic tsunami. The takes are considered sometimes hot but investment advice. This is not the analysis. Brilliant never tring pointed balls and strikes. Someday we'll disassemble the moon. Deal with every elephant in the room. Destroy all financial disparities.
2:23:58Don't sleep through the singularity. Abundance we will foster while we live long and prosper riding the wave of autonomy formed from supersonic tsunami.
2:24:16All right to the beach.
2:24:18That was awesome. Enjoy the umbrella drink.
2:24:22All right for an amazing pod.
2:24:24Hey, make sure you're uploading.
2:24:25It'll never get slower. See you next week.
