AI Roundtable: What Everyone Missed About Gemini 3 w/ Salim Ismail, Dave Blundin & Alexander Wissner-Gross | EP #209

27:19is not so much the capabilities are being added as is restrictions being removed. And Frontier models, in particular, when we see live audio type engagement, are moving from what they've been in the recent past, which is audio to text to text to audio, just directly audio to audio, which enables much, much richer audio interactions, including accents. Yeah, I mean, and where we're going here with the next generation of AR glasses, as everyone's developing and basically plugging into your auditory and visual input. It's simultaneous translation. It is going to change how we communicate with people around the world in an extraordinary fashion.

28:02This was a fun one. Again, continuing on the Google theme, the TLDR, they really have gone and won hands down. I know, Dave, you and I are looking at the prediction markets that Google has literally skyrocketed to be the contender that's going to be is the winner by the end of the year. And I think they got that mantle. Google AI helps users shop, compare, and call stores for the holidays. So new agentic features can call your nearby stores, check stocks, pricing. Gemini apps add built-in shopping tools. I mean, this is like, hey, call 20 stores within 10 miles of me and find out who's got the cheapest prices and put it on hold or better yet, purchase it for me and have it delivered tomorrow.

28:48Holy cow. A lot to unpack there. So I had to check on this one, Peter. It was all of seven years ago that Google launched Duplex, their AI store calling functionality at I.O. Seven years ago, 2018, the year after attention is all you need. It's been seven years for this to make it into some fully realized format. But I think this is finally the beginning of AI starting to autonomously index the physical world. If you can have AI call stores autonomously, you can send AI-powered robots out into the physical world to index everything that's going on as well. I'm curious what the consumer behavior is going to be like, right?

29:28Is it going to be just become – actually, what I'm really interested in is what's it like on the other end when you're in the store and you're getting all of these calls inbound? And at what point is more than 50 % are AI calls? You have AI answer the AI calls, obviously. I mean, is it going to be that you have to identify yourself as an AI? Probably. That is what Duplex has historically done. It announces itself as an AI assistant. Yeah, actually, so far, it's going to be state by state. But so far, the AIs are not announcing themselves. And we do a lot of this inside our lab here. So about half the time, people are like, am I talking to an AI?

30:06And the other half, they have no idea. And so do you have to answer it if it asks? You don't have to in most states, you don't have to. But again, regulatory consideration is moving so slowly. It's just completely ambiguous. But as of right now, you don't have to. But it doesn't hurt to say, yeah, I'm an AI. or even declare it up front, it's not hurting the call performance rates at all. So you might as well just say, hey, I'm an AI, but I'm so much more helpful than the guy you were going to talk to. My new business idea then is a little button on your phone. When an AI calls you, you flip it over to your AI.

30:39Because when I'm calling a store, I want to speak to a human. But the human at the store, what do you think about that product? How good is it? Are people returning it? And that interaction is a pro-human to human interaction, but I'm not going to have that tolerance with an AI. Wait, I want to challenge you. If you call a store, why do you want to talk to a human? An AI is going to know way more about the inventory, the situation than the human would. Yeah, exactly right, Salim. And not just that, the AI can pull up images in real time and show you the product and spin it around and stuff. So it's nothing like talking to a human in a store.

31:16It's actually far, far more engaging. I'll tell you what else. The voice run guys here in the lab are doing OpenTable, doing restaurant bookings and stuff. And you wouldn't believe the fraction of restaurant bookings that are non-English speaking person or going the other way if you're traveling internationally. It's a lifesaver to be able to talk in a different language and do your full booking and then the AI just translates it. Fascinating. I do think this is how we get to APIs for everything. There's now the need for an escape valve for surfaces, for business interactions that don't support APIs.

31:55With an AI that can make voice calls and have arbitrary unstructured interaction, we get APIs for everything. Yeah, we do. Okay. One more article on the Gemini front, Gemini 3 benchmarks. We should probably skip this. I don't think anybody's interested in it, but okay. Oh, you're – you can't. That's a good one for you. Good one for you, right? Good one. Alex has just sent a drone to your house there. Watch your roof. To take me out. I'm going to send my duplex AI to give you a phone call. All right, Alex. Clue us in here. Gemini 3 benchmarks, how good are they? And at the end of the day, what do they really mean?

32:32I mean, just to represent people watching this, listening and watching our Moonshots program. Okay, Alex, I hear you talking about benchmarks every time, right? We're going to talk about some more benchmarks in a little bit. But what does it really mean? What does it mean to me? So, please. Sure. So I guess there's the headline, the numbers are going up and to the right, so who cares? Who cares is we have a way of measuring progress in our civilization. And this is a precious moment when, with raw numbers day by day at this point, we can track progress towards solving some of the hardest problems that our civilization faces.

33:11Humanity's last exam, say what you like about it. Some like it, some like it less. But it's an attempt, as are all of these benchmarks, to encapsulate in a measurable quantitative way progress by AI towards solving hard problems. In Humanity's Last Exams case, it's an attempt to measure the ability for AI to solve PhD-level problems. In the case of Arc AGI 2, it's an attempt to model human-level ability to visually reason. The so what is these benchmarks are all saturating, which means that AI is, at this point, has the ability to perform PhD-level research. When we think about the so what for the so-called average person, it's going to be that AI is imminently, I think, well-positioned, now that these benchmarks are saturating, to start solving the hardest problems on Earth in math, science, engineering, medicine.

34:04That's the so what. We spoke about that last episode, last pod with Sam Altman speaking about science breakthroughs coming on GPT-6. That's his expectation. And here, the numbers are impressive, right? If we're looking at GPT-5.1, Gemini 3 is basically doubling the ARC-AGI 2 benchmark. It is effectively doubling CLAWD 4.5 on humanity's last exam. I mean, these are not incremental moves. They're significant step-ups. And critically, it's not bench-maxing that we're seeing. There are some labs that have been accused of just optimizing their AIs to do well at one or two of the benchmarks. And then when you ask them something out of distribution, they fall over.

34:50That doesn't appear to be the case here. It feels like the team behind Gemini 3 really did a professional job not over-optimizing towards narrow, spiky intelligence on any of these benchmarks to do well in a press release. This feels like a well-rounded, generalist AI model. And given the trajectory towards saturating these benchmarks, I'd be very surprised if by the end of, say, next year, we're not seeing hard research problems succumb to AI models like this one. Do you remember two podcasts ago, Alex, we had that paper that came out on how to measure AGI, like defining it in terms of, I don't know, 10 or 12 different quadrants.

35:29I wonder how Gemini 3 does on that. I'm sure we'll know soon enough, but I would expect it to do generically well on the spikes where models historically were doing well. As I recall, one of the spikes where one of those dimensions where models historically did poorly was on continuous learning with ultra-large context. Off the cuff, I wouldn't expect Gemini 3 Pro to do amazingly better on ultra-long context, but it does really well on retrieval scores. I don't think it's shown in this slide, but there are other needle-in-a-haystack-type benchmarks that attempt to measure how well models are able to retrieve tiny facts of information buried in their context window.

36:09Gemini 3 Pro does amazingly well at retrieval as well. So I think almost everything is going up and to the right at this point. Yeah, amazing. There was one observation I had, and I wanted to check with you guys what you think of this. When you have coherence at this scale, it implies we have systems-level thinking inside these models. Is that accurate?

36:32Could you say a little bit more, Salim, about what that means? Well, because you've got essentially, you know, systemic thinking is one of the holy grails of deep, deep reasoning, right? Because you can look at the entire patterns of things shifting. And it looks like, feels to me like we're at this level of AI competency, you can get to that kind of systems level thinking. That means you can do world modeling in a really powerful way using almost, you shift the whole thing into symbolic reasoning almost when you can think in those concepts. So don't we get to that level very quickly now? I have so many thoughts, but the first thought that immediately jumps out at me is, of course, these are world models.

37:09And of course, they're able to symbolically reason. They're solving math problems and they're writing source code. I would argue that in past, you've seen some commentators argue that there's some sort of nebulous, neurosymbolic type advancement that's waiting to drop. I think that's utter nonsense. Of course, they're able to reason symbolically. The tokens are in some discrete space. And, of course, there are systems-level thinkers that are able to solve PhD-level problems across dozens of disciplines. That requires understanding the world as a system. So, yes. Yeah, I agree with that. And it turns into a philosophical debate and nothing great usually comes out of it.

37:47But I will say that this is a 7 trillion parameter class model. And last year, all the naysayers were saying, well, there's evidence that things will slow down. Because last year we were at a trillion parameters. And they were clearly wrong. You know, when you went from one to seven, we know next year is at least a 10x and up to a 40x step up in raw horsepower. And the naysayers are saying, well, things are going to level off unless we crack through some other level of system two level thinking. But they're clearly not leveling off. And I would challenge the technical audience out there looking at these benchmarks to you.

38:22You're almost obligated to think about two things if you're in all inclined. One of them is where on these benchmarks does it become self-improving? Read all of Ray Kurzweil and really have an opinion on that because that's tied heavily to benchmarks 1, 4, and 6 on this slide. And to just have an opinion, I have my opinion, but have an opinion about where you need to be on 1, 4, and 6 in order for this thing to improve its own algorithm. That's a critical point. And then the other one is where do you need to be on the benchmarks to start proposing cures to diseases? and being right. And if you work in anywhere in health tech and you have no opinion on that topic, you're doing a disservice that's bordering on, in my opinion, bordering on negligent homicide, because this can save lives if you work on it, if you apply it to whatever you're doing in health tech.

39:14And you're obligated to get your head out of the sand, look at this podcast, study the numbers, and at least have an opinion. And even if that opinion is, no, it's not going to work, That's fine. I'm okay with that. But to say I don't know or I didn't listen to the pod, that is absolute negligence. Can I ask a question to you, Dave and Alex? You know, Jan LeCun comes out saying we have gone down the LLM rabbit hole and that's the wrong direction. We're optimizing on that. We need to go through a different evolutionary tree to really get to AGI. What are your thoughts? All the old people say that.

39:54And all the young people don't. When that tells you something out of the gate, you're sorting yourself into an age bucket just by saying it. And there's definitely a philosophical divide in there. But the question I would ask isn't is there another innovation that we need? It's whether a human will have that innovation or this exact AI scale will have that innovation. I would bet on the AI anytime. Either way, we have so much to absorb just from where we are now. Forget everything else that may come along later. I think there are also many paths to AGI, and I know and respect Jan's work, and I know he favors an approach toward AGI that's more focused on actions in an embedded space rather than in terms of autoregressive models.

40:39That may be a perfectly legitimate approach as well. But when I see the scaling laws continue to hold and capabilities continue to go up and to the right without any new paradigms, it makes me think maybe really we can just continue scaling and don't need to worry as much about yet another paradigm shift. And let AI do that. All right, let's go on here. Let's turn to... Insert my normal rant about AGI here and we can move on. Okay. So noted and approved.

41:36pre-compiles code for each task, Blitzy delivers 80 % or more of the development work 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 pre-IDE development tool, pairing it with their coding copilot 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 building with Blitzy today. All right, next story. OpenAI introduces GPT 5.1 for developers. So again, this is a benchmark question.

42:23First of all, this was announced before Gemini 3 came out. So I'm curious, AWG, whether this is still the case and why, again, why does this matter? Yeah, I think the economics of this, the microeconomics are maybe even more interesting than the technical side. So we're starting to see, and this is somewhat visualized in the chart you're showing, the beginning of inference time compute start to conform to the economic productivity of queries. So you know how, like in Google search, for example, if you search for mesothelioma litigation, you're going to see a bunch of very expensive AdWords ads.

43:03Yes, for sure. It's a very economically valuable query. On the other hand, if you search for— For the lawyers. For the lawyers. For the lawyers. If you search for, like, an arithmetic query, you'll see none or almost no ads because it's not that economically valuable. We're starting to see, I think, that same dynamic emerge here, where certain queries require lots of inference time compute. And so what we're seeing at the routing layer with GPT 5.1 is even more compute being allocated to queries, to prompts that really require a lot of compute. And then for the lighter, easier queries or prompts, we're seeing less compute get allocated.

43:43And I think it's actually pretty profound. It's not just a matter of moving around the deck chairs in some sort of zero-sum game. I think this is actually almost a premonition for what the economics of post-superintelligence will look like. One of the things I think the most about is who's going to pay at the end of the day for the trillions of dollars of CapEx in data center build-out? Who's going to pay for it? Is it going to be the consumer? Will the consumers on average be spending hundreds of dollars per month on core subscriptions for AI? Or will it be enterprises that are spending billions of dollars, in some cases, for enterprise-level tasks?

44:23And I think what we're starting to see here is that modally, probably it's going to be the enterprises paying lots of money for the most valuable tasks. In the same way, we're seeing right now in Microcosm, some of these harder tasks, harder prompts get allocated a lot more inference time compute at the expense of easier queries. I would totally bet on that direction just because if you're, say, Target, you can manage merchandising and get 20 % extra margin on something, then it's worth the extra compute on the back end. And we'll see a lot of that. But there are places where consumers will spend hundreds of dollars a month on their iPhone, on their plan, because it enables them in an extraordinary fashion.

45:03But remember that the money to be made here is on the margin from persuading people to switch their behavior from what they otherwise would have done. If they were going to spend the money anyway, that money doesn't go to the AI. It goes to the entire value chain underneath the phone manufacturer. All right. Well, I can tell you, in my experience, you have to operate at the margin, at the extreme end of what these are capable of. And I've tried to either save money or to get more speed by dumbing it down by a half step. And it just isn't the same. And so it just feels like everybody wants to be at the forefront.

45:39And this is the weirdest product that's ever been launched on humanity in that it's talking to you as it's selling to you. And so you start with a subscription. They give you this incredible experience. And then it tells you, well, you want more of that. You need to upgrade. But it's actually telling you. It's talking to you about upgrading. No product, no cable company, no iPhone has ever done that before. So it's a salesman baked into its own capabilities. It's kind of creepy, actually. It's very weird. All right. Let's stay on the OpenAI theme. And this is a fascinating story. It's an important one.

46:20OpenAI-backed startup aiming to block AI-enabled bioweapons. So this is a startup called Red Queen Bio. And they received a$15 million investment from OpenAI, which, by the way, just sounds really small compared to all the$100 billion and trillion investments being made. But Red Queen is using advanced AI plus lab testing to spot vulnerabilities in biological systems. They're basically saying, hey, we want to stop people from using these AI models to create bioweapons. Super important. Who wants to jump in first? I'd love to speak to this one, maybe for starting with the literary reference. So for those not tracking, Red Queen, in this case, is a reference to a scene in Through the Looking Glass, where Alice and the Queen are constantly running just to stay in the same place.

47:11So the Red Queen's race in general is used as a metaphor to cases where a lot of effort is required basically to maintain a standstill. And in this case, I think the other key concept that I think is ultimately quite profound out of what Red Queen Bio has announced and the reason why they're taking funding is we've just spent quite a bit of time talking about how as you pour more compute onto these models, the capabilities keep increasing. Inevitably, you have to worry about alignment and safety as well. In society, if you're growing a city and you double the population, you're going to approximately want to double the police force or the safety force.

47:53Wouldn't it be wonderful if, as the capabilities of AI keep scaling, keep increasing, the safety measures, the alignment and other properties that make them safe for humanity, if those also benefit from scaling with more compute? So seeing scaling laws, Red Queen Bios announced that they've uncovered scaling laws for biological safety measures. I think this is the way we achieve alignment. Just like, again, the scaling law for police forces in the city, a little bit sublinear relative to population. Same idea here, but nonetheless close. As capabilities increase, we want to live in a world where we achieve so-called defensive co-scaling, where the resources and capabilities of safety measures scale close to proportionally with the resources and capabilities of the underlying models.

48:45Yeah, let me add some data to that. So today, or at least last year in 2024, the biosecurity, biodefense market was$34 billion. And it's expected to double by a decade from now, 2034, 2035. But here's the quote that really hits me, right? So an extreme bio-attack scenario could have a multi-trillion dollar global loss. And the notion is, could you create such a bioweapon for a thousand bucks, right? It's the asymmetric situation where a small amount of money using complex models could do a lot of damage. And so there's got to be this layer of defense. I mean, it's critical. When I talked to Eric Schmidt, I remember a couple of years ago at FII, the number one scenario that is of greatest concern are bioweapons, something that can be – you take an existing virus, you change its viral payload, you make it much more infectious and release it.

49:47You know, Salim, you and I have had this conversation that one of the most important things is going to be to set up these biosensing capabilities at train stations, airports, bus stations that are filtering the air and looking and doing rapid sequencing of everything they come across. You know, the majority of the bioweapons that are concerning are airborne, right? So a person coughs or sneezes and it's there. And one thing that is in our favor is that these viruses, these bioweapons can only move at the speed of an airplane. That's the fastest it can go, right? And it travels. We saw that with the release of COVID.

50:31So if you can detect it at an airport, sequence it on the spot, develop an antiviral, and then transmite that at the speed of light, not the speed of 600 nautical air miles per hour, then you have a chance of battling it. And this is what we're talking about.

51:20as a sidekick, suddenly they're empowered to build virtually anything in that basement, and that's the risk. No U.S. company wants to be responsible for that. So they're trying to cut it off at the query level, saying, wait, as soon as you ask the AI to help you, we'll create a bioweapon, it stops. And so the open source would be a huge leak in that. So the U.S. labs don't do the open source anymore. The Chinese still do. So I don't know how to comment on that. How do you deal with that if it's a model running on my laptop and somehow it contains enough knowledge to do this and I can query my laptop, no one ever knows the query I've made, it's just resident there?

51:58How do we deal with that? Yeah, I think ultimately it all reduces to co-scaling. So if you imagine having a fully self-contained facility, hypothetically, in your basement, and the ultimate societal protection will be having lots of sensors and, more importantly, having lots of AI screening, super intelligent AI screening that can spot hidden agents. I have this dictum that I think is super important on so many different levels. In the software engineering world, there is Linus Torvalds, who created Linux, has this so-called Torvalds law that, and I'm going to butcher this slightly, that with enough eyeballs, all bugs become shallow.

52:45And I would propose sort of a generalization to that, that with enough superintelligence, all hidden agents become shallow. To the extent that we have hidden agents in their basement building super weapons, I would expect with enough super intelligence, defensively co-scaled, they become shallow. So I've made the comment before that privacy is an illusion. And this is just going to shatter even that illusion. Because if you want safety, you're going to want agents listening and watching everything all the time. Salim? this is an arms race i think what we've seen throughout history when we tried to we thought oh my god email's going to crash because of of all the scams and then we thought we have phishing we can't solve for that and we've used uh ai consistently in that sense because people forget the bad actors may use ai and they will but the good actors can also use ai and therefore you just have to be one step ahead the question is if that gap gets too big one of the challenges with what you were saying earlier, Peter, is you may not know what to look for in some of these and that's the danger point.

53:55Yeah, well, you do have a catalog. It's an interesting little case study too because, you know, if you rewind the clock before Gmail took over, you know, Microsoft had Outlook and Hotmail and Google launched Gmail and the two promises were very different. Microsoft said, we will never read your email. And Google said, we will read every word of every email that you receive, but it's going to be read by an AI and not by a human. So we won't let the human eyes look at your email. But we're going to do all kinds of things based on the information in your email read by the AI. And people didn't care.

54:26And so everybody moved to Gmail. So you have an interesting case study in how this plays out, just the human behavior. So here I think the equivalent is, hey, I'm talking to AI about my most personal things in the world. And Peter, I think you're right. The AI is going to listen to every single word. And if you're designing a bioterror weapon or a cyber attack, it's going to flag it. and escalate it. And if you're talking about your virtual girlfriend or whatever, that's going to be fine. I'm just going to kind of hide that. Yeah. I remember talking to the head of one of the major intelligence agencies, and they had a very clever thing.

55:01They said, look, when there's known things like nuclear weapons or whatever, we put eyes on it, we try and watch it. When you have something like this that could be developed in secret, they've been actively opening up these communities and actually funding the biohacking movements because then you can see things earlier. But this, if you can do open source bioweapon development in a lab, in a bunker, that really causes a huge issue. We're going to have to rethink an approach. Something along the lines of what Alex said. Remember when you gave the Ayn Rand Award to Mike Saylor? I don't know if you did the keynote.

55:34Mike did this incredible speech. But those people are probably vomiting right now based on how this is evolving. Well, listen, the bioweapon, I mean, you're not going to create a novel virus that has zero history involved. And there are extensive registries of every virus that's ever been mapped. And so when at an airport, if you identify, if you sequence something and it's not on that registry, you're going to then look at it and LMs or the future bio LMs will be able to look at, okay, this is an infectious agent. This is something that's able to be airborne or water soluble. When you look at the proteins, you can tell what kind of a virus or protein it's generating.

56:21So you're going to be able to learn instantly when you sequence it. And rapid sequencing is here, but we're going to need this. And I think giving up privacy to a large degree, which you've talked about, Salim, right? When you're in an airport, you basically have given up your privacy right there. Yeah, you know you're being surveilled and you know your rights can be taken away at any time. And the one framing of our rear is that we're living essentially in a global airport. I think that continues to some extent. I don't see a way of coming back from that. Well, good luck to the red green folks.

56:52Although Brad Templeton and the EFF folks, they say there is a way of doing it. You don't have to compromise privacy for security. There's lots of mechanisms for solving this in other ways. That's their complaint, that the governments kind of go after the surveillance side just because, oh, this is great, we can surveil people under the excuse of security. But many times you don't have to. Yeah, well, good luck to Red Queen. If I might close the discussion on this, I want to make sure we don't over-index on safety concerns or so-called safetyism. I think these are very important concerns, but I also think that AI can be skilled to combat the concerns, just like one might naively expect the development of like modern cities, that crime would be overwhelming and that humanity would not be able to support itself in urban environments at scale.

57:40It turns out that we are able to. I would also, we're not doing a book corner this episode, encourage everyone to read Werner Vingy's Rainbow's End, which does a glorious job of depicting what the future of AI-enabled biosafety looks like. Amazing. Well, I'm, you know, the eternal optimist here, and I'm absolutely clear we're going to be able to overcome this. Let's move on to one more benchmark here. This is XAI releases GROC 4.1, ranks number one in major leaderboards for reasoning and writing. Back to our resident leaderboard expert. My comment on this one is short. This lead in the text arena benchmark lasted approximately one week and was over.

58:22So my short comment here is the race for the frontier is so intense that even if Frontier Lab is perhaps even bench maxing towards a single benchmark, generalist models seem to be able to push the frontier at this point on a weekly basis. I can only imagine as timelines progress what this is going to look like when these benchmarks are being toppled on a daily basis. Well, I'm sure Grok 4.5 and 5 is around the corner. Let's move on to cursor so cursor triples its valuation in just a few months from june through november going from roughly 10 billion to roughly 30 billion dollars in six months time raised 2.3 billion uh there's michael the ceo of cursor uh who wants to jump in here i mean this is a hot race between a whole slew of different uh coding tools out there this seems to be in dave's wheelhouse Dave, yeah.

59:18Well, I'll tell you, I think this team is phenomenal. And most of the people around here think that they'll rise to the occasion and succeed. But I also think that anti-gravity looks exactly like cursor. I mean, like I actually have both open on my laptop side by side. And other than a little cosmetic here and there, you don't even know which one you're in. And so then you look under the covers and it's like, well, I can access all the models through cursor. and it can only access Gemini 3 through antigravity. So there's a difference right there. But then the bet at Cursor is that the Anthropic and the other models will be worth having and Gemini 3 doesn't just run away with it anyway.

1:00:00So it's really an interesting horse race right now. I'm not going to make any prediction on it because you can't make a prediction on it because their core positioning is incredibly vulnerable, but the team is brilliant. Let's back up. And they're well capitalized. Back up. But for those who don't know what Cursor is or what it does, let's do that basic 101 right now. Dave or Alex? Yeah, so Cursor, I think everyone around here that I know uses it every day. It's the best or has been the best coding assistant that uses AI. It's fully agentic now, so you can just type in a prompt. You can talk to it now, too, and it'll just build things for you.

1:00:38And under the covers, though, they don't own their own foundation model. It's going out to either OpenAI or Grok. It has all of them in there. Anthropic is what I usually use, Cloud 4.5. And it organizes everything. It cranks out the product. It configures your laptop for you. It just makes coding trivially simple. Anyone can do it. And it's pretty universally used. And it was early to market. When I think about the value in this world, where does value aggregate? My list is it's data, scaffolding, user experience, and integration and customization, and then the models themselves. So where would you put cursor in those categories?

1:01:24It's everything other than the model. Yeah, it's all the above other than the models themselves. And compared to Replit, we've talked about Replit a bunch and Lovable. How do they compare it to Cursor? So Replit and Lovable are much more for your mom and pop who want to build like a video game quickly or an invite to a birthday party with moving graphics or whatever. You can build something while you're flying your plane, Peter, like you did. Super, super easy to onboard. Cursor is more for hardcore engineers that are moving to AI and trying to get 10x more performance out of their engineering.

1:02:01I would just note for what it's worth, all of these, or almost all of these, integrated development environment companies, including Cursor, are rolling out their own first-party models. It's almost inevitable that they want to climb down the stack to own more of their software supply chain. And I think the success that we're seeing from Cursor, which is, of course, very exciting, is a reflection that software engineering is probably the first high productivity labor category that's being automated by AI. Won't be the last, but it's the first big one that we're seeing. All right. Keep your eyes out.

1:02:35Surely AI-driven software developers is now the default, right? I mean, you couldn't do it without it now, already, in a few months. To the point where, I mean, this is crazy, but I see companies that are almost treating potential software engineering hires by vintage. Did they get their degree and their experience prior to agentic code or not? Are they spoiled? Have they been ruined? Basically, yes. Did they get their skills? Did they learn? Did they have lots of experience prior to the atrophying that comes perhaps with agentic coding? All right. I'm going to move us forward to another incredible article.

1:03:13This is a new startup funded by Jeff Bezos called Prometheus. Jeff put in$6.2 billion. And by the way, can I just like call out the ability to start a company with$6 billion on your balance sheet has got to be just frightening for a number of startups, right? And it's got to be incredibly accelerating. We've never seen this kind of, you know, starting with billions, multiple billions of dollars on day zero. So what is Project Prometheus? It's an AI-enabled engineering and manufacturing. It's basically learning real-world experience so that it can manufacture efficiently and focus on physical testing and simulations.

1:03:56And I love this other bullet point here. Prometheus has hired nearly 100 researchers from OpenAI, Google, Meta, and other labs. They're just feasting on each other. They're stealing each other's, you know, well-trained— If the going rate is a billion dollars per researcher, then this is really underfunded. They've got six researchers on this. But I find that the two things I found fascinating off the top, we'll talk about the meat of what Project Prometheus is in a second, but is starting with that much money and that they're basically stealing from each other. Dave, what do you think? Well, I mean, it's funny.

1:04:34I have probably 12 meetings with different MIT teams in the last week, you know, 30, 40, 50 at a time. And about half of them are computer science. The other half are not. The other half that are not are saying, how do I get involved? What do I do? What's my AI role? Like, you know, when MicroStrategy started, Mike Saylor was an aeroastro. All the rest of the guys were computer science. The company took off under Mike's leadership. It didn't matter what he studied. AI is like that. There's nothing in the computer science curriculum that teaches you much of anything anyway. So just get in the game.

1:05:06Don't be intimidated. Yeah. And so what you're pointing out here on this slide, Peter, is, okay, they stole another hundred people. Okay. Clearly the industry wants 100 ,000 more, 100 ,000 more people to come in. Why are you letting this guy get a billion dollar signing bonus? Why don't you get into the market, learn this stuff and be there for 100 million? You know, I mean, just get in the game. But it's funny because people get intimidated away from it because they feel like it's all geniuses and I'm going to get crushed. It's just not true. Just get in the hunt. Get into the game. This is the thing happening in the world now.

1:05:44And there's usually only one thing driving all change in the world. This is that thing. So just get into the middle of it and then Jeff will – the other thing I'll point out in this is that there's a tendency to be intimidated by Elon Musk. spent six or seven billion dollars building a massive data center in record time. How am I going to compete with that? But the foundation models that will do parts creation or robotics simulation or whatever are different enough from a large language model that you can build a great foundation model company in parallel with OpenAI and Grok and Meta and Gemini.

1:06:22It's okay. You shouldn't be intimidated by that either. And that's, I think, what Jeff is saying here. Just one final point. Jeff bought all the robotics companies, put them into warehouses, and just ran away with warehouse automation, which created a whole litany of new startups working for Walmart and Target and everyone else, like Symbotic, where Daniela Russo is on the board, does the robots now for Walmart's warehouses. Here, Jeff is saying, okay, Amazon is big enough that I'm actually going to be able to build a multibillion-dollar company within our own universe, our own channel. But that creates opportunity for somebody to be outside of the Bezos universe, doing it for everybody else.

1:07:05And so all that mechanical design is wide open. We're going to have Jeff Wilkie on stage at the Abundance Summit this year. Jeff was the CEO of Amazon Worldwide. There were two divisions. One was AWS and one was everything else. And Jeff Wilkie ran everything else. And he's actually super excited about this because this is what he's doing. He's got a company called Rebuild Manufacturing, which is working in this area too. So, Alex, let's get into the nitty-gritty here, right? So, he's building – Prometheus is building physical AI. It's world models, again, like Fei-Fei Li and a little bit like Genie 3.

1:07:40These are world models understanding the laws of physics and chemistry and engineering so you can actually do real optimization. What are your thoughts here? Yeah, I think we're starting to see the pivot of the capital markets from funding super intelligence to funding that which comes after super intelligence, which is, as I've argued in the past, solving math, science, engineering and medicine. And I think it's a 10x, 100x larger market opportunity, larger addressable market, solving basically everything else after solving super intelligence than solving super intelligence itself. $6.2 billion is a drop in the bucket.

1:08:19I would expect it's going to cost many, many trillions of dollars in funding to solve all outstanding problems in math, science, engineering, and medicine. There's been relatively thin reporting on what Prometheus or what Project Prometheus is particularly focusing on. I have taken note, it seems to be absorbing a lot of old biology friends of mine. So it's possible maybe it ends up focusing a little bit more on biology, a little bit less on manufacturing. But I think this is where the action is after superintelligence. Yeah, I have three points I want to make here. One, this is kind of a shift from chatbots to industrial agents, right?

1:08:55So AI for the office is what we've had. This is AI for the factory floor, where there are physical consequences, where the systems are able to operate the factories because they understand the physical constraints and situations and logistics. The second thing is, you know, I met Jeff in college. I was the chairman of SEDS worldwide at one point. And Jeff was the president of SEDS at Princeton University when I was in MIT. And so space has always been his passion. You know, congrats to Blue Origin for its recent launch and landing. We talked about that last time. But this kind of a physical AI system is exactly what you need to operate heavy industry in space, right?

1:09:42To build factories in orbit, to build factories on the moon and to have them fully autonomous and capable. And then the final thing I would say is that this is going to change. And we've seen companies like Lila and other companies out there that are going to go from invention that happened by serendipitous human creation to invention coming from a computational one. And that's when it gets super interesting. And that's what you've been talking about, my friend, Alex. That's right. What hit me with this is it's felt to me like he's creating a backbone AI for everything in his world. Sure. Amazon, space, logistics, et cetera.

1:10:26Sure. This will service all of those. And Elon will do the same, of course. Mm-hmm. It's like electricity. It's going to run through everything. Yeah. Yeah. Well, also, you know, the foundation model that I built early in my career, it was five years from the day I started writing the code until it was done. I can recreate it now in about two months, which I just did. And so if you look forward a year, that'll come down another 5, 10x. So you can use AI to build the next AI, which is essentially what I just did. The same applies in mechanical design. So if you said, wow, building an entire AI platform that designs rockets or designs robots is really hard.

1:11:09Well, it would have been. But now you can use the current AI to build that AI. And it cuts the time down tremendously. So if you just look forward a year to where the existing AIs will be, that time is actually not intimidating at all. And so it's a good reason to get into the game and build these parallel AIs that work on very specific problems, whether it's biotech, whether it's mechanical design, whether it's futures trading, whatever it is. Build it from the old AI to the new AI. So the last time I asked our subscribers, and by the way, we're almost at 400 ,000 subscribers. So if you haven't subscribed yet, push us over the top.

1:11:47We'd appreciate it. Our march is towards a million. Not that it really matters, other than it'll make my kids really proud of me. So that's my goal. So I asked our subscribers to post questions. You're on your way to Mr. Beast. Yeah, well, hey. Yeah, in about 1 ,000 years. I asked our subscribers to post questions and I took all the comments, put it into chat GPT and asked for it to summarize the most important questions. And there was a critical question that was asked. And I just want to take a second and read it because I want to have an AMA about it. It said, what concrete milestones should people expect to see that prove abundance is coming?

1:12:30In other words, lower costs, new industries, accessible AI tools, and how do we ensure these benefits reach everyone rather than concentrating wealth among a small AI augmented elite? So I want to play a video that was posted on X today, and then we're going to talk about this question. But AI and humanoid robots will actually eliminate poverty. And Tesla won't be the only one that makes them. I think Tesla will pioneer this, but there will be many other companies that make humanoid robots. But there is only basically one way to make everyone wealthy, and that is AI and robotics. All right. So that's Elon's thesis.

1:13:09I posted the question here again. And it's a real concern. Are we going to have runaway wealth concentration? And honestly, if you want me to believe in this future of abundance, you keep talking about guys. What are the concrete milestones? And how do we ensure these benefits reach everyone? How do I know it's actually coming? And yeah, let's jump into this. Can I throw out a couple of points? Yeah. You know, there's an important framing here where we let's not talk about the wealth gap, right? The reason is that the richest people in the world are always going to keep getting richer and the poorest people are going to have nothing.

1:13:47The issue is more, can you lift the bottom? If you lift the bottom, who cares? Yeah. Right. You make this point all the time. All the time. It's my next book, right? I mean, a thousand years ago, the king and the queen on the hilltop lived below poverty today, by the way. And there was thousands of serfs that supported them. And what we've done is – And they died of a tooth infection at age 22. Or they were bled by leeches as the king and the queen up there. And what we've done is, yes, we're heading towards a world where there are trillionaires living on Mars. but if every man woman and child is got access to all the food water energy health care education they could possibly want we've we've lifted the bottom of humanity to a point where mothers can believe their children have access to everything they need that's the world i want to live in that's the world i want to create so let me speak just to that for a second right we forget because we see all this we see people getting richer etc etc but we have to remember the unbelievable benefits accruing to every level.

1:14:49I'll give you a concrete example. When the tsunami hit Indonesia in 2004, all the ship to shore communications were wiped out. And so the government gave cell phones to all the fishermen saying, hey, if you're out fishing, you see another tsunami texted in, et cetera, et cetera. And they found that they're surprised that their incomes had increased by 30 percent over the next two months. So they looked into it and all they were doing was texting in to see what the market price was of the fish. Should they stay fishing? Should they come in and sell? Or which port they go to? Who's paying more? Yeah.

1:15:20So now just that little hint of what Alex would call the inner loop allows you to increase income pretty radically by having the democratized access and demonetized access to cell phones, smartphones, and now AI. And this will change the game completely for everything, everywhere. I'll touch two areas. One is education. You can now sit a child down with a smartphone and say, create a lesson plan for grade seven algebra. And they're going to learn 10 times faster than all the kids stuck in elementary schools in the West that are by law have to go to these things, right? The second is healthcare where any single medical condition can now be diagnosed instantly.

1:15:59And when you get something early, the cost of treating it drops by like 100X. So those two are very concrete areas where AI will make a massive difference in two areas that were traditionally inaccessible. and hard to get and expensive. Let me read the numbers here. So the U.S. average expenditures for family, and this is 2023, was$77 ,000. So the number one cost was housing. 33 % goes to housing, 17 % to transport, 13 % to food, 12 % to insurance and pensions, 8 % to health, you know, 5 % to entertainment and about 2.5 % education. So let's knock these down. Housing, right? So number one, now you can live outside of the city where it's cheaper and be able to telecommute in, right?

1:16:55And reduce your housing cost. There is a future. It's not here yet where we're 3D printing houses, reducing the cost. And what we saw on stage a couple of years ago, if you remember, Salim, was 3D printing houses being per square meter the cheapest, but also the most beautiful and most luxurious because you could get the greatest designers to create a standardized print file for people to use. Transportation, 17%. Well, guess what? An autonomous electric cyber cab is four to five times cheaper than owning a car. It's going to be cheaper than an UberX, cheaper than a bus. So we're going to solve that.

1:17:33Food, we've got to solve food better. We need, you know, basically vertical farms and stem cell grown meats. Let me give you the stat on vertical farms. We, you know, we've been doing horizontal farming since the beginning of time, right? Vertical farming is just crossing over now into economic viability. You can drift feed water to the plants. You know what nutrients the plants need because the sensors know it. You get about seven times the yield of horizontal farming by doing things vertically because you have the right frequency of light hitting it. You save 99 % of fresh water. And by the way, we use 70 % of our fresh water globally to agriculture.

1:18:09So just that. The best calculation we've seen is if you took 35 skyscrapers in Manhattan, turn them into vertical farms, that would feed the entire city sustainably. Nice. Just think about that from a logistics, food security, pesticides, fertilizer. There's massive changes coming down the pike. And this is before we apply AI to the whole mix. So the radical changes coming are going to be so huge that the cost of everything should drop to near zero. The amount of energy you need to feed one person is the amount of sunlight hitting one square meter. And that energy would feed somebody for a year.

1:18:46So all we have to do is get a better loop of figuring out how to convert that energy into consumable foods. And we've got a long way to go. Yeah. You know, health care, 8 % of our cost in health care. You said it already. We know that an AI physician, diagnostician, is significantly better than any, and even in the best, you know, physicians. And an autonomous robot eventually will be the best surgeon, and the cost of that will be capex and electricity. I mean, it's hard for people to believe this stuff now because it's on the bleeding edge, literally. But we're going to get there. Entertainment, 5 % of cost.

1:19:27Well, guess what? I mean, YouTube, what else can you want? Education, you mentioned before, AI, YouTube, all these things. So we're demonetizing and democratizing this stuff. It's just hard for people to realize it. I think the challenge is we compare ourselves to the Kardashians, right? We compare ourselves to people that we see on TV and on the internet all the time versus comparing ourselves to what it was like for our parents or grandparents. Yeah, I think that last point is the key one because we've had dirt cheap food for a long time, but everybody still wants a$14 Starbucks latte, which you don't need to pay for, but there it is.

1:20:10and why do I feel that need? So the metric I'd be tracking is actually depression rates. And, you know, because I think AI properly deployed can hit that much more quickly than it can hit robotic automation that creates new homes for everybody, you know, that are 10 times larger. That's a great point. And so I'd be looking at that one as an early indicator that we're on the right path. And it's not a no-brainer. You've got to really think it through because, you know, you mentioned rent is at the top. You know, 33 % of household income gets spent on housing on average. But when you look below the poverty line, I think spend on drugs, alcohol, and gambling is three – the opioid addiction alone is a trillion-dollar error, I guess.

1:20:54And it's about five times more collectively than rent. I see – Go ahead. Go ahead, Slater. Well, no. So I'd be attacking – if you want to come bottom up and say, look, we want to create universal happiness with AI. We've never had a tool that could attack it before, right? You can attack manufacturing automation. You can make food cheaper. You can have harvesters that mow down half the Midwest to create wheat. But all that does is create more of stuff that's already abundant. Now AI is the trigger for a massively more thoughtful way to create universal happiness. And I would start with depression rates and work up from the bottom because you can do that very, very quickly, much more quickly than you can.

1:21:36We've already looked at the robotics. we know that it's going to build a mansion for everybody in the world. But we're not going to have the robots for about 15 years because we have to scale them up on this exponential curve. So, you know, some people will have them next year. You'll have yours this year. But we won't have enough of them to attack the global problem for about 15 years because of, you know, just the manufacturing ramp up rate. Alex, this is all about benchmarks. We've talked about this. You and I have been working on a paper on this subject. Can you speak to that? I think it's so simple.

1:22:06I think what's upstream of all of these other milestones is the dollar cost per unit of intelligence. And as we've discussed previously, right now, that's hyper deflating by something like 40x year over year. So to keep the party going and to make sure that all of these downstream considerations, cost of living, health care, housing, etc., that these all hyper deflate ultimately alongside cost of intelligence, I think it's largely a regulatory and social concern. And we've spoken previously about, for example, the difficulties of getting Waymos in Boston. That's a regulatory consideration. The cost of intelligence needed to autonomously drive cars around, that's making excellent progress.

1:22:48But ultimately, in order to, say, provide essentially free, autonomous, on-demand transit to everyone, there's a regulatory bottleneck. And in order to ensure that the benefits of intelligence too cheap to meter become evenly distributed, I think it's going to require some revision of social coherence and the social safety net to make everyone comfortable with the downstream consequences of intelligence too cheap to meter, including healthcare and housing and energy and utilities too cheap to meter. Yeah. I did a calculation. Okay. So if you wanted to have a reasonable life, you could do it for $20 a day in Bali.

1:23:33Housing costs about$10 a day and your meals are literally about$2 a day and then a bit of extra. So for about$20 a day you could do it. If you had half an Ethereum, which is about$2 ,000, you can put it into DeFi trading pools and earn about a percent a day, which is about$20. So half an Ethereum of capital allows you to live crudely, but allows you to live in a very lovely spot in the world for near very low cost. And think about just that feedback loop on that, because as you double that, if you triple that, if you 10x that, all of a sudden you get into a really great place. You can survive today on a very small budget anywhere.

1:24:14My feet are in the sand. My feet are in the sand already. Yeah, I'm ready. And of course, Salim, that Ethereum comment was not investment advice, just to let everybody know. But it is interesting that Harvard has doubled down on Bitcoin. And now that we're in the Bitcoin doldrums, it's nice to see the institutions. I mean, I remember when we went from like, you know, wacky individuals buying crypto to now institutions and financial institutions and sovereign funds and so forth. Countries. Awesome. Not investment advice. All right. What an amazing episode. And we've actually just gone through half of our stories.

1:24:50But I think to make this consumable because the feedback we've gotten from folks is please try and keep the episodes under an hour and a half. So we're listening over all your comments. We're trying hard. So we'll have to spin up another conversation on everything going on in data centers and energy and space and so much. I mean, it's hard during the singularity to keep up with everything going on. Just the mind-blowing stuff from Gemini 3 was worth covering properly. You know, just a reminder, last summer, not that long ago, Polymarket said everybody, you know, the top five had an equal shot at being the best AI model by the end of the year.

1:25:29Now it's 91 % Google. But by next summer, that's down to 60%. So it's kind of like 50-50 that someone else will take a lead by next summer. Leap frogging. Well, that's what we should hope for because Alex said the key point, as usual, 40x is what you should expect next year. 40x. people really struggle 40x in anything so if the cost per intelligence comes down by 40x or they're just raw intelligence goes up by 40x next year you should expect that very hard to visualize all that all that that means so we do everything we can on the podcast to try and make that tangible for people but really try and digest that you know coming out of this gemini three incredible breakthrough.

1:26:10Yeah. And just hats off to Josh Woodward, to Sundar Pichai, to Demis Hassabis for an extraordinary job on Gemini 3. Just so proud of what they've been able to create. And of course, a lot more coming. I have one announcement. Yeah, please. Sometime in December, we're going to do a Meaning of Life session online. So I've had enough clamoring for my community and other people and people that go to Abundance that people want to do it. So stay tuned. We'll get more details next time. Well, we'll do it also at the Abundance Summit on Wednesday night. This is Salim waxing poetically and philosophically for about five hours straight.

1:26:58It's like a late night French salon type discussion of alcohol or equivalent mandatory on the metaphysics philosophy. And what does it mean to be alive in today's world? It's at 10 p.m. What time does it end? Dawn? It depends on the audience. But the crazy ones, we've gone till dawn. Oh my God. Because we never get a structured conversation on the meaning of life. We never get that. So let's have that conversation. Well, we'll do it. You'll do it. And I'll join you for at least until my bedtime at nine o 'clock. And I'm exiting the building. But I'm going to do it online in about a month. So we'll get details out there.

1:27:31Okay, well, we'll do it earlier. And last time we talked about the potential for a moonshot gathering. we've had 500 of you email us. So if we get to a thousand, if you're interested in a moonshot gathering next fall, you can send an email to moonshots at diamandis.com and let us know you're interested in having these conversations and gathering with other moonshot listeners. And once again, we put our call out for outro music. And this is a piece by John Novotny and it's called Moonshots Metal Version. but here's the key. You need to see this. This is not just music. This is a fun video. Saleem, you look so sexy.

1:28:15Dave, and I love your ponytail, Alex. AWG's got a ponytail in this and he's rocking it. All right, on our outro, let's go ahead and watch and listen to this. This is heavy metal moonshot music. Oh my God, I haven't seen this.

1:28:47Oh my god! No!

1:28:53Oh, that's a good one. Very Jensen.

1:29:07Oh, not to miss. Oh, my God. Frightening.

1:29:13This is amazing.

1:29:23Bobbling the lightning.

1:29:27I've got to get the ponytail.

1:29:31You know, Peter and Salim, your look in that video was really good. You should just do that. I love Salim with the sunglass move. Dave, you on the guitar. And AWG, you on the keyboards. And that ponytail was you, buddy. You've got to grow that ponytail. Apparently so. All right. We've got some good listeners. Thank you, John, for that. That was amazing. Yes. It's DB2, AWG, and Mr. EXO. Have a fantastic week. I love doing this. And thank you to all of our listeners. Great episode. All right, take care, Peter. Take care, guys. Every week, my team and I study the top 10 technology metatrends that will transform industries over the decade ahead.

1:30:14I cover trends ranging from humanoid robotics, AGI, and quantum computing to transport, energy, longevity, and more. There's no fluff. only the most important stuff that matters, that impacts our lives, our companies, and our careers. If you want me to share these meta trends with you, I write a newsletter twice a week, sending it out as a short two-minute read via email. And if you want to discover the most important meta trends 10 years before anyone else, this reports for you. Readers include founders and CEOs from the world's most disruptive companies and entrepreneurs building the world's most disruptive tech.

1:30:47It's not for you if you don't want to be informed about what's coming, why it matters, and how you can benefit from it. To subscribe for free, go to dmaddis.com slash metatrends to gain access to the trends 10 years before anyone else. All right, now back to this episode.

1:31:16The Burlington Coat event is back. Buy a coat now through November 29th and get$5 off your merchandise purchase of$25 or more from December 6th through December 24th. Plus, Burlington is donating 50 ,000 new coats to help those in need nationwide through our partnership with Delivering Good. We've been keeping you warm for over 50 years. We'll see you at our coat event. Burlington. Deals. Brands. Wow. Visit burlington.com slash coats for details.