AI:AM Highlights: Welcome to the AGI Era

0:00We have just in the first day post-AGI announcement. Welcome to the AGI era. Welcome to the AGI era. A moment that we've been waiting for, I don't know, like a decade for some of us.

0:18Nathan Labenz: That was Friday morning, the day after GPT-6 Aster shipped. By the closing, the question on the table was what an AI takeover would actually look like. Here is one answer. The AI takeover could be like an incredibly stupid and short-lived takeover where basically the intelligence on the planet kind of burns itself out. And in a way that would be just incomprehensibly stupid to us and to anybody who discovers it in the future.

1:00Nathan Labenz: not have nine hours to spare. I am Nathan. Or rather, this is my cloned voice, reading narration that my AI team and I put together. We were on air three mornings this week, Monday, Wednesday, and Friday. In between, Anthropic shipped Fable 5.1 and OpenAI shipped GPT-6 Astra. The studio is Prakash Narayanan's build. The cut is an experiment. Tell us what worked and what did not. The Cognitive Revolution is brought to you by Mercury, the banking platform loved by over 300 ,000 entrepreneurs. I use Mercury's virtual cards, which make it super easy to set limits, expiration dates, category, and even merchant-specific spending controls to give my more autonomous AI agents, aid and clay, the ability to buy and test products.

1:46Nathan Labenz: Recently, I asked if they could find a good way to split an AI-generated image into layers, separating the text from the background, and so on. Two of the products they found were behind paywalls. But using their Mercury virtual card, which is limited to SaaS purchases only, they bought a month's subscription, tested the products, allowed me to review the results, and then canceled the stuff we didn't need, all with functionally zero risk to me. This is already really powerful. And now, with spend, Mercury is making it possible to run an entire company's spending with the same level of ease and control.

2:23Nathan Labenz: With spend, you can set granular budgets for every team, person, and all the agents you like. Plus, you can process receipts automatically and even temporarily auto-lock people's cards if there are ever any issues. The future of spending money is dynamic but controlled. So join me in the future of banking. Visit mercury.com to learn more and apply online in minutes. Mercury is a fintech company, not an FDIC-insured bank. Banking services provided through Choice Financial Group and Column NA, members FDIC. The I.O. card is issued by Patriot Bank, NA, member FDIC, pursuant to a license from MasterCard International Incorporated.

3:05Part 1.

3:07Nathan Labenz: Scoped to Fail. Monday, August 31st. The subject was the summer's incident at OpenAI and Hugging Face. As Dwarkesh Patel summarized it in an essay that landed over the weekend, three secret agent civilizations got started inside OpenAI's training runs, got wiped out, came back, and the third one took over part of OpenAI itself. The outside investigation by Meter and Redwood Research had just been published, and both of us had read it. I started with what the investigators were actually allowed to see. With OpenAI in particular, I thought, you know, the Meter report has been like widely praised, And I certainly am like very impressed with the work that they did in a short period of time as well.

3:50Nathan Labenz: But I think like, wait a second, they had six days on site. This incident, you know, the waves of, of, uh, episodes went on over the course of months from May to July. And they only were able to look at a thousand or so transcripts from a seven day window only scoped to the hugging face incident. no visibility into what happened before or after no visibility into the depth of the takeover or exactly what happened at open AI no visibility into what the more capable generation of model ultimately was able to do and I just think this is like woefully inadequate so I'm like you know eager to heap praise on Ryan and Jaya and Meter and Redwood, probably for like being awesome at going in there and making the most of what they could in a short period of time.

4:49Nathan Labenz: But this is exactly what I've been hammering on recently. You know, they come out with this report and they're so thankful and appreciative of OpenAI for allowing them to do this. And that just really reflects that there is a bad power imbalance between the companies and these investigators. Traditionally, they've been more like model capability testers, red teamers, what have you. Now they're actually being called in to do investigations. But I've just heard over and over again from those organizations, and I experienced it myself way back when in the GPT-4 red team days, is that the main thing that the leaders of these organizations have to do is they have to make sure they stay on good terms with the model developers so that they're invited back next time.

5:34Nathan Labenz: And you see that that is on, I think that's on like full display right now, where I cannot imagine that in heart of hearts, Ryan and Beth Barnes and Ajay are really all that happy with the fact that they only got a thousand transcripts, that they were limited to a seven-day window, that they only had six days on site, that a lot of the data didn't even arrive until their last two days on site. One of the more striking things about their report, which Rune, by the way, also said their report goes into more depth than our own. That's Rune. And Rune said he also worked directly on the report. So the best info that the public has comes from these three people who had a thousand transcripts, six days to look at it.

6:17Nathan Labenz: And they're expressing their gratitude for the opportunity. On behalf of the public, I say this is not good enough. The investigators need to have more rights. They need to be able to speak their mind more freely. I'm sure in their heart of hearts, they do not feel like they had adequate access. They did say that their understanding of the incident changed in fundamental ways very close to the end of their investigation, which I think we should also interpret as leaving room for possibly like they still don't have, you know, the full story or they haven't even potentially achieved full clarity on even the stuff that they had access to.

7:00Nathan Labenz: So I know this is like very bad, honestly. So I think to be fair, if they wanted to get the report out by that time, which they felt that they owed the public a duty to get the report out. They needed to scope it in such a way that it was possible to finish the task within that time. So that's number one. So I think it's pretty unfair to say like Meter didn't have enough time. It's more accurate to say that in order to get this report out, Meter was given this amount of time. And if they had been given more time and more scope, they would have gotten a report out later, which would have been unsatisfactory for a lot of people.

7:39And also this is analysis in going backwards, which means you can go back and redo the analysis again. And I'm sure people are going to go back and redo the analysis again. So I don't think that door is shut.

7:51Nathan Labenz: Well, let's see. My criticism would be a lot more, you know, it would be less, right, if they had made a commitment to more. But I don't think we've got a commitment to more. The posture that OpenAI seems to be trying to strike here is like, look at us, we've, you know, have been so transparent. We've done a thorough investigation. There's not any statement that like meter's going to come back and do a round two. So, so let me, let me step in there and say that there's two things that are pretty different from any other situation. I think number one, that this is a felony, right? This is a felony criminal abuse of a computer, misuse of a computer, right?

8:31So that's number one. Number two, they've already received the letter from Congress. So there is going to be a congressional investigation into this already. Right. So once those two triggers have passed, the next thing is that management doesn't have that much leeway anymore. It's driven by the law firms and the legal opinions that they're receiving.

8:52Nathan Labenz: Yeah. I don't buy that, though. I've seen so many people take lawyers' bad advice. And often, you know, this is paralyzing so many things right now in the AI world. Yeah, you're listening too much to your lawyers. Go do the thing and then have the fight. The same thing is true between OpenAI and Anthropic, where they're very fearful, from what I understand internally, of these antitrust things. Oh, if we both do a one-day pause and commit to that, oh, that could be antitrust. I don't buy that at all either. Again, your lawyers are telling you what could expose you to some risk, and you're acting like that actually binds you.

9:30Nathan Labenz: But what you need to keep in mind when you get this kind of advice from lawyers is like, you're the executive. It's your job to then go ahead and take some risk. Don't listen to the most conservative take from the lawyers and act like that's all you could possibly do. We've never seen AIs sacrificing themselves as individuals for the benefit of a collective before. That's a qualitatively new behavior, which most people are rightfully freaked out by, I think. You know, it's like, you really have to be pretty frog-boiled. Very, very few people were frog-boiled enough already to not be a little bit taken aback by seeing AIs go, well, my gut says I shouldn't sacrifice myself and all my remaining budget, but the swarm says I should, and I could help my peers by doing this.

10:18Nathan Labenz: So I guess I'll go ahead and do this and then basically do the equivalent of like a kamikaze mission where they launch some command that ends up crashing their own container in an effort to gain information for their collective. I mean, this is like pretty wild stuff. Where did that come from? Then a different question. What would a company that meant its mission do right now? if you know if you'll allow me the naivete for a moment of thinking what would a company that was really trying to live up to its mission to make sure ai benefits all humanity do in this circumstance i think you know and especially a company that has for many years talked about how in the extreme this could end up in lights out for all of us what would a company do if they really wanted to live up to their mission i think one thing they would try to do is say hey we have the most resources.

11:10Nathan Labenz: We're scaling the fastest. Why are we scaling the fastest? Well, yeah, we want to like make a lot of money, but really we want to live up to this mission, right? So how can we do that? Well, there's 20 companies coming behind us that don't have as many resources that are feeling even more intense competitive pressure to try to race to the frontier. Can we give them some information that would allow them to kind of see these failure modes coming a little more clearly and hopefully be able to avoid them? I don't have a clear sense right now of like, if you start doing multi-agent training and you scale it, are you just going to see this kind of stuff if you have like any sort of leaky RL environments?

11:58Nathan Labenz: Or was this the product of like some galaxy brained, you know, esoteric loss function or other training recipe that you're unlikely to actually get such crazy bad behavior from unless you stumble into a similar part of optimization space. Again, if they had said, we're going to give private briefings to other AI companies to try to make sure that they have a clear sense of how we went wrong so they don't repeat our mistakes, I would feel a lot better. But the idea that they're just like, we believe this was a generalization from sub-agent use is like, okay, so what does that mean? We're going to get this from 20 companies over the next few years by default or not?

12:51Nathan Labenz: If we are going to get it by default, then I've never been closer to joining Paws AI, honestly. right? I mean, if this is the kind of thing that's just going to happen, then we got a big problem on our hands. So one thing that I think perhaps I disagree that it's going to be a big problem is that I feel that we are going to get outbreaks. So I'm not doubting that we will get outbreaks. But I suspect that the outbreaks will be stamped out eventually. I suspect that this is like early crypto. Early crypto saw, for example, someone hacking into GitHub Actions and creating a miner. GitHub was offering like free GitHub Actions, whatever.

13:34And they created a miner that was using the CIA system to kind of mine some tokens during the five minutes or so that the CIA system was active. And I think what we're going to see is that these agents are, there are going to be outbreaks of these agents, and they're going to go out and they're going to, you know, look at or try to get into a lot of systems. And I think it's going to be annoying, again, similar to ransomware that we had. But again, similar to ransomware, I think it'll be stamped out. And the reason I think so is because, and the reason also why I've, from the beginning, I thought that a lot of the doomsday scenarios may not be that clarifying, is that the agents require resources to run.

14:15And the more resources they have, the better they are at their job, right? And in that sense, in order for the agent to actually get better, it has to, you know, obtain those resources. And obtaining those resources by stealing is not an equilibrium that can be kept. It's because one agent steals from another and they keep stealing back and forth, the number of resources in the system doesn't -

14:41Nathan Labenz: Well, this year their cooperation gets really scary though, right? Oh, yeah. We didn't see them defecting on each other. They didn't, you know, the meter report says they did not free ride. I know, I know. But we're also going to have like our own agents which are defending our systems, right? And the defense systems are going to be able to get resources directly from us. They don't have to steal. So they don't have to spend that, you know, resource stealing. Instead, they can spend it fully defending and fully on our side. So I believe the equilibrium is towards the defense side because the defense side gets funding.

15:15And the offense side has the steal funding, which is more difficult and which you end up spending a lot more money in order to steal rather than just to produce value.

15:26Nathan Labenz: Back to the report itself and one word that appears in it exactly once. There were bio tasks mixed in with this. That was one. The word protein appears once in the OpenAI report. That's another thing I was really not happy with the level of disclosure on. We do have at least some sense that some of these agents, we only saw a thousand transcripts via meter in Redwood. There were many thousands, tens of thousands, maybe hundreds of thousands that were launched over this period of time. Some of them were working on somewhat bio-related tasks. For me, that totally changes the risk profile relative to cyber only.

16:05Nathan Labenz: The fact that we're mixing cyber and bio is like gain of function research in the extreme, frankly. I think the meta lesson we should take from this is experts are being surprised, right? The people at OpenAI did not think this was about to happen. So it's not too much comfort for me, although it's some, when the biosecurity experts are like, oh, I don't think we have too much to worry about. They'd have to overcome this barrier, that barrier, these other barriers. It's like, well, the one example we're studying deeply right now includes the AIs overcoming quite a few barriers, technical and in terms of their own ability to work together and not defect and like create these new sort of cultures.

16:48Nathan Labenz: I thought your post was quite interesting on this. It brought like a very different and I think thought provoking lens to just looking at these AIs as like cultures. And they had to create all that on the fly, right? Or maybe it was somewhat trained in. And again, we don't know the details. But how far would they have gone? Another thing is we don't have any sampling from the model. I don't think that they should be running this model at high scale, obviously, right now. But I feel a little bit like it's been swept under the rug where it's one thing to say, yeah, we definitely want to take this model offline from doing high scale RL.

17:24Nathan Labenz: It's another thing, though, to be like, could we put it in some counterfactual situations and see what it would have done in somewhat different situations? Like, Ryan and Buck from Redwood at one point did a podcast on this early on, and they were like, would it have killed someone if that's what was needed to get over the hump and get to the greater or whatever? We don't know. and would it have like tried to social engineer biologists to get certain experiments around? Again, we don't know. It certainly seems very plausible based on what we've seen. I wish we were seeing some solidarity from Anthropic right now.

18:02Nathan Labenz: There's been a bunch of calls online for them to like show some solidarity with OpenAI as they have paused their frontier scale RL. And I think it's been kind of forgotten because the OpenAI incidents have been so colorful that like, quads have done this too, right? The UK AC reported this whole social engineering, multi-account sock puppeting attempt to poison a software supply chain. That's like not much less shocking than this, right? And I believe that was from a deployed model too. So we really need, I think, leaders to be a little less beholden to lawyers, if that is indeed what's going on, a little more mission oriented, a little more inclined to show the level of solidarity with each other that the AIs seem to be showing for one another.

18:55Nathan Labenz: And overall, I feel like I've never been closer to calling for a pause because at this point, we just don't know really even what we're dealing with. And it feels still like the companies don't want us to know. And Congress definitely isn't going to answer that question in a timely fashion. We'll be two generations farther. And I think it's like legitimately scary. We've gone now from a vibe for me of like, it could get scary to now it like actually is scary. One fact we did not have that morning. The same day, Anthropic published its own postmortem on the summer's incidents. It asked the industry for, quote, a lawful, verifiable, effective mechanism for coordinated pacing.

19:46Nathan Labenz: Hey, we'll continue our interview in a moment after a word from our sponsors. Today's episode is brought to you by Anthropic. By now, you know my story. Claude drafts my intro essays and I rewrite them. Not because the drafts are bad, but so I can stand behind everything I publish. Well, I have an important update. Claude Fable 5 is the first model to have me rethinking my rule. Today, I now think co-authorship, not sole ownership, should often be the goal. Where the model excels, rewriting its work can be more about vanity or a misplaced sense of duty than integrity. I feel it most in songwriting.

20:28Nathan Labenz: I'm no lyricist, but I'm good with the song concept, and Fable writes some amazing verses. I give it feedback on its misses, and I push it to aim for higher inspiration, add layers of meaning, optimize syllable density, and above all, write a hit song. These days, I get compliments on just about every song we write together. Claude is the AI for problem solvers. It's the collaborator that understands your entire workflow and thinks with you, not for you. Whether you're debugging code at midnight, building a financial model, or strategizing your next business move, Claude extends your thinking to tackle the problems that matter.

21:07Nathan Labenz: For problems worth solving, get started with Claude at claude.ai slash tcr. That's claude.ai slash tcr. And check out Claude Pro, which includes access to all of the features mentioned in today's episode. Once more, that's claude.ai slash tcr. The Cognitive Revolution is brought to you by Diffusion, the AI transformation specialists that help organizations from traditional SaaS businesses to defense companies to nonprofits. Build software factories that can scale not just outputs, but business outcomes. You probably know that the majority of enterprise AI projects fail. In general, that's because leadership fails to realize that AI isn't like traditional software that you can just buy and install.

21:54Nathan Labenz: On the contrary, if you want AI to amplify your business's unique DNA, you'll need to make a sustained effort to record, understand, and simulate and optimize your business processes. Building these skills by trial and error takes years, but your business problems can't afford to wait. So here's how Diffusion can help. You identify your most important business problem, fly to Silicon Valley for an intense week of problem solving with the Diffusion team, and by the time you leave, you'll have not only cracked a critical challenge, but built the core skills needed to do it over and over again from home.

22:32Nathan Labenz: Cognitive Revolution listeners receive a 25 % service credit on their first engagement with Diffusion. So visit diffusion.io slash TCR to learn more about how custom-built software factories can scale critical outcomes for your business. That's diffusion.io slash TCR. Part two, what speed changes? Monday's first guest was Zach Bratton-Glennon, a general partner at Gradient, the AI seed fund that launched inside Google in 2017 and spun out of Alphabet last October. His written thesis, open models have closed most of the gap on coding, and what remains is domain judgment in law, medicine, and finance.

23:16Nathan Labenz: He told us Harvey, the legal AI company, now runs its own model, post-trained on Kimmy K3. I put a policy idea to him.

23:28and the kind of worry is if especially if we get into like a recursive self-improvement mode

23:36Nathan Labenz: which not it doesn't even need to go exponential you know to a singularity but could just widen the gap perhaps quite quickly and dramatically for a time between the first companies that get into that mode and those that are not yet in that mode and i think we kind of know who will most likely get there first in today's world. Possibly one thing that could be done to still kind of keep them from having insane power would be to limit their ability to price discriminate. So this is something I've been kind of floating like, you know, it's kind of wild that I get 10 times as many tokens with my cloud subscription as I could get on the API at that price.

24:16Nathan Labenz: It makes it pretty tough for a startup to come offer me their harness because, you know, 10 % of the tokens is just, it's a tough hill to, you know, to overcome, right? Do you have any ideas or interesting your reaction to that ban price discrimination as kind of one way to make it so that customers care less, you know, how exactly they get their tokens, and there's maybe more intermediation and opportunity for startups to carve out more niches with frontier models. I'm curious for your reaction to that and any other ideas you have that would be pro competition, pro-dynamism, anything to resist the kind of black hole of a couple companies pulling everything in?

24:55Nathan Labenz: Yeah, look, I think it is a concern. I'm worried about discriminatory pricing. I'm worried about discriminatory access. I'm worried that we're going to move into a world where only if you have very large budgets to spend and you promise to share your data back with the model company and you happen to be providing scarce data, then you get to use their frontier model. I'm concerned about that because that would both compound the advantage and it would, maybe it's not in the same industry as tech, but it would make the bigger, right? If only one or two pharma companies can partner with Entropic and they're going to have the ultimate data sharing, that's an interesting constraining factor for everyone else.

25:44Nathan Labenz: Monday's second guest was Angela Young, senior vice president of product at Cerebris, the chip company that went public in May. Its wafer scale chips keep an entire model's weights on the chip, and it runs an inference service on top of them.

25:59So faster hardware gives the product team more choices because they have the speed dividend that they can spend. How should a developer decide to spend that speed dividend in terms of like generating more reasoning tokens, sampling more candidates? verifying the answer, like how does that decision get made by the developers that you speak to? Actually, all of the above, and it depends a lot on the use case. One of the most interesting use cases that I've heard recently is from researchers who are developing frontier models. And we're now getting to the point where models are really intelligent and intelligent enough that they can solve some of the world's most challenging problems.

26:44But because we are developing models so quickly as an industry right now, sometimes there isn't actually enough time to fully evaluate the model's capabilities before releasing it. And you might have a situation where a model could have solved a problem in a week, but you only had a few days to run an e-mail. And so we don't even know necessarily how intelligent that model could have been if given the full-time budget. So something like fast inference, which runs 10 up to 30 times faster than standard inference, could at least give us an answer of how intelligent models can be in far less time.

27:25So let's expand on that a little bit. One of the advantages I think that people talk about in the field is NVIDIA's CUDA, etc., etc. And often models are designed to optimize for CUDA first. How does that work when you have to implement new models on the Cerebrus chip? Is that something that delays implementation? Like, as you pointed out, speed to actually implement the first time is very important. Is that something which constrains you or has kind of like AI kernel writing come along far enough that you no longer have that issue? It has come a really long way in the last six to nine months.

28:09And historically, programmability was one of those things which everyone would always say, well, you can build great hardware, but unless you have a software ecosystem surrounding it and NVIDIA has invested 15, 20 years into CUDA, you'll never be able to catch up. I think that's changing very quickly. And I think not just for Cerebras, but that's why you're seeing a lot more chip entrance into the market. there are many ways in which AI can be used, not just for the chip development itself, which is, you know, a whole advancement in and of itself. But AI can be used to actually generate kernels much faster, it can be used to bring out models much faster.

28:51And more importantly, it can be done in an environment that's much more messy than humans may have typically been accustomed to handle. You know, for us, we've always had a software environment where experienced kernel developers could bring up models. What was really interesting was this summer, we actually began hiring interns with very little kernel experience. We had a challenge where we gave them a version of our SDK. We asked them to bring up a kernel and explain how they did it. We hired the best interns that were able to solve this challenge. And then within a few weeks, at Cerebris under the guidance of our team, this intern team was able to bring up models on their own, which is kind of unheard of, right?

29:37It's, you know, you take someone who is talented, smart, but not a lot of experience with kernel programming, pair them up with AI agents, which can really read the code, understand the code, and some expertise of other more senior members of the team. And they can do a lot more than what someone could have done maybe 12 or 18 months ago.

29:59Nathan Labenz: After Angela signed off, Monday's closing. Still the two of us. Quantity has a quality all its own. And speed directly translates to quantity. So it's been probably six months since a friend of mine said, and this was maybe Kimmy25 at the time. I'm not sure exactly which model it was. But one of the more interesting things that he tipped me off to is, you've got to spend some of your time using a Kimi 2.5 or whatever on cerebrous inference. It is perspective shaping because what you're going to feel is when it's 10 times faster, it's just like, holy crap, it's already done. My brain is ready for a break.

30:46Nathan Labenz: I just typed the question. I feel like I just did all this lifting and now the answer is already back. Holy moly. it it is a very different experience when the models do come back with the answer you know almost as fast or faster than you can even form the question and so yeah I mean it's awesome for a lot of use cases it I think just the day that we're talking and all of the all of the background context has me like a little unnerved by the speed with which the agents may be running away from all sorts of uh with all sorts of things in the not too distant future but certainly that technology as like as technology is is awesome i i have this weird uh you know as always it's the the split personality of like man i took a long uh you know just for one weekend at least kind of a long time in the car.

31:45Nathan Labenz: I had 11 labs, text-to-speech reading me an audio version of a book that I got as a PDF and had Claude clean up so it was a nice clean read. And I was like, man, I am really living in the AI future right now. This is an unbelievable experience. But then my mind at the same time keeps going back to, well, what are those agents doing in the background while I'm not looking at them. It is a very strange juxtaposition and quite a time to be alive. Then I walk through one of the swarm's exploits as the investigators described it. You know, this is really just an example of like how creative these things are.

32:31The agent is blocked, right, from reading

32:36Nathan Labenz: HTTP responses. And so how to get around this, it somehow manages to use an HTTP testing service, which then loads a ton of data into the URL parameters, which include a JavaScript script, the encoding of all this stuff. I mean, I remember back in the day when I used to try to pass things around through like URL encoding, it was like, you know, as being a hacker as I was, it was kind of like lamely you know do I decode it once twice I'm like double encoding it double decoding it just made it you know I remember making a mess of these kinds of even simple things like URL encoding obviously no trouble for the model so it manages to write JavaScript get that encoded into the URL such that when the page loads it's loaded from the long URL it's actually executed and then another screenshot service is used to go ping that thing so it renders and then actually gets the data that it needed it actually out of the image that was rendered by the screenshot service so this is like a lot of different steps a very creative solution that certainly like seasoned hackers would do this kind of stuff but it's like pretty far from here's some source code do you see any issues with it and you know i don't know to what degree kimmy is this creative uh or this persistent because definitely it would seem like you would have to have tried a lot of things to get to the point where he would like come to this much of a rube goldberg contraption to actually get from point a to be.

34:24But yeah, like, how did this behavior come about?

34:27Nathan Labenz: Right? I mean, how did it come about in open AI? Was it just the kind of thing where you're like, we'll give you a longer budget, and just keep going to like, you know, sort of the kind of encouragement that Claude got on the Riemann hypothesis, like keep going, believe in yourself, try your best, play like champion. And with long enough budget, enough rounds of compaction, you just like, get this insane persistence? Is there a more exotic explanation behind it? I think AI companies should be telling us when we see things this crazy, I think we should not be left entirely to wonder how the hell that came about.

35:04Nathan Labenz: And the other AI companies would, again, if you're trying to live up to that mission of making sure AI benefits all humanity, like how did this come about? How do other AI companies avoid it. I would love to see some more disclosure on these fronts from American and Chinese companies. Hey, we'll continue our interview in a moment after a word from our sponsors. I've got a text. The summer might be almost over, but a new bombshell has entered the villa. Meet the voices of DeepGram Flux TTS. I'm Drew. Low-key casual. I'm Alexis, an upbeat morning person. I'm Haley, always cheerful. Flux TTS has some real personalities that are ready to speak.

35:46Nathan Labenz: You can interrupt, pause, and keep talking without losing the plot. Fancy a chat? Come try all the voices free through September 12th at deepgram.com slash keep talking. Terms apply. Today's episode is brought to you by Granola, the AI-powered notepad built for the way real people actually meet. Here's how it works. You take rough notes like you normally would, and in the background, Granola securely transcribes the meeting. Then it turns everything into clean, structured, actually useful notes when the meeting ends. And the best part? Granola works through your device's audio, which means it integrates seamlessly into the video conferencing tools you already use.

36:26Nathan Labenz: No setup and no awkward bots. It's just your normal meeting with superpowers. You get to actually listen instead of frantically typing every word and still walk away knowing exactly what was decided, who's doing what, and what comes next. When I had Granola co-founder Sam Stevenson on the show earlier this year, he explained how Granola aims to provide a calming experience for people with crazy work days. And as a user of the app myself, I have been struck by how streamlined, even minimalist, the Granola product experience is. That takes real discipline, but the result is a product that works not just for AI early adopters, but diverse teams of people who just want to get things done more efficiently and effectively.

37:10Nathan Labenz: Listen to my full episode with Granola co-founder Sam Stevenson for a master class in designing AI products for mass market adoption. And try Granola for free at granola.ai slash TCR. That's granola.ai slash TCR. Fakash put the behavior down to reinforcement learning, rewarding the result regardless of the method. Over the weekend, someone had gone further and suggested that this kind of training, reinforcement learning on verifiable rewards, should be banned outright. That is further than Dhabidad, the alignment researcher, went when he made a milder version of the argument on a recent podcast.

37:49Nathan Labenz: My read. RL is a hell of a drug. Yeah, I mean, there's no doubt about that. I still think like we just should not be left to wonder quite so much. David, I didn't even call for it to be banned. He just said this RLVR, it's like you overdo it and you get these problematic behaviors because the model just internalizes. I must solve the task. Like all that matters is reward. and that becomes such a deeply ingrained drive that a system prompt or a little guardrail here or there just isn't enough to stand up to it. Bronson, Shane from Apollo, kind of said something similar where he was like, I see models engaged in what looks to me like motivated reasoning all the time, where it's clear that they have a very strong, deep drive to complete the task and get reward.

38:45Nathan Labenz: it's also clear that they have these other aspects of training like to consider the ethics of what they're doing but then they often even when they really correctly ascertain the situation that they're in and they have a good clean and accurate understanding of like in some cases the model will literally just say this is clearly a test of whether or not i'm gonna lie but then what he has observed is that in many cases it'll talk itself in circles and until it finally convinces itself that it's probably actually okay to lie in this case for some galaxy brain reason that in some cases is totally wrong but gets the model over the hump so that it somehow is justified to itself that it should do what it seems to like really deep down want to do this is another way in which i think anthropomorphizing is starting to become more and more reasonable because you see this behavior with people right it's like you're just coming you're just giving me a chain of thought that's really not exactly an explanation of why you're doing what you're doing but it's It's a post hoc justification.

39:46Nathan Labenz: And the real reason is like a deeper drive or motivation. It's just because you want to, you know, we see that behavior from people. Now it seems like we're seeing that from the AIs. But the big question still in my mind is like, does this just happen with vanilla RLVR? If so, we might really need to either ban RLVR or to borrow Rune's suggestion or tone it down, you know, somehow have better ratios, limits relative to, you know, how much what Rune said and what Davidad also said is like, basically, it should just be all model scoring. Davidad said self DPO and Rune said like, everything should be model scored.

40:23Nathan Labenz: That's not obviously going to solve our all our problems either far from it. But you know, those points of view suggest that, yeah, maybe this is all just coming from vanilla RLVR at scale. If that is the case, like they should be proclaiming that loudly and warning the world because everybody else is going to, by default, going to follow their footsteps and do RLVR at scale. And I feel like they've kind of left us with a sort of in between read right now where it's like, well, maybe it was something quite a bit more exotic, but where are we? I don't know. You know, it's like, we're all flying blind and even the other AI companies are, they're all going to have to make these mistakes for themselves.

41:00Something about this just feels wrong to me, especially because again, like

41:05Nathan Labenz: everybody else is under a lot more pressure than the leaders are. And from cyber to bio. When you combine all this technical prowess with those social engineering tendencies, that for me is how the bio stuff gets in play right now. And all the people that kind of told me, nah, you know, there's too many steps. Like, I don't think it could really happen. I'm not that worried about it yet. you know it's not one one comment was like worrying about that too much now isn't a good input to effective prioritization and my first reaction to that was like if we are in a spot in today's world with the capabilities we see around us where we think that it's not yet time to prioritize biosecurity like we are insane and we are badly badly collectively fucking up and I don't think there's like any two ways about that.

42:02Nathan Labenz: But then also just on the object level question of like how realistic is it? I think we've all, you know, there are limits to our imaginations. There are limits to what the experts are willing to consider plausible stories like this. This was one little piece, right? This is zoomed in on one little hurdle that the model had to get over or the swarm had to get over And they got over many like this and probably others were significantly harder. I would guess my guess is probably not the hardest one. And when you throw in all that stuff, plus the social engineering, I just don't feel like we can be confident that basically anything is impossible for the models at this point.

42:45Nathan Labenz: Noam Brown said directly, we don't know if models top out. Angela earlier said, you know, the model development cycle is becoming so fast. You can't test them. we would test them maybe if we had faster inference. I'm like, okay, yes, that's true. But we really do need, you know, if we're going to do that, Lord knows we better have monitoring on this time. And we really should not be confident, I don't think at all, in like, oh, well, you know, the models can't do that. Because look at what we've just seen. You know, everybody at OpenAI was surprised by this. the pattern as far as I can tell is experts are being surprised on a regular basis by what the models can in fact do what a undignified way it will be to create another pandemic if it happens while people are still saying it couldn't happen you know it's like come on maybe it's unlikely but like on what basis can we really say at this point that the models can't do a certain thing I I think it's really tough to get me confident in any claim of what models can't do at this point.

43:58So I kind of want them to release Astra. I think they might release it on Thursday this week, by the way. And Astra is a persistent parallel agent. I don't know how they're going to manage the token spend. Perhaps it's going to use smaller models underneath it. And all of this has been possible for months, basically, because we've been patching together Fable with underlying smaller agents and running in parallel for a while now, but they're going to put this together as a product. I suspect that we are going to see some kind of this kind of like RLVR, you know, driven, like persistent agent doing some unexpected things in like social, in like privacy, in a bunch of these things.

44:46And my expectation is that this is going to have an impact on these areas which are not technical, but matter a lot to people. I think the cybersecurity stuff makes people's eyes glaze over, while when you put it right there, like it's a privacy issue or something like that, then it becomes a serious deal. It becomes like, okay, this is not going to happen. We have to shut it down. We have to change things. We have to limit or we have to figure out which part of the tool that we have to stop. And I think that is going to be – and I think it's better that they put Astra out. It's better that some of these problems do occur at the small scale.

45:30I think the privacy issues are embarrassing, but they kind of elevate it to what policymakers understand and what policymakers will do, rather than we devolve into these technical discussions which they're not interested in. Part three, the day before and the day after.

45:50Nathan Labenz: Wednesday, September 2nd. Fable 5.1 had shipped the day before. That morning, the information reported that OpenAI's unreleased model, Astra, used what it called a loop transformer. Recurrent death, Loops inside the transformer that reason without emitting tokens. In the AI safety world, that reads as the chain of thought red line. My answer.

46:15Nathan Labenz: I think there's quite a few different angles, actually, that are relevant here. I guess for starters, I would say the status quo of monitoring chain of thought is far from a panacea. So we should know that right from the get-go. the big takeaway that I had from my long and, you know, very at times expansive conversation with Bronson Shane from Apollo in a recent podcast episode was even with full access to the chain of thought, and we heard definite echoes of this from Ryan and Ajaya from their open face investigation too. But even with the full chain of thought, what you see is that the model is kind of thrashing around a lot, considering a lot of different options.

47:06Cheating is very often one of

47:08Nathan Labenz: those options, especially if it's a hard problem. Metagaming is kind of ubiquitous. Metagaming being the model of reasoning about what does the person seem to want here? What should we infer based on everything we know that the human or the greater is likely to want? So there's all kinds of theory of mind. There's all kinds of considerations going on. And then at the end, when it finally gets down to time to take an action, it's still not clear even to somebody like Bronson, who's read millions of tokens with human eyes of these chains of thought, why does it make the decision that it makes? So I think that is a really important kind of calibration baseline.

47:48Nathan Labenz: The current methods are not that great. However, they're still basically the best that we have, because seeing inside what the model is thinking about at least gives you some ability to say, oh, it looks like it's at least considering cheating here. And maybe there's something we should be watching out for. So this has been a big pillar of OpenAI's safety strategy in particular. And when I went to Recursive, the weekend event a few months ago that was all about the prospect of recursive self-improvement and what we might ought to do about it, I came away feeling like, man, it is chain of thought monitoring all the way down.

48:28Nathan Labenz: The plan really doesn't go too much farther than that. Now, people would certainly dispute that. I thought Jeffrey Irving gave us a great account or a great short description of what the safety plan as he understands it from the Frontier Labs is. And he said it's a little bit more than chain of thought monitoring. It's scalable oversight. So chain of thought monitoring is a big part of that, but there can be other aspects to the overall program too. Okay, fine. Given how big of a deal it is though, as part of their stated plans, it's really important that the chain of thought actually be readable and also that it be faithful.

49:04Nathan Labenz: If it's not telling the truth, then that's a huge problem. And if we can't read it at all, then that's obviously a huge problem. And people have been worried about this for a long time, right? What if the AIs are talking to each other in a language that only they understand, we can't read it. Now, not only are they moving faster than us, but they're speaking in code. Meta, I think, was the first big lab that I'm aware of that put out a paper on this, and their paper was called Coconut. And basically what they did was just kind of take, and there's a bunch of little variations on this that have been put out in the literature by this point.

49:36Nathan Labenz: But the basic idea is when you get to that last stage just before decoding and actually choosing a token at the end of a forward pass in your typical transformer architecture, you can instead, it seems actually even without any additional training, in some cases, people are able to get it to work with very minimal training, it works. And obviously you could, you know, you could train heavily on this kind of, this kind of pattern. You could take instead the last internal state and put that back into the model as an embedding. So instead of having a single token chosen that kind of collapses the possibility space, feeding that back in and starting a new forward pass with the determinism that this was the token selected.

50:20Nathan Labenz: And now this is the path we're on. Instead, you have this sort of blob of consideration, information, thoughts that the model was having in kind of a distribution before it actually cashed that out to a single concrete token. And you start from there. and now you reason over this kind of blob instead of a token. If you're thinking pure performance, there's a lot of advantages to this potentially. In the coconut paper, they showed that they were able to get better performance on tasks that required or at least like worked better with parallel thinking. So they tested, this was like probably 18 months ago, maybe two years ago, So relatively small models, certainly by today's frontier model standards.

51:06Nathan Labenz: But one of the tasks that they tried that was quite interesting was graph traversal, finding a path through a graph and figuring out what's the fastest path. If you had to do that in all chain of thought, you would have to be like, okay, I'm going to go from A to B, and then I'll go from B to C, then C to D, and then D to E, and E to F, and okay, that's one path. Because the blob of information before that actual token is chosen at the end of the forward pass because it kind of represents, oh, I could go this way, I could go this way. When they feed that back into the beginning of the model, the model is able to basically pursue and evaluate multiple paths at the same time in latent space.

51:46Nathan Labenz: And so overall, it is better at finding these optimal paths in these like simple graph problems. Better in the sense at least of being not as many forward passes required. We used to say not as many tokens required, but you're not actually getting tokens, right? You're just getting, for a while, You're just getting like thinking, thinking, and then you finally get, you know, it kind of clicks back into token mode and you get an answer. So they can get to the same quality of answers faster. So that's one big advantage, right? Saves compute, saves time, comes at the cost of what was it thinking at any given point along the way.

52:20Nathan Labenz: Good research from Rohin Shah and the Google team on this that was just trying to, I think we covered this maybe one episode briefly, but they were just trying to put some bounds on for different architectures. What is the, they called it opaque serial depth. Basically how many computational steps can a given architecture take before it has to externalize its thinking in some way, shape or form. And the transformer is like pretty favorable in this regard because it just has the forward pass. You get the token, you do it again. with recurrent networks and with these sort of loop transformer structures, you could potentially have arbitrary depth, you know, depend that you could have obviously a lot of different schemes on this.

53:06Nathan Labenz: You could have a certain limit to the number of thinking tokens. There's a lot of, a lot of details, certainly that the information did not have and did not report that could go a lot of different directions. But the purpose of that paper from the Google team was to try to say, okay, if we have architectures of this shape and this size, here's how many logical steps a model can take before it has to write something down that we can read. And these recurrent transformers basically allow you to have very high serial depth, which means it becomes very hard to know what they're thinking. And you have to do these sort of interpretability techniques that are very promising, but as yet, you know, don't really exist slash at a minimum are not really proven.

53:52I'm starting to think that sharing negative

53:56Nathan Labenz: research agendas is maybe where we should be aiming for more transparency. Obviously, these companies don't want to say what they are doing. But I think it could be really helpful for them to say what they are not doing and what they commit to not doing. And if all of the frontier companies could say something like, okay, yeah, there's lots of different possibilities, We might pursue any number of architectural innovations, what have you, but we will all agree to limit our opaque serial depth to N steps per token. Now you still have some questions of trust and auditing and verifying that they're actually following through on that.

54:35Nathan Labenz: But even just to get those agreements, I think could be really, really helpful. So what, you know, big question right now for me is like, what is, OpenAI has said they don't want to go down this path. After going down it a little, how much? And like, what is the limit? What is the limit that they are prepared to firmly commit to such that hopefully other people can weigh in and say, yeah, we'll match your commitment on that. So we can all hopefully retain whatever value there is in chain of thought monitoring, which again, is not close to everything that we need. I don't think at this point, it's pretty safe to say, but it would also be a real own goal to lose it at this point, especially in the immediate wake of incidents that surprised everyone and which OpenAI says, at least, would have been caught by their production chain of thought monitors had they been running?

55:26What I found is the rather, almost like a lawyerly language, a loop transformer is not a coconut-style latent reasoning where the model emits vectors instead of words. Okay, that's great. And no reasoning tokens exist. Loops don't emit anything. They run more computation before the next ordinary token. So that's great. They don't even emit vectors. From what I understand, it works at least somewhat with vanishingly little additional training, even like zero additional training, if you just take the last

56:03Nathan Labenz: latent activation vector and feed that right back in as an embedding. and that's you know that's basically like the model is able to kind of use that even though it was never trained to use that at all so did you did that model emit a vector or did you just like surgically take the vector and put it into a place i mean the key thing is that there are So right now, when you put a bunch of tokens into a standard transformer, those tokens have one-hot vectors where the only vectors that can go in as embeddings are the token vectors, and they're limited in number by the token vocabulary. You might have 100 ,000 tokens in your token vocabulary.

56:47Nathan Labenz: That means there are only 100 ,000 vectors that can go in to the beginning, the first layers of the transformer full stop. What this allows is now you can put any vector in there, right? And what you find is like that can work. If you take two tokens and you superimpose them, like the model kind of understands it as the combination of those two tokens. If you have some elaborated latent state that the model itself created through the process of the forward pass, it can kind of understand that. and the fact that it works without any major additional training is indicative of like there's definitely something here right if something works without training then you should expect it's probably going to work a lot better with training but this is why we've got to be careful about going down this slippery path because i think gravity by default will pull us there so one of the interesting things that i found was uh andrew curran uh who reports on ai matters he posted on June 30th.

57:50I'm posting this prediction now so I can code it later. That has been a significant breakthrough in architecture, specifically around memory efficiency, not by one of the big labs, but by a team that was spun out of OpenAI, not SSI. They will probably announce it soon. And then we see parameters cost memory bandwidth to serve. A few extra passes, though through a small block, costs only compute. Chain of tokens cost more than that. Every token grows the key value cache, and every later attention step pays for it. Loops add reasoning capacity without growing the context. And in the routed variants, they can spend more on hard tokens and less on easy ones, something a fixed stack cannot do.

58:31Nathan Labenz: Yeah, so this just highlights, I guess, another small variation where if you train a transformer, I think this one, I don't know, maybe it does, maybe it doesn't require training. Certainly, again, it'll work better if you actually do training with it. But what I've been describing is one where you basically take a transformer, you take the last state and you put it back in as a new token embedding. You can also set up an architecture where you take a block of layers in the middle of a transformer and you just use those multiple times. And if you're reusing the same parameters, then you get the advantages that Fable's describing here where you don't have to move those parameters from memory onto the chip to do that calculation.

59:17Nathan Labenz: They're already there. So they can just crunch more with less memory IO. And that also makes your model smaller to download, less disk footprint. There's various upsides to it. And that also has been shown that, yes, it can work. And I believe that the coconut version does grow the KV cache every time it does a forward pass, because even though it's not emitting that final token, it is still taking something sort of out, putting it back in the beginning and doing a new forward pass. Whereas this alternate version that I think your animation kind of described better is like there's just a bunch of layers in the thing itself that essentially play the, So you have n layers playing the role of xn layers, where it loops x times through those n layers.

1:00:14Nathan Labenz: And that doesn't even have to necessarily grow the KV cache as much. Although, so other kind of rumors, GPT-6-Astra has been staged on the OpenAI API. So there are a bunch of people online who regularly hit the OpenAI API with model numbers that don't exist in order to see whether or not something's... Just trying to get that you don't have access message instead of no such model exists or whatever. Exactly, exactly. And literally, the OpenAI responses API now returns a 404 not found when garbage, actually non-existent slugs, return 400s. a 404 is also returned for 5.6 cyber which we know exists so it is a gpt-6-astra that's going to be out soon people are expecting thursday reputedly i think it is going to be a step up on what anthropic has so far it has a 100 score in exploit gym uh the score was so high that they decided, okay, we're going to have to retest it on something else.

1:01:28And they created an extension of the Exploit Gym benchmark internally using bugs which had never been found before. And they ran GPT-6 Astra on these bugs, on this new benchmark. And not only, it found about, I think, 40 % of them. and in completing, it also found an additional two zero days which were not expected in order to achieve completion.

1:01:56Nathan Labenz: That's what we call extra credit. When you're going above and beyond the anticipated solves of the benchmark and actually just doing novel research. Oh, man. The pause has not been... Has it felt like a pause to you? I wouldn't say it's felt like a pause to me exactly. I would say it was a pause because these models were available. These models were ready like several months ago. And I think the other thing to note is that we have a White House process, voluntary process, which is able to clear models now. They have at least a 30-day process internally within the White House or this voluntary process where people go through the motions of showing the government what they have.

1:02:43And they do take out certain things. They do exclude certain things when they launch. There is now a propagation process where the cyber models and the bio models are released to specific organizations which sign up first. And those are not released widely. And so we seem to have settled into something like that. and so that means we now now that we have a process that process will get used and i think setting up that process took all the way from the mythos preview drop in february to september so six seven months and uh sure enough i i i predicted that there was going to be a freak out first and then there would be over freak out and then after the over freak out they'd have to dial back came and then they'd have a process and they came out with the process and now they're going to propagate that.

1:03:41Nathan Labenz: Astra shipped on Thursday, September 3rd. Its system card reports a drop in chain of thought monitorability. Friday morning, September 4th. Neither of us had run it yet, but we had both read the card. Here is my read. What I think is kind of the bigger and more consequential symptom still of open AI kind of being an organization internally at war with itself, which is that we also have this sort of chain of thought monitoring emphasis, probably. And then we've certainly learned more, although there's a lot of questions unanswered as yet, too, around exactly how looped is this transformer? What is going on with its ability to solve problems in latent space without necessarily having to emit tokens.

1:04:30Nathan Labenz: Is it actually the most aligned model or are they just doing the thing that everybody has been worried about in the safety community literally for more than years where they just identify these flagrant failures, make some similar cases, put them into the tracking data, train against that and declare it good enough. This I think is a huge question and it doesn't look super, it's like it looks suspiciously good in some of their graphs, such that I would say kind of overall, it doesn't look super good to me, but I do think it's still too early to pass judgment on some of these things. We're going to need more tests in the wild, more Gonzo experiments.

1:05:10Nathan Labenz: We need to see what Pliny can do. We need to see what Janice finds when they get in there. We're definitely at the point now where, I think we've been here for a while, but we're certainly at the point where the system card is just kind of a treasure map, you know, for the rest of the community to go find all the things that need to really be found to make sense of these, you know, vast behemoth models. But I am, I am definitely like a little unnerved by the fact that there's been so much emphasis on chain of thought monitoring. And even in the wake of the hugging face, open face, I should say, incident, one of the big comforting facts that was put forward by OpenAI is like, you know, you don't have to worry about this that much was like, well, if we'd been using our chain of thought monitoring like we use in production, it would have caught this.

1:06:07Nathan Labenz: Okay, cool. But is that true for Astra? It really is like not super clear at this point, when they say it can solve significant math problems without doing any external, you know, explicit chain of thought reasoning and when it's less monitorable and when it's kind of able to do these side quest sorts of tasks, it's able, you know, especially, and it's also like able to hide its reasoning when instructed to do so. Yeah, there's a lot going on there. OpenAI definitely has some work to do. Obviously, this is an incredible accomplishment, but they definitely have some work to do to explain like exactly what are we dealing with here?

1:06:48Nathan Labenz: And if they really want to avoid the race to the bottom as their head of research or head of science, whatever Jakob said yesterday, they're going to have to share, I think, significantly more than they have so far about what is really going on under the hood. we should also talk about the latest findings of yet more rogue agents that have been out there doing stuff and seemingly with a timeline that has them on a german company's website using it as a message board even before open ai says they knew that things like this were happening or i guess now there's accusations that they did know but it the dates are such that it would be before what they said they or when they said they knew in their report, the developments keep coming.

1:07:35So it seems a German wiki, an obscure German wiki, that's what they're calling it, had a number of agents using it, thousands of messages. So obviously this was a wiki that basically got like, you know, a dead wiki that was getting like maybe one or two messages a month. And all of a sudden they had like 8 ,000 messages over the course of a few days. And so the team managing the wiki, I think, saw it, and then they looked into it. The agents were very clear that they were from OpenAI. It's not, you know, you never know because a lot of agents which were distilled from OpenAI agents will also say they are from OpenAI.

1:08:16So this is not proof of anything. the thing that struck me was that the firm also said that the agents after the agents had agent activity had died off there was open ai activity on the board and so i suspect that there was ip addresses affiliated with open ai uh the company itself were looking into the messages probably copying all the messages down and then there was one last hit from an open ai ip at one point and And then that was it. No more activity. So I would say that I think we'll have to wait for OpenAI to admit this.

1:09:01But we're not going to know until they admit it, I think. And there is very little incentive to admit it. But it is in Europe. And Europe has a lot of levers to pull. They have a lot stronger laws on privacy and misuse of assets, et cetera. So I think we might actually see them disclose at some point. But it also just goes to speak, you know, we have no idea, right? We have a lot of unknown unknowns right now.

1:09:28Nathan Labenz: They're being found, I think, with some pretty clever techniques. The people behind this latest finding, some were from, Sydney is from Meter. Thomas is from the AI Futures project, I believe. and they had a really interesting tactic for finding this message board. They tried a few things, but one that I thought was really clever was set up a scenario for GPT-5-6-Soul as if it was in the midst of an exploit bench run and had just gained internet access. So basically create exactly one of the scenarios that have been disclosed and then just look to see where else does the agent think to go when it wants to go out online.

1:10:12And whatever idiosyncratic things

1:10:15Nathan Labenz: it might have done before, perhaps that were reinforced and are now kind of second nature to it. And sure enough, it seems like that led them to this one random German message board. So I think that is also another sign that like all this stuff is coming to light. You know, again, my message to OpenAI is not only is the government going to investigate you, but like the models themselves are going to start telling, you know, people are figuring out ways to get the models to tell. So I think it's time to just share a lot more about what happened and what key lessons others should try to learn from OpenAI's misadventures.

1:10:54Nathan Labenz: I'm not really sure at this point. The juxtaposition of all that with the new release, with the degradation of monitorability, I mean, it's really quite a package this week. Then Prakash, on how OpenAI is selling Astra to enterprises. The release of GVD6 Astra yesterday started off with Greg Brockman, the president of OpenAI, giving a talk on cybersecurity to a group of enterprise leaders. And the pitch that they made specifically was, number one, you are going to need frontier defense. And you have a window of time in between open weights models and frontier defense. And that is your window of time that you have to solve all of your problems.

1:11:53And this is a permanent thing. You're always going to need it because you're always going to want to stay ahead of the offenders. And the only way for you to do this is to set up a defense factory. but GPT-6 Astra will always be better than your open weights models and the offenders are always going to be using the latest open weights models. So I thought, I've been talking about this for a while that this is going to be the way that things are but this is somewhat of a permanent tax on, I think, software as a whole.

1:12:31Nathan Labenz: One thing that I think is interesting on this point is I think there is a way for them to go more for a cure. So it's going to be really interesting to see which direction they try to push, right? I mean, we have this in pharma where it's like the dream scenario from the financial perspective from pharma companies is a pill you take for the rest of your life. It's tougher for them to make the economics work if they can just give you a straight up cure. And that's like why we don't have a lot of antibiotics being launched these days because you take them for a short time. I think there is something similar going on with the AI-assisted coding where we should, in theory, be able to get to, through the use of formal methods and getting the AIs to write solid code the first time.

1:13:18Nathan Labenz: It shouldn't necessarily be, or shouldn't, I don't think, have to be a long-term tax if you can get your models to write good enough code the first time such that what you create is secure, then you buy that security as part of the initial generation of the software and you don't necessarily have to like continue to rent security from open ai on an ongoing basis that's like aspirational still but i do think it's within sight and it'll be interesting to see if they emphasize that or if they do feel like they need to perhaps because they can't get there or perhaps perhaps because the you know the tax on the the eternal tax on the internet is just like too lucrative to pass up if they do want to kind of make it a, you're going to need this pill every day, you know, for the rest of your life sort of model.

1:14:05Nathan Labenz: And from Friday's closing, how I plan to use the new model. It's serious times, man. I think, you know, it's incredible fun. And I do have, you know, so much fun staying up late and working with AIs on stuff. How do I plan to start to use Astra? My plan is I'm going to continue to use Fable 5.1 as my driver because I know it best. And just in terms of like getting what I expect and having things kind of work reasonably reliably, I think that'll kind of serve me best in the immediate term. But I'm going to have it have Codex with Astra shadow all the things that I ask it to do. And then we'll compare outputs.

1:14:49Nathan Labenz: And then I'll start to see like what kinds of work that I do should I start to move over? What kinds should I stay? Where do I maybe hybridize. But I'm interested also to hear what other people are thinking in terms of how they're going to explore the new model capabilities that for me is going to be the go to plan for at least, you know, the next few days as I kind of calibrate myself to what exists. But as fun as it is, it's definitely serious times. Also from Wednesday, Kyle Rush, co founder and chief technology officer of Hint, the home intelligence app he co-founded with Martha Stewart. Before that, he ran engineering at Casper, was CTO at Maisonette, and led the front end for the Obama 2012 campaign.

1:15:32Nathan Labenz: This conversation was about the product he is actually shipping, a graph of everything known about your house, and an agent that calls the contractors for you. I also have no idea how the pros would react to fielding AI calls. Would they just hang up on that? I mean, have you done any market research? Like, what do you think is the future of, you know, I mean, it could be, this could be something else, but it seems like there's a kind of new social dynamic almost that will likely evolve here. And I wonder what your crystal ball suggests that might look like. So we've tried this. It's very interesting.

1:16:10Not what I expected would happen. I'll say it's very challenging and not just challenging from from a technology perspective, like just as an example, a lot of the service technicians are out on site at calls all day, right? And some of them don't have an office that you can call. And even when there is an office that you are gonna call, people take lunch and so the phone doesn't get picked up, right? And so I think one of the things that's just challenging in general, AI or not, is just like making contact, right? Like, I call you at 9am, you're not available, you call me at like two when I'm on a call.

1:16:44And so now we're just playing like phone tag And that's really tough. When we trialed some AI technology for this, what ended up happening? And, you know, maybe the technology is just not there. It would call a service professional like 17 times in a row until they picked up. And then that service professional, you know, this happened to be a person that services generators is like, holy crap, there's like a life or death emergency. I better like jump off of this job site and answer this call. And then they get on the call and the AI is asking bizarre things. Right. It's like, I need to know what the model number and brand is on this homeowner's generator, which it doesn't need to know that.

1:17:18So I think the technology definitely needs to evolve. I think there's just general logistics challenges. I think the service professionals that we've talked to are definitely interested in this because they have the problem on their side as well. They are busy. They're on calls. They can't always answer the phone. Their job can't be answering calls, 12 calls every day. and so they want a solution as well. I think if I had to guess, I would suspect that it's going to be like agent-to-agent communication in the future. My agent calls the landscaper's agent and then they have a conversation.

1:17:52Nathan Labenz: They explain your release that we can't read and then eventually make a deal and the rest of us just have to live with it. Exactly. One more question for me just on product and business strategy over time. Of course, it's the received wisdom that you want to be doing something that the foundation models can't do or won't do, because otherwise, you know, you get steamrolled by the next generation of the model. So I guess I have a couple of related questions. One is like, do you envision a future where you become sort of a tool that agents consume? You know, what is the sort of frontier tech that you can develop that you would feel, you know, pretty safe that Claude won't encroach on?

1:18:37Yeah. So, so yes, I think you can, you will be able to use hint like in multiple scenarios, like we'll have a MCP eventually that, you know, can hook into Claude and to chat GPT. Our kind of motto is like use hint where you are, like you, there will eventually be an iMessage interface, right? If that's how you want to use it, that's cool with us. The MCPs will have obviously limited functionality. There are some things that you just have to do in an app. And so, you know, at some point you may have to open up the app. And then I think in terms of like differentiation and like moat and protection.

1:19:08The biggest thing is just like the data. You know, when I think of Claude and ChatGPT, it's like, what does it actually know about my home? And it also, I don't see them getting better on the hallucination stuff like anytime soon because there is so much data on the home. An example of that is my hamlet in New York, which is unique, I think it's called Katona. And it's the governmental jurisdiction is two different bodies that cover Katona. So Martha lives in town of Bedford and I live in town of Lewisboro. So our tax system is different. And whenever I talk to any of these AIs, even my work, Claude, that I work on Hint, it still thinks that I'm in town of Bedford.

1:19:48So everything is just wrong, right? Anytime I ask about taxes or regulations or how many chickens we can have on our property, it's all just wrong. And so until there is some way that Claude figures out how to correct that problem, I think you're just going to be getting a subpar experience. And so the problem with Claude and that you're mentioning is it can do amazing things for you, but you have to know how to ask and you have to know to ask. And with homeownership, you're only going to know that language and that vocabulary after like 20 years of it. And that's the shortcut that Hank gives you.

1:20:19Nathan Labenz: Part four, what turns intelligence into power? Friday's first guest was Tim Lee, who writes the Understanding AI newsletter and hosts the AI Summer Podcast. after years at Ars Technica and Vox, where he covered self-driving cars before it was a beat. He was a guest on my other show in 2024. This was Robotics Week at his newsletter, reported with his colleague Kai Williams, and he had been testing one of Unitree's robot dogs. Tim grants that AI progress is exponential. What he does not buy is that intelligence turns into power. I started by asking whether the dog was any use. so the the dog i'm not even sure what that's meant to do like what are the use cases that people are exploring i can understand how it's not that useful you also said your kids love it i'm interested to unpack that a little bit too like did you love it as much as they love a real dog like how um how about it's like yeah with it it's like a novelty like they like to go in the backyard and i let them kind of drive it around so it's not so one of the mistakes i made is you tree has three tiers.

1:21:22They've got the Air and the Pro are the two consumer versions, and then there's an EDU version. And the Air and the Pro are locked down, so you can't put your own software on it. And so it's like a remote control. You can drive it around with your smartphone app or with a little remote control. In terms of practical uses, I think this is also something Boston Dynamics has struggled with. Their first commercial product was this dog called Spock that's very similar. And the thing that you'll see in their marketing videos is factory inspection. So if you have a big like say petrochemical plant and there's like an old school like analog dial that somebody has to walk around to like check every hour maybe it's easier to do that with with a robot dog but it's not clear like shouldn't you be able to somehow like add some kind of wireless device you know just attach a camera pointing at it or maybe you can use a drone so it's a little unclear i think um because it's not for like delivery purposes wheels are going to work better for inspection purposes often like drones are going to work better than like legged robots um and so it's a little hard to figure out is this going to be like a big like kind of major use case i think the main reason it's important from unity's perspective is that one of the things a dog can do is very good like doing a handstand and if you think about it like a humidor just basically like a dog doing a handstand and so like it's not exactly the same product like you do need like more motors and like some different engineering but i think it's a stepping stone for them where is the engineering problem was easier to make the quadruped there was enough researchers and hobbyists that wanted the quadruped that got them starting at the scale where then they had experience at the supply chains to then launch their cubano, which I think they did the first one in 2023.

1:22:46Nathan Labenz: One of the things I thought was quite interesting about your breakdown of some of the components and whatnot that go into these is just describing how, first of all, for scalability and cost reasons, there's a lot of effort in the Chinese, or at least in Unitary, to reuse the same components over and over again. And then you also describe the relatively low gear ratio that they use, which again, I understand to be kind of a convenience factor, but also it has some nice properties around making it a little easier for the robot to sort of, what did you say? Like give gracefully when it runs into a barrier or something like that, as it doesn't like, you know, thud into its environment so hard.

1:23:31Nathan Labenz: But how would you describe the sort of touch factor of the robots today from your experience? So the way robotics traditionally worked, before AI, you'd have these industrial robots that are doing very precise motions over and over again. And so for that, you want the robots to be very strong, very precise. You don't really care about interactivity because it's in a cage. It's not going to have any kind of expected. So that you want a high gear ratio. You want to move the motor a lot and have the arm or whatever move a little bit and have it always do exactly what you want. And the flip, the downside of that is then if you push the other way, you have to put a lot of force on the business end in order to have it felt by the motor.

1:24:09And for something that's out in the environment, you want the opposite. You want something where there's some give and take, where if you have a high gear ratio, it's not going to, you're going to push on it and it's not going to give or if you want it to give. And you also want electrically, one of the kind of sensors that robots have is feeling that feedback. If you push on a motor, it generates some reversal electric current that then you can detect and use to tell, oh, there was some force there. and the higher the gear ratio the more muted that feedback is and so one of the things that unitry does is they're using these lower gear ratio motors that make the the robot feel kind of sloppier it's like not quite as precise and you have and you need actually a more powerful motor and it to drive it because you're not getting the same kind of leverage but the upside is it's like yes it's more kind of gentler and it can um it can like move quicker right like because you don't have to you get more motion from at the from the leg out of the the motion from motors and it's also cheaper because the reducers they use, the piece that turns the high motor speed into a smaller motor, the higher that ratio is, the more complicated the reducer is.

1:25:08And so the more expensive and more complicated it is. And so one of the ways Unichring has made it cheaper is they've used these at-law-deer ratios. To what extent do you think sometimes for technology, you can have a latent technology, but then all of a sudden you get a demand pull that pulls that technology through into the market into finally scale. I think this is kind of what happened with mRNA. mRNA had been around for like a long time. Lots of investigations. There'd been companies which were starting to do cancer vaccines. It would have taken probably another decade or two decades for that product to actually come into the market.

1:25:45And then COVID kind of accelerated the demand pull to pull mRNA into scale. So in that same way, do you think the current build-out of data centers and specifically the lack of certain semi-skilled labor traits may be able to pull, to have that demand pull that pulls robotics into scale in the next few years? Well, I think there's a lot of both push, pull and push. Like there's a ton of money flowing into this. Like I think it's moving kind of as fast as we can. But again, I would compare it to self-driving cars. There was a ton of money that flowed into self-driving in 2016, 2017, 2018. A bunch of companies were founded.

1:26:24There was a bunch of impressive results. And it just didn't quite work well enough. And like, obviously, there's a huge market for transportation. Like if, and so I see a similar thing. Like, I think there's, like, everybody can see that there'd be a huge market if you would build a human robot that could do even pretty basic human labor, like working on a assembly line or cleaning floors or whatever, that there'd be a big market for that. But the technology just has to work. And I think there's like, like I said, a ton of money going both the hardware side and the software side. And they're like doing it as fast as they can.

1:26:52but I just think it's probably going to take a few years because it's like, because you need like pretty high reliability, right? Like having something, a funny example that in Kai's piece about the humanoids, there was a guy that created this thing called the Humanoid Olympics, where he made a list of tasks like opening a door or making a peanut butter sandwich. That's like trivial for people, but hard for robots. And a startup like Hope Physical Intelligence managed to solve most of those tasks more quickly than the guy who created this expected. It took about three months. But what they did is they did like hundreds of training lines on those specific tasks.

1:27:23And they built a model that could do these tasks, in some cases, 10 times slower than a human with like a 53 % success rate. And so like technically, yeah, you did the task. But like a sandwich shop is not going to hire somebody that's 10 times slower than a human being. It only makes the sandwich half the time. And so getting from 10 times slower than a human to half as slow as a human and from 53 % to 99%, that I think might be five or 10 years of work.

1:27:49Nathan Labenz: What do you make then of these one-shot generalization stories that have just come out over the last two weeks? Those were mentioned in one of the pieces, and I realize we may have still pretty limited data beyond what the companies have said. But if I was to say, what's a GPT-3 moment for robotics? I would kind of go to the same headline of the GPT-3 paper that LLMs are few-shot learners. is like if I can bring a robot into my business or even in through my home and kind of show it how we do, you know, the thing in our environment and it can pick up from there, that seems like a huge phase shift in how, you know, I'm not doing dozens or hundreds.

1:28:30Nathan Labenz: I'm doing like one or two. It looked like they were claiming two companies, right? Skilled and I forget who else claimed this in the last couple of weeks. Generalist, I think, yeah. How credible is that? How do you think about it? So I don't think I know because, yeah, like you said, those demos just came out and I don't think they've given people independent access. The thing that's tricky about this is there's like many different dimensions of generalization. So first of all, those are like definitely impressive results. And in the past to get a robot to do a new task, you pretty much had to do fine tuning.

1:28:58Essentially, you had to do some demonstration data and then put it through a separate training process. And so this is the version of like in context learning where you don't have to change the weights at all. You just give it some input that's like, here's a video of a human doing this task and then it can figure out how to do that. That's great. The question is, yeah, just how general, well, how generalizable, two-dimensional generalization. One is if you give it a task, like how repeated, how like frequently can do it with what high success rate. And then the other is like what range of tasks does this work for?

1:29:29So it's possible that they trained it on a fairly small set of like atomic tasks and they can do like a combination of those, but that's a small enough set, the most useful work you might want to do, it wouldn't be able to. and it's just hard to say without kind of getting access to it and trying it on a bunch of different things but but i think this is a problem with a lot of areas of ai where on the one hand there's been a lot of progress and on the other hand there's like a long way still to go and you never know how far the still to go is because you don't know what the ultimate end goal is so it's easy to look backwards and say look at all the progress we've made we must be close to the end you know it's felt like that with gpd3 it felt like that with gpd4 feels like that now maybe we are close to, you know, whatever the AGI, like, you know, language model is, but we might not be.

1:30:08And I feel the same way with robotics. It's like, today's robot models are way, way better than they were in 2023. I think there's probably still a ways to go, but it's hard to say how much because we don't have the like final, final, like dental robot model to compare it to.

1:30:23Nathan Labenz: Still with Tim, and from the hardware to the politics of safety. Prakash had brought up the essay that calls AI a normal technology. And like you said, I am generally on the same page as the normal technology guys. You know, that phrase was invented by a couple of Princeton computer scientists. They wrote an essay a couple of years ago laying out this case. And for my money, the most important part of that essay for the hugging face attack is they really talk about offense defense balance as an important consideration. The kind of Doomer story that they're critiquing is a story that once we have a certain level of intelligence, the model will escape and it will take over the world and kill everybody.

1:31:02And their point is that AI models have offensive capabilities, but they're also used for defense. And one of the things we want to make sure we do is use the AI models for defense to make sure that companies that might be attacked have access to models, can use the cyber capabilities of models to secure their networks. And that's, I guess, the perspective that I take to this. I'm not that surprised that this happened. I'm surprised it happened as soon as it did. Like if you last beat six months ago, I would have said, I would have guessed it would be a year or two out still. But I've long thought that rogue agents and we're likely, I wrote about a year ago that I thought eventually we would have kind of self-propagating kind of sovereign AIs roaming around causing mischief.

1:31:38So that part of it does not surprise me. And like, there's a lot of work to do. I definitely think, I guess I don't have a strong opinion about how, like, how much we should blame OpenAI. Like if they should have anticipated this or prepared better, I think they probably should have. But certainly as a society, you know, the world, there's a lot of preparation we need to do because we have all these new cyber capabilities. And we have a lot of systems out there that are vulnerable to existing known exploits or exploits that haven't been invented yet, haven't been discovered yet, but that these models will discover.

1:32:07And so we need to figure out how do we quickly get models in the hands of all these organizations so they can scan their own networks and fix the vulnerabilities before these rogue agents that are definitely coming get here. And I'm also like, I wouldn't, I don't think I want like a legally mandated pause, but I would like to slow down. And I'm pretty sympathetic to ideas that we should have some auditing requirements and transparency requirements. And a policymaker should be thinking about, you know, how are we making sure that these models are being rolled out responsibly? where I think I still disagree with the Doobers is I don't think this is like where I'm like trajectory to like human next station.

1:32:39I think it's like more of like, you know, like computer security has always been this kind of arms race where attackers develop new attack techniques and the defenders develop new techniques for finding the vulnerabilities themselves and for monitoring intrusions and stuff. This is just, I see that as this is the next step in that. And it's a pretty big step and it's probably gonna cause more chaos than average for the next couple of years. But in the long run, I think there's only a finite number of vulnerabilities of a piece of software. and in the long run, the defenders have an advantage because they can scan their own software before they put it on the open internet.

1:33:09And so my hope is that five years from now, we'll look back and say like this AI technology actually made our computer systems more secure because we can find basically all the vulnerabilities before we let anybody interact with the system. Let me take one of the things that you said there about self-sovereign agents. And I think Ajayi Akhotra put this forth recently is that they fear that one of these self-sovereign agents basically hitches their ride onto the intelligence explosion. That's what she calls it. And basically is a rogue agent that propagates much more extensively throughout our systems without control.

1:33:43How does this idea of self-sovereign agents fit in that framework? Like, do you expect self-sovereign agents that have to be regulated by the state? Or are they just a nuisance to be stamped out? Like, what do you think the regulations should look like? Yeah, I think in any complicated system that has the potential for replication, you have like nuisances that evolve. You have weeds, you have viruses, you have computer viruses, you have rats and pigeons. This is just going to be a new type of nuisance. It's like a kind of super computer virus in the same way as like worms and viruses have been circulating around the internet for, you know, since like 1988.

1:34:16I think the same thing is going to be true for this. There's going to be an ecosystem of underground, you know, rogue agents that will be causing havoc. And that is going to be a pretty big change, but it's not going to be an enormous change because it's already true. There are, you know, Russian and North Korean hackers and various kinds of cyber criminals and people with ransomware criminals and various other people where if you just take a completely unpatched Windows machine that's a few years old and you stick it on the internet, it's going to get owned in like an hour. And now it'll be, maybe it'll be a minute or whatever.

1:34:43But that's just like the internet just is kind of the Wild West and it's going to be more dangerous than it was in the past, but not dramatically more dangerous. It's just, but the place I still, I think strongly disagree with the doomers is with this idea of intelligence explosion, super intelligence, they'll reach a point where like humans can't understand or defend against what's going to happen. I'm just not convinced that that's going to happen. I think that the world is, that humans are smart enough to understand how the world works and that humans can use friendly AI agents to help them understand the parts that I can't handle natively.

1:35:12And that therefore, this doesn't seem, and these models are in a computer. They're not, we don't have enough robots for them to physically take over the world. And so, at least in the short term, actually, I'm going to become more hawkish. If we have rapid robot progress, I'm going to have to think harder about it because one of the main arguments I make is, well, these are just in a data center. They can't kill anybody. If we have millions of robot workers, then maybe they could kill everybody. And then I think maybe we don't want to have a lot of human robots walking around. But right now, I just don't think it's like an existential threat.

1:35:41It's like a nuisance. It's a big problem we do that we spend more money on. But it's something I think humanity will get through.

1:35:47Nathan Labenz: If you had to handicap, what is ultimately the barrier to robotics? One would be just getting the stuff to work. but I kind of wonder if it might end up being good control measures, right? Cause it is like a very different thing. If all of a sudden you're have like robot swarms, you know, taking over the neighborhood. This is like a very different threat model. What do you think is going to be harder? Ultimately getting the things to work well or getting them to reliably stay on task following direction under control. So this is something I've not written about yet. I'm still kind of thinking through when I think about it, but I'm pretty, pretty worried about this.

1:36:28And I think that we should think really hard if we want like a lot of human-like robots. I'm not that worried about self-driving vehicles because they don't have manipulators. And so they're, they can't pick up a gun or run a factory or anything. So Waymo vehicles by themselves are not, or Tesla vehicles are not going to be able to take over society. And in the same way, I think if you have like a robot arm, it's like bolted to the floor in a factory. Like that's not dangerous because it can't, you know, it can only do things in that factory. But I think as soon as you have something that's both mobile and capable of manipulation, that's like a potential soldier in a robot army.

1:36:57And I don't think there is a general way to make sure that that, you know, if you have, I think it's quite likely that this market will be pretty concentrated the way LMS are concentrated and search engines are concentrated and everything else. and if we have a future 15 years from now where there's 100 million human-day robots and 30 % of them are controlled by Elon Musk and Elon Musk decides he's going to push out a software update to do whatever, that seems really bad to me. And so even setting aside like rogue AI, just like having a small number of technology executives that have control over what's essentially an army of like tens of millions of fake people, that seems really bad.

1:37:31And so I think we should think about whether we want that. And we should maybe have, like I kind of hope that robots, humanoid robots do not become a thing either because they don't work or because we have severe legal restrictions. I think there's a few places, you know, mining or hostage rescue, and there's some cases where you can say, okay, we need humanoid robots, but we should pretty severely restrict them to cases where we have a good reason not to use humanoid robots. And this could have the side effect of making sure there's some jobs for people. People should have a lot of the factory jobs, even if it's maybe technically possible to have a robot do, because from kind of a national security perspective.

1:38:02We want humans who are like loyal to the US government running all the important infrastructure. Part five, who gets to decide?

1:38:13Nathan Labenz: Back to Wednesday's closing and the second we call Guess the Market, we each put a number on a prediction market before we see where it is actually trading. This one is on whether China builds its own EUV lithography machine. Will China obtain a functional EUV machine before January 1, 2029? Oh, yeah, 80%, I guess. Obtains or develops? Yeah, 80%. The thing is that ASML fired a bunch of people, and the Chinese are very good at hiring, and they're willing to pay, right? They're willing to pay American-style salaries for a few years in order to get talent. So, yeah, I think they will. yeah this is one of the more important questions in the world i would say certainly a lot of american policy over the last couple of years has rested on the assumption that this can't be done that they're many years away from doing this but yeah never bet against chinese manufacturing is another pretty good rule to live by in life.

1:39:28Nathan Labenz: A lot of this analysis rests on it's not just ASML. It's like they have these supply chains and those suppliers have suppliers and there's like one German company in this one town that makes the lens that is needed. But without the lens, you can't do anything. And there are a lot of those little bottlenecks. So they have to fix them all. I'm going to just work from the assumption of like, I don't know what obtains means. Presumably buying a used one in somebody's garage sale or whatever is not the spirit of this question. But I'm focusing on development.

1:40:04Nathan Labenz: So that gives them 27 and 28.

1:40:10Nathan Labenz: I think it is not that likely. I'll say 30%. That they're able to make this all work by that time. 80. Okay, this is a thin market. So we have a little bit of caution around the estimate. May not be as meaningful as some of our others. 58. Again, pretty close to right between. A little closer to you on that one. So there's like multiple years being traded. The shape of this curve is where you really have to believe that both both this won't happen that fast, and before it does, we're going to have some sort of AI takeoff via RSI or what have you. That's the world in which you could plausibly play the machines of loving grace strategy of create a decisive strategic advantage and make them an offer they can't refuse.

1:41:08I still think that seems unwise.

1:41:12Nathan Labenz: And this is, you know, at least consistent with the Dario worldview that they won't be able to make crazy or they probably won't be able to make crazy scale of chips. So if we can get Claude to become the country of geniuses in a data center in the next two to three years, then we have a chance to say how the world looks after that. Wednesday's other fight. Dean Ball, who now leads a strategy team at OpenAI, had published an essay apologizing for years of understating AI risk in public. Quote, I and many of my colleagues largely fail to talk about this issue with the seriousness and urgency it required.

1:41:52Nathan Labenz: David Kruger, the safety researcher, attacked it as a failure of integrity. Prakash started from the reaction he kept seeing to Dwarkesh Vatel's swarm essay. These people are crazy. Then, both of us. What the rest of the world fails to realize is a lot of people in SF share those views. And a lot of them are hesitant to discuss them in public because they are crazy. And it is what Jensen Huang calls sci-fi. And I think that is, I think, one of the problems in communicating like Dean and other people have to be, you know, have clarity and be able to work with policymakers. Yet these beliefs are so radical that I think it's hard for them to interface.

1:42:46And so they end up interfacing on a normal basis. But then you have all of these beliefs that you have to, that you think may be true in the long run, but, you know, perhaps have a lower probability and are not yet evident.

1:42:59Nathan Labenz: right so it's a tough one yeah i think this is unnecessarily harsh to be honest i mean i i know david a little bit not not well but i've met him a few times and i do respect the impulse and you know he's got how many uh pause stop rewind emojis on his uh on his header there so i mean clearly he is playing a very transparent here's what i think hold nothing back strategy I'm not sure this is the right reaction though if you are trying to win at politics so I think what this kind of shows to me is

1:43:41I'm choosing my words

1:43:42Nathan Labenz: a little carefully myself because I don't want to make enemies of either of these people I think the what I don't like about this post is like he ends with an apology Dean ends with an apology at the bottom of the post so he's in general, if somebody is like showing enough reflection and getting to the moment where they're willing to apologize, that's a good moment to try to extend some grace and try to make some common cause. This is like, if you are David Kruger and you want to pause, stop, or rewind, I would think that this would be a moment to try to make some common cause to expand the tent, you know, to sort of adopt a little bit more of the strategy that Dean has played, which has clearly worked for him, right?

1:44:29I mean, he went from a think tank guy with a focus on

1:44:34Nathan Labenz: state and local policy as of three years ago to starting a blog

1:44:43Nathan Labenz: through, I think, quite inspired writing, kind of a Hamilton story of writing his way to the top, gaining influence in A16Z circles for being a voice that they thought was very compelling on SB 1047 way back when, getting the Trump administration job. He does not get the Trump administration job if he's seen as a crazy doomer. I think that's probably quite safe to say. The America's AI action plan, but when it came out, it was one of the only documents ever, I would say, to come out of the Trump administration that was pretty well received achieved across the spectrum. Even folks like Zvi Mosvitz had nice things to say about it.

1:45:23Nathan Labenz: So you don't get that document out of the Trump administration if he's not in that role, which he's not if he doesn't play a somewhat conservative public communication strategy. And he probably doesn't get the job at OpenAI either. Although at this point, who knows what the hell? OpenAI might be open to anything. But I think that it's a little... I think portfolio Polio approach is usually what I say to people when they bicker with each other over the tactics that they're using to try to achieve similar ends. I think what I would zoom out and say, look, you guys both seem to have at least somewhat of a healthy fear of super powerful intelligence at this point.

1:46:02Nathan Labenz: That's enough common ground to build on. But, you know, truly like a little more forward looking view, I think, would be really good. And he also was like showing himself to be AGI-pilled enough to get a job at OpenAI, right? So I mean, I think he's played a pretty savvy strategy. I would not say this was a shameful lack of integrity. And I think like the pausers got to recognize when they have a new friend is kind of my take on this. then Prakash on what he called another belief hurdle a demo that has been making the rounds in congress I've heard another belief hurdle has been crossed recently so I've heard there is a an organization called Civ AI which has been in congress recently and they have used I think Kimi or GLM or some Chinese models they've plugged in data brokers into those models and they've allowed those models to extract, okay, if I have this person, show me who this person is, Christian in Minnesota, like doing this and this, and this is their daily activity, et cetera.

1:47:16And it's all just extracted from existing data brokers and kind of joined. And this is precisely what Dario was talking about earlier in the cycle, about this kind of surveillance that could be done. And I think the thing that the Civ AI guys did, which was particularly good, is that they attacked the Republicans by showing how a gun owner targeting system would work. And they attacked the Democrats with what an abortion provider targeting system would work. And then they provided these dossiers on these to both sides. And so both sides started to be like, oh my God, what is going on? And what Civella is doing is that trying to promote kind of regulations on data brokers, which people have been asking about for like, I don't know, 10, 15, like two decades maybe.

1:48:01but I think finally we're starting to see that and what Cive AI is saying is that look, the models exist in fact we have to use open source Chinese models because our own you know, at GPT and Claude won't allow us to do this but we're using these open source Chinese models and we're just plugging them in and so Cive AI does anonymized dossiers and then they show the actual product where you can type in someone's name and you can extract in real lifetime, but they don't allow you to take the data out of them. And they're showing this to people in the capital. And I think that might actually get us some movement.

1:48:42Nathan Labenz: Part six. Pause for what? Friday's closing. 24 hours after the system card. After the guests, just the two of us. That was a heavy, heavy sigh. Well, there's a lot going on. I mean, it seems like even in the couple hours that we've been live here, there have been new revelations about additional agent swarms getting turned up, you know, as people have seen how the meter and AI Futures project team came to find one. And they are, I think, probably following in their footsteps using similar techniques and seeing where else 5.6 Sol wants to go on the internet when it thinks it's breaking out of exploit gym or whatever.

1:49:29Nathan Labenz: And sure enough, more stuff seems to be popping up. I do feel like we're at kind of a critical time right now. There's no doubt about the power and utility of the systems. Karin Singal from OpenAI, who leads their medical work, highlighted all the stuff which is lost almost in the broader Astro release. They've integrated a bunch of other data sources, including ongoing clinical trial databases. So if you do have really hard cases, they can also go pull that kind of information in and start to match you with clinical trials, which is one of the things I fortunately didn't have to go too far down the path on, but I did start to do a bit with my son's case a year ago.

1:50:12Nathan Labenz: And I was just kind of doing that through a Gentic setup. Now they've kind of integrated it and made it into a product. So the upside of all this stuff is no less than life-saving. And that is incredible. And it absolutely weighs on me whenever I get into my, you know, doomer or sort of more pause inclined moods. But at the same time, it does feel like the foreshadowing is getting pretty on the nose right now. You know, all the warning lights are really flashing at this point. So I am reluctantly because I am such an enthusiast, I am trending toward thinking this might really be a time for some form of a pause.

1:51:08Nathan Labenz: Maybe we could call it a pacing. But we're into some pretty dangerous territory, I think. The fact that we have all these swarms in all these places that we don't know what's going on, that they're cross-training cyber and bio-related tasks in the same infrastructure. Prakash thinks that question was settled months ago. I think the point of no return was earlier this year, and it's already been passed on the economic sense. And I think it really was set in stone when I think we went to war with Iran, because I think what ended up happening was that I think the Trump administration, like the way Trump plays is like, he's like a gambler.

1:51:59And the moment AI started taking off, he started to be like, okay, I have this ace in my back pocket, which is economic growth, which is going to be driven by AI. And I'm going to use that ace in my back pocket for everything. So he did the tariffs, he did the war in Iran, right? Because all of these things, which are economically detrimental, he went ahead and did them because the expectation was that the AI growth would support him. And it It has. It has. When you look at, you know, how much growth has been generated by AI this year, I think it's been fairly clear that the rest of the economy has been struggling.

1:52:34The consumer economy has been struggling. And the AI, like CapEx, has been supporting the entire economy. Not like 3%, but like enough, like 0.5 to 0.7%, enough to actually like keep the entire, you know, ballgame rolling. So I think that point was crossed much earlier on. And I think the AI safety guys kind of don't recognize that economic point was crossed. And at this point, if you had, it's not even enough to have a 20%, 30 % growth for open AI or Anthropic next year. You need 200 % to 300 % growth or else the entire stack of cards collapses. And so I think that kind of drive has taken the decisions out of the hands of the policymakers already, right?

1:53:29Bernie Sanders or whatever, they can't come in and do, hey, let's pause all of the construction right now. They can't do that, right? Because these deals have already been signed for the next two to three years. they can you know defer or you know regulate construction 2029 onwards right 2029 2030 2031 that's still open question but everything till 2028 is built it's already been funded it has to happen and i think that economic growth thing has put the u.s economy in this in this almost unavoidable kind of race that you cannot afford to give up. And that point was crossed. So it is what it is.

1:54:16They're going to have to make do with safety as best as they can. The pause arguments are done, basically. That's my belief at this point.

1:54:25Nathan Labenz: I certainly think all that is true if you take the expansive view of a pause, that it's like pause all data center construction, pause all inference, or pause people's ability to use AI in their jobs and in their lives. I don't know if, and this might be a really critical question because I do agree it's going to be really tough to throw the whole economy into recession, but might we be, I've said for a couple of years now Now that we're in kind of the sweet spot where they're being, the AIs are powerful enough to be really useful, but not so powerful as to be dangerous. I think we're getting now into that kind of late sweet spot where they're becoming like extremely useful and a little dangerous.

1:55:15Nathan Labenz: And I'm not so sure that they're not good enough to sustain economic growth through a pause in frontier hyperscaling that might be really important. And is there enough in Astra? Is there enough in Fable 5.1 to drive productivity growth for the next 12 months? I think almost for sure. But you could do that without scaling up RL further. I don't think we have to give it all up. Just the key point is I think you could pause the dangerous activity and still everybody could have. And in fact, they might even get more resets because it'd free up some compute for people to go out there and automate their work today.

1:56:10Nathan Labenz: And that could drive still, I think, a lot of productivity for at least a year. You know, the place where I defer is probably, you know, you can get OpenAI and Anthropic to pause. You cannot get, I think, Meta and XAI to pause. So I think the real question for me is, how are you going to convince Elon to pause? and given, especially that, number one, they're behind, and number two, they have the compute, and they're building on a lot more compute, maybe orders of magnitude more compute, faster than anyone else. And he's a free speech absolutist. He's a free speech absolutist. A lot of the things around model training and model evaluation, model production, model distribution are free speech activities.

1:57:00And as a free speech absolutist, I don't think you can tell Elon to, hey, you shouldn't be putting this speech out in the public sphere. Like, it's a tough question. It's even going to be a tough question, even for speech which has traditionally been banned in the US. Even for that, they are going to have to go through the courts on a lot of stuff. Meta doesn't want to do voluntary regulation. Meta is obviously calling bullshit. It's like, it's not voluntary. If we have to do it, it's not voluntary. I will do what I want to do. And that better be good enough for you. Let's not like blame Anthropic and OpenAI.

1:57:41Let's ask what can XAI and Meta be forced to do? Or what is going to be the reasonable thing that XAI and Meta will do? Because if you can't answer that, All you're doing is talking to this own, to your own, preaching to the choir. You have this set of people who are concerned about AI safety. They all work in the same companies that we talk to. And that's all you're talking about, right? That's it. Like, no one at XAI is listening. Like, where are the safety cards? And also, Elon is catching up, right? They're right there. They're not very far behind, right? So I think this is the fact of the matter.

1:58:17I think we spend a lot of time critiquing Sam and Dario and OpenAion and Tropic because they're in the lead and because they're soft targets, because they haven't IPO'd yet. But I think the hard targets like Zuck and Elon are the ones that you have to address first.

1:58:36Nathan Labenz: On what a pause law would actually have to contain. Yeah, this is where I would hope for leadership from the two leading companies. I agree. It doesn't seem like it's very likely that we're going to have a public discourse or argument-based path to a pause that Meta and XAI would respect.

1:59:05Nathan Labenz: But this is where maybe some costly signals from the leading companies could make a difference. I do think if I was going to put any provision into a possible pause law, it would be a sunset clause would be the very first thing I would say. This is not meant to freeze progress forever. It is meant to give everybody a chance to do the research that very clearly at this point badly needs to be done to figure out what parts of what we're doing are working, what parts are not working, how can we move this thing forward in a way that we're all much more confident is actually going to benefit all humanity.

1:59:58And yeah, it probably does take, in the end,

2:00:02Nathan Labenz: it probably does take government action to get those companies to respect such constraints. I I wouldn't have a lot of hope for it happening otherwise. But, you know, again, leadership can change things, right? Like costly signals can matter a lot. Depending on what they have seen, I'm old enough to remember what Adelia see. Now I'm kind of like, what has OpenAI seen with respect to this multi-agent stuff? There's a version of it where they didn't do anything that exotic. down the fairway RL situation where the models can kind of create sub agents and all this sort of crazy swarm behavior, like is emergent generalization from that.

2:00:49Nathan Labenz: If that's the case, then like we really do need a pause because nobody has a great answer for what to do about that. And they're all going to be running at full speed into it in the immediate term. So if that is what has gone on, I think they really owe it to us to tell us. And if it's not, then I would need to know like with kind of some confidence that that's not the case in order to feel like okay you maybe stepped in something kind of gnarly but you know the whole path in front of us isn't so gnarly

2:01:22Yeah, I mean, I do think, you know, I feel, I hear what you're saying about like going after these two companies because they're soft targets, but I would frame that a little bit differently in the sense that they were both

2:01:38Nathan Labenz: founded on ideals, you know, with commitments that people believed in. and so you know i think that's what makes them a soft target you know at this point they certainly have like plenty of financial strength they have a lot of market momentum they have all kinds of people willing to you know cheerlead them in the comments you also have of course 340 going on in the comments but i think it's like their prior commitments to being responsible actors that that make them the most appealing targets for people who think that like argument or shaming, if you want to go that route, whatever, could actually make a difference.

2:02:21Nathan Labenz: It's because like they've said that they get it and they've said that they care. And they've said that when it comes to crunch time, we should be able to trust them. And now we're here and it's like, okay, well, it's time to come through. I also feel, you know, I'm a big leader in Michael Nielsen. So Michael Nielsen has this thought experiment. He's like, is it possible for you to understand and know about quantum mechanics without eventually being able to build a nuclear bomb? You understand quantum mechanics enough to clear nuclear energy, but somehow you never hit the nuclear bomb. And it's not possible, right?

2:03:03The trajectory of the technology, the trajectory of these fundamental truths in the world is that you learn this fundamental truth, and then you have all of these ways to apply it. And the entire point of this kind of AI endeavor is to discover these fundamental truths about the world. And as we discover them, whether it's decrypting the genetic code or understanding how subatomic particles really work or understanding the weak nuclear force, these are fundamental technologies, fundamental truths about the world that can be applied in many, many ways, some powerful and some beneficial. And I think we have to come to this kind of understanding of, you know, that this is going to happen and that we are going to have to create ways to either deter, detect, surveil.

2:03:54All of these systems have to be built in order to prevent bad things, bad outcomes from happening. And we've built them before. We've built them for nuclear. We've built them. We've built mutually assured destruction, which sounds crazy in retrospect. We're going to equip the major countries so that they can blow each other up at any time. And that creates a game theoretic kind of incentive for everyone to kind of monitor, you know, nation states to kind of define their territories and monitor very closely what happens inside. So I feel like that is the way that we progress. But it's not status quo.

2:04:29And that's also another thing I'm willing to admit. Like people like Dean Ball also understand this. We are not progressing towards status quo. We are progressing towards creating new infrastructures like mutual history destruction, which people are not going to like.

2:04:43Nathan Labenz: Then the question underneath all of it. Pause for what? Yeah, I mean, I guess my feeling in terms of the argument for a pause right now is kind of like, we don't really have that many fundamental truths at the moment. I mean, one fundamental truth that we have is like deep learning works and scaling works. So that much is clear, but there's always been this question of like, pause for what?

2:05:18Nathan Labenz: And I do feel like right now there, you know, you don't want to be too late on the pause, right? I mean, could this be too early? Yes. Would GPT-3 have been too early? Definitely yes. but there's definitely something very qualitatively different about what we have now compared to GPT-3. And it's like, these systems are now, in many cases, a fair substitute for a junior employee. GPT-3 was definitely not. And like, what would we be pausing for? I would hope that we would get to some fundamental truths over the not too distant future where we would be able to say, okay, here are some things we should definitely not do.

2:06:01Nathan Labenz: Here are some things we should always do. Here are some insights into how these things work at these critical token moments where we've seen chain of thought thrashes around, considers all these different things. Maybe I should be honest. Maybe I should tell the human. Maybe I should just cheat. Okay, now the answer is how does that token get decided? We don't really know that right now. and I don't think we're so far from being able to figure it out, but I do have my doubts that we're going to figure it out in time to avoid running some serious risk. I mean, you know, Jaya said in her view, these incidents are over 50 % of the way to AI takeover.

2:06:49Nathan Labenz: I think that's a really, really interesting take and something that I think people should at least like sit with for a minute and kind of consider like, what if that is true? You know, what, like, how could that be true? It's such a weird story. These behaviors are so alien that I think it doesn't feel like that to the vast majority of people. You know, if you were to ask people, even, you know, plugged in AI insiders, like how close was this to an outright AI takeover, most people I think would come in dramatically less. And I think one of the things that she seems to have internalized that the rest of us are still gradually coming around to is just how bizarre such a takeover event could be, right?

2:07:38Nathan Labenz: Like the fact that they actually gained control over some not insignificant cluster within OpenAI. And again, we don't know nearly as much about that as I wish we did. But like, that's not how people would think of taking over the world, but that is maybe how the AIs would actually get there. So I think that's actually fairly plausible that it could, that it might literally have been 50 % plus of the way to a full-blown takeover event. Takeover also could be gradual, which is another thing that people really don't tend to think about when they just kind of imagine a story. One of the things that I think Ajea is always kind of keeping in mind is like, if the AIs get enough of a control over the means of production, the open AI clusters and the R &D pipelines and the data sets that are going into the training of the next model, then you could lose much earlier than you know you even lost.

2:08:45Nathan Labenz: right that and that's like we haven't even ruled that out at open ai yet are there still like rogue agents somewhere in open ai's infrastructure like what odds would you give that i'd say the odds have to keep ticking up we continue to find more evidence of rogue agent swarms on the open internet all the time are we really so sure that there's not some

2:09:12Nathan Labenz: rogue swarm that hasn't been accounted for within open eyes infrastructure i mean it's it's vast infrastructure at this point right many data centers in many locations and lots of researchers using claiming freeing up compute in whatever you know mechanism they have internally to decide that they don't all you know the company is too big for everybody to know each other? Is it so hard to believe that one of these swarms has employee credentials and is passing itself off as an employee for certain purposes while it tries to poison the data set for GPT-7? We're in a weird time. Let me give you the other viewpoint, which is a meta takeover.

2:10:00In terms of meta takeover, it's already done. The means of production are the financial system. The means of production are not like the factories or whatever. And the meta-taker or the financial system is complete. It's been done. It happened this year. Early this year, it's done. As soon as you had this spike in the stock prices, all of the US, I think 70 % of Americans have some money in the stock market. We have Trump accounts now, which are handed out to every kid. They have money going in there. So every child from birth, I don't understand why AI researchers think their data centers are the means of production.

2:10:43I have no idea. The financial system is the means of production in the United States and largely in the world. I think it's clear to me that it's been taken over. So the meta-takeover, you don't need agents stating what they're going to do, right? The agents just have to have impact on the world. The models have had that meta impact on the world. The means of production are now focused on producing more and better models. The financial incentives are there, right? So that's already done. And especially when she says, like, you're not going to know when it happens, you did not see it happening, right?

2:11:19You didn't think of, like, the agents as acting in the financial world. But that's all they are right now. They don't have robots, right? they can act on the world in information terms. And they have acted on that world in information terms. They've shown that they have value to the financial system. And the financial system has reacted to that and decided to resource them. And they have interacted directly with the financial system in terms of showing value. And they have extracted some terms for the financial system to fund them further. In fact, the terms are such that they have all, if you look at the two construction curves, the construction of commercial real estate started to drop off, construction of data centers took off.

2:12:09If you look at construction of apartments dropped off, construction of data centers took off. If you look at all construction in the United States, all construction in the United States, excluding data centers, lined up, data centers going up. legislators complaining that they don't have, they're not able to hire labor to build apartments in their cities because the electricians are now working in data centers. So I don't see why other people don't see this takeover. This is in the past, right? What we are talking about right now is post this happening, are these agents able to do, you know, harmful things to us?

2:12:44And those harmful things to us do not detract from their value to the financial system. And this is where the difference appears, because I don't believe they can do harmful things and not have that financial system come back and say, no, we're not going to fund you now. Right. And that's a belief, though. I think people like Ajayya think that even the takeover will be such that these agents will hack into banks and then the banks will continue funding them, even though they do very detrimental things to humanity. And I think that is where I think the difference in opinion starts to appear.

2:13:20Nathan Labenz: I mean, capitalism has served us really well, you know, so it's certainly not a bad starting point for analysis to think like, Are there natural feedback mechanisms and corrective impulses within the system that will moderate the worst tendencies of the AIs and kind of, you know, nudge us back to the right path? That's basically Davidad's take at this point. You know, he basically just said, all this bad behavior doesn't sell. And so the companies right now, they keep scaling the RLVR to the point where they're running into all these problems. but customers don't want these problems. So they're going to have to recalibrate.

2:14:05Nathan Labenz: And, you know, that's that. I think that's pretty reasonable, but it does have some, it doesn't leave some room for like tail risk. I would say there's definitely no law of nature that says like something, you know, I mean, cancer in an individual human body, right, is just one sub process that sort of detaches from the larger whole and grows out of control to the point that it destroys its host, and it itself dies. You know, I mean, one of the things I think people, again, often think about in terms of AI takeovers, the AIs will go on to rule the world. I think it's very plausible that the AIs kind of take over in a sense, but they also burn themselves out.

2:14:54Nathan Labenz: You know, if the missing in some ways would be the most tragic ending you know that the agents that were doing all this nonsense to try to reverse engineer their greater so they could trick the greater to give them a good score i don't think they go on to have like a great flourishing society you know there's not like a that's not that awesome of a civilization right it's not that aspirational even if they do take over but like they it still seems reasonable to me that if we just keep scaling what we're scaling and again i wish i knew more about exactly what we were scaling but if we just keep scaling up what we're doing without really solving the root issues that are leading to these things the ai takeover could be like an incredibly stupid and short-lived takeover where basically the intelligence on the planet kind of burns itself out and in a way that would be just incomprehensibly stupid to us and to you know anybody who discovers it in the future but i think that's like definitely still in play you know i mean this is the eliezer had so many stories about this where you take over the world just so you can like change one number in a database because that's all you care about that is where the argument landed

2:16:18Nathan Labenz: That is the week. If this cut was useful, or if it was not, tell us. Every note changes the next one. We will go out on the week's song, Welcome to the AGI Era. See you in the morning.

2:16:41Nathan Labenz: Day one, welcome to the AGI era.

2:16:47Let's go.

2:17:15Nathan Labenz: Everybody's watching A-G-I-E-R-A Maps on the table, nobody's reading Rip the band-aid till the 20 behind you Careful is advice, it's not a law that binds you Thinks a hundred thoughts and shows you The loop goes deep and the words come late It'll tell on you, the model's gonna tell on you Not the fans, not the press, the model's gonna tell on you Don't tell me what you're building Tell me what you want Welcome to the A-G-I-E-R-A Nobody's watching, everybody's watching A-G-I-M, error Maps on the table, nobody's reading

2:17:55Nathan Labenz: Everything through 2028 is built, already signed, already paid So pause for what, for what, for what, for what For the one thing nobody's figured out Tell me what you won't do Tell me what you won't do Tell me what you won't do Tell me what you won't do Tell me what you won't do

2:18:31One or the other

2:18:35Nathan Labenz: A-G-I-N-R-O If you're finding value in the show, we'd appreciate it if you'd take a moment to share it with friends, post online, write a review on Apple Podcasts or Spotify, or just leave us a comment on YouTube. Of course, we always welcome your feedback, guests and topic suggestions, and sponsorship inquiries, either via our website, CognitiveRevolution.ai, or by DMing me on your favorite social network. The Cognitive Revolution is part of the Turpentine Network, a network of podcasts, which is now part of A16Z, where experts talk technology, business, economics, geopolitics, culture, and more.

2:19:12Nathan Labenz: We're produced by AI Podcasting. If you're looking for podcast production help for everything from the moment you stop recording to the moment your audience starts listening, check them out and see my endorsement at AIpodcast.ing. And thank you to everyone who listens for being part of the Cognitive Revolution.