$RXRX Recursion Tape Reports
Per Ticker.id: $RXRX Recursion Tape Reports — 2 podcast mentions across 1 podcast (30 days), latest 2026-08-13 23:57 UTC.
And I worry that kind of to the extent that it's still incentivized for the models to keep getting better and better. There's just so much incentive to just keep cranking up the RL, stamp out the visible misalignment you can. Yes, there's some big theoretical concern about layout. There's some misalignment remaining that's harder to catch, but this becomes harder and harder to catch. It's just like you get like rare and rare cases of this really egregious stuff happening. And yeah, I think it's a pretty worrying equilibrium just because especially as the models, like all the labs, given that they're openly targeting RSI, it's like, there, there aren't market dynamics there. I have been very confused by people in the discussion being like, as they release the RSI models to the public and as the market forces around the RSI models, and it's like, if they're doing true full-blown internal AI R&D recursively, no humans in the loop, that model does not go to the public the next day. That model does not go to enterprise and have some feedback loop where it was reward hacking a lot. And so I'm like pretty worried that if we If the world was like fairly static at the current ability levels and they grew slowly, you can imagine the dynamic of reward hacking is bad for customers having a stronger effect. But I think even if it stays pretty bad, the labs are very explicitly like racing for this target. I think September is the target for OpenAI's automated AR&D intern. And then 2028 is their full-blown automation target. Anthropic is even sooner. Theirs is, their frameworks usually say is early 2027. And like, it's very soon, but, and you don't have to take any of these companies at their word, but to the extent that this is a thing that they're aiming for, reward. If it works and they're successful, there's not a feedback loop where the country uses it in a data center or whatever, reward acts a lot, and so they're used less. And so I think a lot of these things sense in worlds that stay a similar trajectory that we're on now and don't have, or that, that stay a similar capability level or on now and somehow don't keep this trajectory of capabilities going up.
Bronson Schoen — "The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis · RL's a Hell of a Drug: Metagaming, Reward Seeking & Motivated CoT Reasoning – Bronson Schoen, Apollo · 2026-08-26But as you're saying, it's already the case that the labs are generating way more revenue per megawatt than everybody else. If they stay as far ahead as they are currently, you would expect that to be the continuing case. If there's some kind of recursive self-improvement where the AI labs are relatively uplifted or they have models internally they're not releasing externally that are helping them make the next model better, you'd expect that to be even more the case. And aren't you already seeing this where SpaceX or whoever's slightly further behind will just sell compute to the highest bidder if they can't internally monetize it as well as the labs? I feel like it's continue expecting them to be able to gobble up, like bid for larger and larger shares of the compute.
Dwarkesh Patel — Dwarkesh Podcast · Dylan Patel – Anthropic & OpenAI will have most of the world’s compute by 2028 · 2026-08-25Yeah. And in that 6 months, they do recursive self-improvement internally. They just have all kinds of crazy shit happening in the company. Meanwhile, the rest of us are stuck with models that are like, at current pace, years behind.
Dwarkesh Patel — Dwarkesh Podcast · Dylan Patel – Anthropic & OpenAI will have most of the world’s compute by 2028 · 2026-08-25Can I, can I just tell you something that I got wrong on this? So, um, I love the question that you asked, which is how's our thinking evolved on like, you know, whether these things, you know, like their capabilities in generality, which is, um, I was responding to this Bostrom notion of recursive self-improvement, fast takeoff. You create one of these things, you step back and it takes over the world, right? And so I kind of poo-pooed that because that's clearly not what's happening. And I think most, a lot of people agree that that's the case, right? But here's what I got wrong. What I got wrong is I did not know that we could effectively just continue to pour money in this. Like the scaling laws are holding and I, I don't, you know, I don't know what it means to just, let's say we do a $100 billion training run to like have this thing that you're putting $100 billion in. And then that, that money comes from this metaeconomic machinery that may be able to want to solve whatever they may want to solve cancer, but they may also want to create a weapon. Like, who knows? And so this concentration of this many resources in a useful way, I think is very new. I don't think we understand the implications. I think you could reasonably argue that that's very dangerous if you kind of apply that $100 billion in the wrong way. And so I think that's kind of where this conversation needs to evolve to. So less the foom, you know, and more the what does it mean to be able to concentrate resources?
Martin Casado — The a16z Show · The New Economics of AI | Martin Casado & Steven Sinofsky · 2026-08-25100%. Now that it's here, yeah, or at least one version. And then all the conversations about LLMs versus world models versus, you know, I mean, there's a lot that's going to happen in the next few years and we may really wind up in a very different place to where we expect. I mean, almost certainly we will. I think an observation that I sort of started and then was bounced back and forth with a friend who's leading one of the leading companies was we spent a day working with his team talking about what you can do with these things, what are the applications for it? Because I think that is one of the bigger questions. And it occurred to me, it's like, I think for the first time in a very long time, there may have been moments like this in the middle of the Second World War with so many technologies just kind of pouring out of the war effort. But this feels like one idea obviously expressed, you know, the amount of brilliance over 40 years to get to this place. You don't want to, you don't want to, it's kind of like the overnight success, right? The actor, ingenue, singer who, you know, finally gets the Oscar, Grammy, whatever, has been doing this for 20 years. So these, these, you know, neural nets have been around for a very, very long time, not an overnight success, but pretty much to your point, 2.5 years ago, suddenly went from an obscure talking point for computer scientists to an everyday fact for all of our kids, everyone, everyone all over the place. And it occurred to me that unlike some of these technologies over the years, this is more like an alien spaceship crash-landed on the face of the planet. And each of these companies and all of us are exploring the ruins of it. And we kind of walk into one room and we're like, oh shit, I can do teleportation. You go to another room and it's like, okay, make movies, right? Yeah. It's less like we're making these things and more like we're sort of stumbling upon them because we've created something so powerful and so recursive that it can spew out almost weekly or monthly these wonders.
Jonathan Nolan — Uncanny Valley _ WIRED · Fallout Director Jonathan Nolan Says He's Still a Techno-Optimist (ReAir) · 2026-08-25That's still a concern. We've heard any number of different answers on that question, certainly from the companies in question, which would be a lot more positive than perhaps those who are negative on their future would be. But it's still unresolved. One key thing would be if you don't get as many data centers built and you're not offering quite as much compute, and it is possible that AI is not moving quite as quickly and so it will will slow whatever is coming their way in terms of a competitive threat, Jim. But it doesn't mean it's not still there. You know, I do wonder sometimes about the so-called race, the race with China, the race that is underway that we have to win. I still, I still don't get answers sometimes when I— well, what about when we get to— when we finally get to superintelligence or AGI, artificial general intelligence, as some describe it, and then it's all recursive self-improvement? I'm not sure what actually happened with the race.
David Faber — Squawk on the Street · 9AM HOUR: Big Market Week: Bessent, Warsh and Earnings from Nvidia 8/24/26 · 2026-08-24I'm most worried about this geopolitical part of the pause idea, which is just, well, frontier AI development is like one of the few things that the US does extremely well in comparison to China right now. And I think in a very sort of, if you're like sufficiently sort of AI-pilled, you might think that like one of the few strategic strategic edges that the US has right now is this idea of like getting to these ever quicker sort of increasing, like recursive cycles of like very, very intelligent AI development that will actually give you like a decisive strategic edge in some sense, maybe not in like the strict decisive strategic advantage sense, but like definitely it's gonna be like very helpful.
Anton Lect — ChinaTalk · How AI Becomes a Political Crisis · 2026-08-24So, well, it's— I've been on some painful calls with like a, a team of lawyers on the other side and our one lawyer. So this definitely can happen with, with some developers. The stance we've usually taken is to negotiate terms that talk about the intended outcome rather than particular model releases. So if we could have uncovered a vulnerability from testing a publicly deployed model, then we can disclose it even if we first uncovered that testing in an internal-only model. And I think that's the clearest, but yeah, this is part of why developers don't always want to do business with us for sure. I think where I see most of a lawyering in details is actually around developers' own internal evaluations or commitments where somehow it seems like no model is ever high risk according to internal evals, it's either low or medium. And that just is suspicious, but these thresholds are not clearly defined and the developers get to change them over time. So I think there is this sort of broader problem of, of grading your own homework basically. And, uh, it's good that these voluntary commitments exist, but it has created this almost perverse incentive for developers to, to sometimes downplay some risks of a— voluntary commitments don't actually kick in. I've actually avoided signing on open letters about pausing or slowing down AI, cuz I'm just not convinced that's the right approach, but Choosing the speed that you go at deliberately and not accelerating into recursive self-improvement when we're already seeing safety incidents where we don't know how to stop. I think that is very reasonable and, you know, about time. What I can see happening with voluntary commitments is abstaining from certain parts of a technology tree that could give you capability benefits, but won't immediately. And which have real bad properties for safety. I think NeuralEase is a good example of this, where a big part of why we're able to understand what was going on with these recent incidents is we could read the model's chain of thought, and it's not always perfectly faithful, but it's a pretty good window into what's going on. There are alternative model architectures that, you know, have been proposed, have been actively developed, where you lose that, where the model's just reasoning in this opaque, continuous, high-dimensional vector space.
Adam Gleave — "The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis · AI in the AM — Weekly Highlights: Relaunch Week (Aug 17–20, 2026) · 2026-08-22