“If you look at DeepSeek R1's outputs, they'll say "wait" 50 times or more in a single reasoning chain. That's a lot of epiphanies to have when you're working through one problem.”
The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis · AI Research & Frontier Labs · October 2026
Bigelow, an interpretability researcher at Goodfire, was asked how models have changed over the last two years. His answer is that reasoning models get around their inability to backtrack by listing every possibility they can think of and choosing among them afterwards, which is why he calls "reasoning" close to a misnomer for what they do.
Transcript
…best guess is that we're going to see more context engineering and more of this kind of thing, more harnesses, more memories. And I think that's really what we've seen in the last year or so with the rise of agents is a lot of that engineering is just going into putting a lot of stuff in context.
…researchers based on just how dialed in they seem to be? And I'm also just interested, do you even know like how sampling is being done these days? It's a very in the weeds question. But yeah, tell me the story of the last two years. How do you understand how models have changed during that time?
I'll say first that I wish I knew what the sampling parameters were for the Claude API and for the latest GPT models. I think it's very unfortunate that they've talked these away. I expect that it's probably similar things happening. It's just that they want to limit how much people can reverse engineer the models and play with temperature to do a lot of systematic. Like I think a lot of the people that are doing really systematic prompting are using like a lot of hitting the cloud API with millions and millions and billions even of token requests. There's a high risk that they're doing something, you know, a little bit nefarious or trying to distill models or extract model weights in some way or another. So I think that's probably the main reason why they did this rather than that they're actually doing a different sampling scheme under the surface. Although I would love to learn more and I'd be happy to be surprised. As for the story for the last couple of…
When you do research, how do you choose what model to work with? Obviously, a lot of people are choosing based on what the compute resources they have available support. And I'm sure that's a factor for you too. But at Goodfire, you guys have significant resources and you've built some…
So I think I think of them as in large part standing in for each other. There are idiosyncrasies of like Quen really likes to think in Chinese. And if you do the logit lens on Quen, sometimes there will be a lot of Chinese characters sort of in the middle before it starts out putting English tokens…
does that mean that interpretability research today outside of proprietary model companies is becoming dependent on chinese models is there an american model that you could use as a substitute for chime k3 that you feel like wouldn't lose that much of the value or are we like we have to use chinese…
More from The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis
