Bowen Baker

So basically like the number of sequential logical steps it can do, chain of thought actually increases that for a transformer.

And so for some problems that like require, you know, complex thinking and many like serial hops of logic,

it might actually be just like necessary for the model to use the chain of thought.

And then like, so that's like, I think like the hard line case for, you know, if we were to rely on it as like a safety layer and you really wanted to only rely on chain of thought, you could probably only rely on it when the model needs to use it.

So for really complicated, like logical processing.

But the hope maybe is something like that really, you know, like anything for the model to do like actual big harm in the world might require a lot of that and a lot of a complex thinking.

So maybe that's maybe the kind of hope that, yeah.

Yeah, I guess like my I don't really know, like my guess for how it would break down for a model would be like things like move my arm, you know, from like this position to one centimeter above would not probably end up being represented in the chain of thought because the plans that people are training their models with are a bit more high level.

They're like, oh, go grab that object.

And that might be like the level of representation and planning that appears in the chain of thought.

Whereas like that move, move me, my arm one centimeter up will be like a bit more kind of like automatic or instinctive to the model.

Yeah, definitely the physical world has a lot more like, I think, hard problems than, you know, knowledge work.

And so there's going to be have to be like a lot of work on safeguards there that are that are not only chain of thought, you know, monitoring.

You definitely have to like I mean, even in robotics, you have to do like simple things, just like limiting, like figuring out what your limits for torque are in your motors so that you can.

You can't like accidentally kill someone if like, you know, you, you know, the, the, the arm just moves way too fast.

You have to like limit those things.

And so, yeah, definitely like, and I think the analogy to, you know, to knowledge work and online things is similar.

It's like, you have to kind of, there are chain of thought monitoring will not be the only thing.

safety layer in any for deploying models.

We'll have to rely on other things too, probably.