Tangent: This is often true of humans as well.
replyWe often make a decision based on a gut feeling, and then backfill a logical reason supporting our feeling, without even realizing we're doing it -- rationalization.
And we all know some people that rationalise poor choices and misbehavior, hide their mistakes, etc, to an unacceptable degree. Sometimes the individual knows they are rationalising but continues anyway, other times they seem incapable of seeing that.
replyWhen you ask people who are rationalising poor behaviour about the scenario, but it is someone else doing it, they may arrive at a better answer. Can we use multiple LLMs to achieve self criticism and critical thinking?
isn't this happening already? There's the concept called "thinking" where the models talks with itself before giving you the final answer
replyTangent on the tangent: I think that's true in a minority of cases and in a majority of AI cases. Though in principle I think it should be possible for an LLM to have access to and faithfully represent its own reasoning.
replyOn the contrary, I would argue that it's true in a totality of AI cases.
replyTo your point, I agree that nominally there should be a way to give conceptual names to paths of weights, and when answering a question, notice which weights were and were not applied and retrospect on that.
That's not what reasoning traces as they currently exist are, though.
This was beautifully shown by asking a model to explain how it added two numbers together (something like 45+21), and it told a plausible story, when in fact they showed it was some rotation on a helix living in some internal manifold.
replyLike asking a human "how did you catch that fast ball coming at you?"
It could be that the rotation in the helix manifold whatever is a low level representation of the logical steps (carry the 2, add the next column,...) it's describing. The point stands that the explanation it generates doesn't necessarily in all cases reflect what it "actually did" but your counterexample doesn't hold.
reply