@ShcChy@florian_tramer@satml_conf Not necessarily. If we exclude slops, a poorly written paper can still offer a strong contribution that's worth having expert peers review, even if it gets rejected. We shouldn't forget the whole point of having a community; after all!
@kotekjedi_ml True! That's the funny part. If you want to see the sad part, check how much a PhD student makes (easy to find) compared to what they publish
So many discussions going around this topic. Not going to details, as I’ve always said, and as Nick’s article also points out, we should take a holistic view of mech intp, much like how we understand humans.
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
https://t.co/zHij5zwKGT