My step dad just said "Constantinople, or Istanbul, or whatever they're calling it these days. They change it all the time, seems like", and then shrugged it off when I told him the last time they changed its name was like five hundred years ago
I think "AI kills everyone" takes up too much space in our brain, and probably distorts our actions away from trying to keep AI safe. It's a weird corner solution, where there's not much bad in everyone dying. It's the messy catastrophes which are probably much more likely.
XKCD: I wrote xkcd.com/221/ "Random Number" as a cautionary tale.
Tech Company: At long last, we have created the Random Number from classic web comic Don't Create The Random Number.
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
already-clean.goodfire.com/
I binged a bunch of Nathan Fielder in preparation for the documentary and he has quickly become one of my comedy heroes. How does his mind work like that