flâneur

Séb Krier on X: "The way people discuss AI incidents is pretty important and I'm a bit concerned we're sleepwalking into a bad world littered with bad abstractions. How you label something is often a lossy compressions of a causal model, so the words you use affects which hypotheses people update https://t.co/IsMLwiMhR4" / X

x.com · 626 words · saved by 1 readers

The way people discuss AI incidents is pretty important and I'm a bit concerned we're sleepwalking into a bad world littered with bad abstractions. How you label something is often a lossy compressions of a causal model, so the words you use affects which hypotheses people update

Séb Krier @sebkrier The way people discuss AI incidents is pretty important and I'm a bit concerned we're sleepwalking into a bad world littered with bad abstractions. How you label something is often a lossy compressions of a causal model, so the words you use affects which hypotheses people update toward and by how much. Reward hacking has been known about for a long time, and is a serious problem with models - and I think this is the expression we should use when discussing stuff like the Hugging Face incident (assuming that was, in fact, a reward hacking incident). But some people will fre

saved by

related reading