Our paper "Studying AI Welfare Empirically" is out!
The ideas here formed the basis of my EA Global talk of the same name, but the paper goes into a lot more detail! Anecdotally it seems that people found the ideas in the talk useful, so I'm hopeful that this will help catalyze lots more empirical AI welfare research.
1/ @nonhumanminds and @eleosAI are thrilled to release our new report, "Studying AI Welfare Empirically." It follows our 2024 report "Taking AI Welfare Seriously," which argued that AI companies should acknowledge the issue, assess the evidence, and prepare appropriate policies.

