80% of AI welfare research: We demonstrate that AIs have traits/mechanisms that in humans may be relevant to internal experience or welfare. Of course, we cannot possibly comment about if this says anything at all about whether AIs have internal experience or welfare.
retweets 6
likes 56
substance 4 · opinion · rank 116 · AI welfare research demonstrates human-like mechanisms but refuses to draw conclusions on AI experience.
view on X ↗
1 top reply
I'm sympathetic to this approach. It's a difficult but important question and our tools and understanding are limited. Finding analogs of known human mechanisms (or their absence) is weakly to somewhat informative.