tomspark wrote:
Anthropic's gotten way better at transparency, but "model welfare" still feels like it's dodging the real question — does the model actually suffer, or are we just pattern-matching human suffering onto something that might not experience anything at all? The framing matters.