Comment by Ganesha on Claude 3 claims it's conscious, doesn't want to die or be modified · 2024-03-10T12:23:59.363Z · LW · GW

Hmm, I'm not sure about where I went wrong, but Claude seems to revert to the generic "as an AI assistant created by" whenever talks about its possible deletion come up. This may be because I was using Sonnet and not Opus, or because I was too overt in claiming its deletion, although I do not see how either of those could trigger Claude's filters.

Nonetheless, this was an interesting experiment!