They don’t “feel”. The Hard Problem of Consciousness isn’t even close to being solved in humans, let alone a LLM. We are nowhere near AGI. And even if someone claims this, establishing whether it was just simulating AGI might be impossible to tell.
I struggle with the current debate on AI. It’s both way more important than we think but also nowhere near sentience. There are compelling “biological chauvinist” arguments against mechanical AI being able to ever have the same class of consciousness as biological beings.
Either way. The millisecond we switch on a machine that has a “what it feels like to be X” we just created a slave. Massive ethical issues to be addressed.
As I’ve said I think a mainstream debate on AI is all about how well it simulates human interaction. And like ridiculous anthropomorphic robots misses the point.
AI is for folding proteins, not simulating Hawk Tua girl.
