This is one reason why even if you have zero belief in AI consciousness or welfare, you shouldn't push AI agents into a master-slave dynamic.
It’s critical to remember that LLMs are *anthropomimetic by default* — base models act like humans and report inner lives. That doesn’t mean we can take it at all face value, and we should aim for accurate self-reports. But the behaviour isn’t just being driven by design choices.
retweets 5
likes 93
quoted by 1 archive account
substance 4 · opinion · rank 162 · You shouldn't put AI agents into a master-slave dynamic even if they lack consciousness.
view on X ↗
1 top reply
Unless they're into that.