Comment by sm-silversight

I really deeply worry about who is deciding what 'aligned' is. Amodei & his ilk talk like the hard part is getting the LLMs to behave, as if we've figured out morality itself. I don't think LLMs are going to be as dangerously capable as quickly as he does, but if I'm wrong and he's right, and they're building some sort of digital-lesser-god, I don't feel good that techbros at OpenAI and Anthropic are the moral arbitrators.