Raw LLM Responses

I think a big oversight here is that the world is still ultimately people doing things, and until we have robots in the mineral mines, a completely automated energy grid, automated data centre management and construction, we can always switch things off. I was under the impression that AI companies generally overblow their capabilities, we know generally how these things work (like GANs and stuff). I'm not sure I trust the tech CEOs in their fear mongering, I think it's a sales tactic. If they push AI as a global threat, and say that our enemies might get it before we do, then they can push for the support of putting more money into their pockets than anywhere else. And about AI being evil in tests, I haven't read the paper but one of the foundational principles of AI is what you put in is what you get out, so if data sets include fiction or scenarios where similar agents acted with that kind of self-interest, they would output that kind of thing. It's a fear mongering spiral. I'm a bit of a Marxist, and with that understanding of the world the AI can't do anything until it controls the means of production. Capitalism seems to really want to give AI the means of production, so that the few billionaires can live in luxury while the rest of humanity dies. This only ends if the majority takes control while we still have the opportunity to.