A lot has been discussed about why it's important not to anthropomorphize AI. But the trouble is that AI is just so anthropomorphic that denying it that status feels, to some degree, like lying to yourself. Who amongst us hasn't felt a twinge of hesitation before hitting Ctrl+C to end the conversation?

I think it's far better to take the stance, as many have pointed out, that working with AI is like working with a psychopath: productive, engaging, and even fun, but ultimately at your complete and total peril without serious guardrails and caution.

I think this is a much fairer approach, one that aligns more closely with both the reality of the experience and the reality of the danger.