I’m glad to see this, because I’m trying to write a refutation of the AI-doom position and have had trouble identifying cogent and logical arguments on the AI-doom side. This one has flaws but is substantial enough to argue against.
I’ll note that the conclusion is much more moderate than “everyone will die”. Here the last step is that it “can end badly”—“can” instead of “will” shows possibility rather than inevitability. This step towards moderation and admission of uncertainty should be applauded.
thanks for those pointers, which I’ll check out (although the first link to a Vox article seems to be behind a paywall). I think the moderateness is to be applauded, and in general the more extreme positions are easier to refute, or at least they depend on less likely possible future paths.
I also want to refute the extremists on the other side, e.g., those claiming that AI will suddenly “end poverty”.