I’m not sure how to coherently think about AI civilizations scaling. Too many likely spurious assumptions to make for the thought experiments to be compelling for me. In particular I don’t endorse applying the reasoning above.
Just trying to be transparent that I don’t think the tools here address this generalization of the problem, which is fine: scoping problems helps make them tractable, ideas can exapt in surprising ways, etc
That is a good point. I would find the work more useful if you scoped it that way explicitly.
Pretending AI progress doesn’t exist because it’s hard to think about is what the majority of the world is doing right now. I find it pretty frustrating.
I’m not sure how to coherently think about AI civilizations scaling. Too many likely spurious assumptions to make for the thought experiments to be compelling for me. In particular I don’t endorse applying the reasoning above.
Okay, well, you can’t really just skip important parts of a problem because they’re hard.
Just trying to be transparent that I don’t think the tools here address this generalization of the problem, which is fine: scoping problems helps make them tractable, ideas can exapt in surprising ways, etc
That is a good point. I would find the work more useful if you scoped it that way explicitly.
Pretending AI progress doesn’t exist because it’s hard to think about is what the majority of the world is doing right now. I find it pretty frustrating.