What are the main things you agree with Leopold on? Maybe:
āThe Projectā is a significant possibility.
Improving cybersecurity is very important.
Neutral pro-humanity observers should prefer the US/ā the āFree Worldā to have relatively greater power than China/āother authoritarian countries, all else equal.
I think these are important points that I agree with Leopold on. But I agree with you (and your piece moved me a bit in this direction) that national securitization is risky.
Yes, the Project is a significant possibility. People like Aschenbrenner make this more likely to happen, and we should be trying to oppose it as much as possible. Certainly, there is a major āmissing moodā in Aschenbrennerās piece (and the interview), where he seems to greet the possibility of the Project with glee.
Iām actually pretty unsure whether improving cybersecurity is very important. The benefits are well known. However, if you donāt improve cybersecurity (or canāt), then advancing AI becomes much more dangerous withg much less upside, so racing becomes harder. With worse cybersecurity, a pause may be more likely. Basically, Iām unsure and I donāt think its as simple as most people think. Its also not obvious to me that, for example, America directly sharing model weights with China wouldnāt be a positive thing.
Certainly according to my ethics I am not āneutral pro-humanityā, but rather care about a flourishing and just future for all sentient beings. On this axis, I do think the difference is more marginal than many would expect. I would probably guess that US/āthe free world would be better to have relatively greater power, although with some caveats (eg Iām not sure I trust the CIA very much to have a large amount of control). I think both groups āas-isā, particularly in a nationally securitised āraceā are rather far from the optimal, and this difference is very morally significant. So I think Iām definitiely MUCH more concerned than Aschenbrenner is about avoiding a nationally securitised race (also because Iām more concerned with misalignment than I think he is).
Great points, I hadnāt thought about the indirect benefits of poor cybersecurity before, interesting!
And yes, your point about considering non-humans is well-taken and I agree. I suppose even on that my guess is liberalism is more on track to a pro-animal future than authoritarianism, even if both are very far from it (but hard to tell).
Nice post, this is a useful critique I think!
What are the main things you agree with Leopold on? Maybe:
āThe Projectā is a significant possibility.
Improving cybersecurity is very important.
Neutral pro-humanity observers should prefer the US/ā the āFree Worldā to have relatively greater power than China/āother authoritarian countries, all else equal.
I think these are important points that I agree with Leopold on. But I agree with you (and your piece moved me a bit in this direction) that national securitization is risky.
On these three points:
Yes, the Project is a significant possibility. People like Aschenbrenner make this more likely to happen, and we should be trying to oppose it as much as possible. Certainly, there is a major āmissing moodā in Aschenbrennerās piece (and the interview), where he seems to greet the possibility of the Project with glee.
Iām actually pretty unsure whether improving cybersecurity is very important. The benefits are well known. However, if you donāt improve cybersecurity (or canāt), then advancing AI becomes much more dangerous withg much less upside, so racing becomes harder. With worse cybersecurity, a pause may be more likely. Basically, Iām unsure and I donāt think its as simple as most people think. Its also not obvious to me that, for example, America directly sharing model weights with China wouldnāt be a positive thing.
Certainly according to my ethics I am not āneutral pro-humanityā, but rather care about a flourishing and just future for all sentient beings. On this axis, I do think the difference is more marginal than many would expect. I would probably guess that US/āthe free world would be better to have relatively greater power, although with some caveats (eg Iām not sure I trust the CIA very much to have a large amount of control). I think both groups āas-isā, particularly in a nationally securitised āraceā are rather far from the optimal, and this difference is very morally significant. So I think Iām definitiely MUCH more concerned than Aschenbrenner is about avoiding a nationally securitised race (also because Iām more concerned with misalignment than I think he is).
Great points, I hadnāt thought about the indirect benefits of poor cybersecurity before, interesting!
And yes, your point about considering non-humans is well-taken and I agree. I suppose even on that my guess is liberalism is more on track to a pro-animal future than authoritarianism, even if both are very far from it (but hard to tell).