Any reason to protest scaling instead of hardware or algorithmic development? (IDK if a comment on this post is the best place to say this, but I couldn’t think of a better place.)
I’d probably be in favor of slower scaling at current margins, but I don’t feel super confident in this. However, I’m strongly in favor of stopping hardware developement which seems more clearly bad. It’s probably harder to stop/regulate algorithmic development than either scaling or hardware.
I suppose frontier scaling is more obviously connected to danger which makes it a better political message?
Also, stopping frontier scaling is probably best for reducing short term risk (<2 year out) per unit of economic damage to the existing economy. Quite strong hardware pauses can also prevent short term risk and have other positive effects, but they also have stronger and more obvious economic damages. (In fact, I think very strong hardware pauses are probably more robust than scaling pauses for reducing short term risk.)
As far as what I mean by “very strong hardware pauses”:
Stop all H100 manufacturing (and ideally A100 manufacturing)
(Ideally) goverment buys back H100 and A100s and locks up this chips (or destroys them)
Improving hardware technology is banned, maybe nationalize nvidia and similar companies and mostly dismantle them.
This is obviously quite expensive, but slowing down hardware in other weaker ways still seems robustly good to me (ideally, slowing development globally instead of just in the US).
Personally, I’m interested in targeting hardware development and that will be among my future advocacy directions. I think it’ll be a great issue for corporate campaigns pushing voluntary agreements and for pushing for external regulations simultaneously. This protest is aimed more at governments (attending the UK Summit) and their overall plans for regulating AI, so we’re pushing compute governance as way to most immediately address the creation of frontier models. Imo hardware tracking at the very least is going to have to be part of enforcing such a limit if it is adopted, and slowing the development of more powerful hardware will be important to keeping an acceptable compute threshold high enough that we’re not constantly on the verge of someone illegally getting together enough chips to make something dangerous.
Any reason to protest scaling instead of hardware or algorithmic development? (IDK if a comment on this post is the best place to say this, but I couldn’t think of a better place.)
I’d probably be in favor of slower scaling at current margins, but I don’t feel super confident in this. However, I’m strongly in favor of stopping hardware developement which seems more clearly bad. It’s probably harder to stop/regulate algorithmic development than either scaling or hardware.
I suppose frontier scaling is more obviously connected to danger which makes it a better political message?
Also, stopping frontier scaling is probably best for reducing short term risk (<2 year out) per unit of economic damage to the existing economy. Quite strong hardware pauses can also prevent short term risk and have other positive effects, but they also have stronger and more obvious economic damages. (In fact, I think very strong hardware pauses are probably more robust than scaling pauses for reducing short term risk.)
As far as what I mean by “very strong hardware pauses”:
Stop all H100 manufacturing (and ideally A100 manufacturing)
(Ideally) goverment buys back H100 and A100s and locks up this chips (or destroys them)
Improving hardware technology is banned, maybe nationalize nvidia and similar companies and mostly dismantle them.
This is obviously quite expensive, but slowing down hardware in other weaker ways still seems robustly good to me (ideally, slowing development globally instead of just in the US).
Personally, I’m interested in targeting hardware development and that will be among my future advocacy directions. I think it’ll be a great issue for corporate campaigns pushing voluntary agreements and for pushing for external regulations simultaneously. This protest is aimed more at governments (attending the UK Summit) and their overall plans for regulating AI, so we’re pushing compute governance as way to most immediately address the creation of frontier models. Imo hardware tracking at the very least is going to have to be part of enforcing such a limit if it is adopted, and slowing the development of more powerful hardware will be important to keeping an acceptable compute threshold high enough that we’re not constantly on the verge of someone illegally getting together enough chips to make something dangerous.