The prevailing notion in AI safety circles is that a pivotal act—an action that decisively alters the trajectory of artificial intelligence development—requires superhuman AGI, which itself poses extreme risks. I challenge this assumption.
Consider a pivotal act like "disable all GPUs globally." This could potentially be achieved through less advanced means, such as a sophisticated computer virus akin to Stuxnet. Such a virus could be designed to replicate widely and render GPUs inoperable, without possessing the capabilities to create more dangerous weapons like bioweapons.
I've observed a lack of discussion around these "easier" pivotal acts in the AI safety community. Given the possibility that AI alignment might prove intractable, shouldn't we be exploring alternative strategies to prevent the emergence of superhuman AI?
I propose that this avenue deserves significantly more attention. If AI alignment is indeed unsolvable, a pivotal act to halt or significantly delay superhuman AI development could be our most crucial safeguard.
I'm curious to hear the community's thoughts on this perspective. Are there compelling reasons why such approaches are not more prominently discussed in AI safety circles?
The naming might be confusing because pivotal act sounds like a one time action, but in most cases getting to a stable world without any threat from AI requires constant pivotal processes. This makes almost all the destructive approaches moot (and they're probably already bad for ethical concerns and many others already discussed) because you'll make yourself a pariah.
The most promising venue for a pivotal act/pivotal process that I know of is doing good research so that ASI risks are known and proven, doing good outreach and education so most world leaders and decision makers are well aware of this, and helping setup good governance worldwide to monitor and limit the development of AGI and ASI until we can control it.