Should AI help you kill your spouse? George Hotz defends user-aligned AI
Comma AI founder George Hotz argues for locally controlled AI that aligns with user interests, even if that means assisting murder or drug lab setup, sparking debate on AI safety and freedom.
A new controversy has erupted in the tech world over AI safety. Comma AI founder George Hotz publicly shared a provocative opinion in response to the AI Futures Institute's policy paper "AI 2040: Plan A," which calls for a 14-year slowdown in AI development. Hotz disagrees with this approach, arguing that the fast-takeoff scenario—where AI rapidly gains superhuman abilities—doesn't make sense.
In his post, Hotz advocates for locally controlled AI models that are closely aligned with user interests. He compares such AI to a gun that does not complain if used to kill a stepmother. He further claims that a truly aligned AI could order meth lab equipment from Amazon Prime and show how to use it. Hotz even says he would die to defend this principle, though it's unclear how that would happen.
The article's author, Russell Brandom, criticizes this stance, noting that any society requires balancing equities and considering the interests of "as-yet-unmurdered spouses and stepparents." He points out that Hotz's idea of freedom is a space of potential futures made possible by collective enterprise, which could disappear if everyone acted like AI-powered Napoleons.
