AI Security is Harm Reduction

In a post on LessWrong, researcher Quinn defends working on AI security while questioning broader AI development, drawing a parallel to harm reduction in drug policy. Quinn points to nineteen-nineties needle-exchange programs in Philadelphia that assumed addiction was inevitable and sought to minimize harm. Quinn applies the same reasoning to AI: if recursive self-improvement will happen regardless of disapproval, making it as secure as possible is the right move. The post explores the logic behind reducing risks from technologies one doesn't fully endorse.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton