Pausing AI at human level seems harder than pausing ASAP

Per LessWrong, pausing AI development now may actually be easier than waiting until AI reaches human level. The conventional argument for delay is that human-level AI would be so economically valuable and useful for safety research that a future pause becomes possible. But the piece argues those same economic incentives would make enforcement nearly impossible once AI is that advanced. It also flags a key tension: if AI is smart enough to accelerate safety research, it's probably already smart enough to accelerate capabilities development even faster. One potential counterweight the analysis raises is that widespread job displacement could create enough public pressure to override industry interests in continued development.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton