Ask HN: What Would Slowing Down AI Look Like?

Summary

A Hacker News discussion asks what the second-order effects of slowing artificial intelligence development would be. The post raises possibilities including higher token prices, canceled IPOs, a burst AI bubble, the return of laid-off jobs, and delayed or resumed data-center construction. One commenter interprets Dario’s call for slower progress as a moderate reduction in pace, from roughly 90% throttle to 75–85%, while continuing to advance AI quickly. That commenter suggests using about 20% of each development cycle to build defensive capabilities. The original poster argues that this could amount to industry self-regulation designed to get ahead of legislation, while another commenter says even a slower pace could still transform society. The discussion distinguishes a broad slowdown in AI’s social impact from a narrower effort to gain one or two years for alignment work and reduce the risk of models compromising critical systems or contributing to catastrophic actions.