Comment by David Weir

“All AI research would be made public under this plan, and an international regulatory regime ‘of mutually assured compute destruction’ would come into being. [...] The outcome of the debate is the new President will declare that we must slow down AI development. To everyone’s surprise, then other superpower leading the AI race, China, agrees. But neither side wants to slow down their AI development unless they can verify the other side is too. A verification scheme emerges involving the U.S., China and other countries as well. [...] What gives Plan A credibility in my eyes is it sounds realistic and achievable. [...] Viewed that way, how can we not give it a shot? [...] Plan A offers us a way out.”