Comment by Michael David Angel

Plan A is the alternative. Delay superintelligence to 2040. Make all AI research public. Let dozens of companies in dozens of countries catch up to the frontier. And enter, deliberately and with eyes open, a regime of mutually assured compute destruction. [...] Under the deal, all post-deal frontier training happens in new, purpose-built, highly secure R&D datacenters. And when the Consortium builds them, the United States and China agree to site them in the third-party countries least secure against the other’s military intervention. America’s clusters go to Mongolia. China’s go to Canada. [...] I don’t know if Plan A is right.