flâneur

AI 2040: Plan A

ai-2040.com · 9,263 words · saved by 1 readers

A detailed forecast and recommendation for how the US, China and the rest of the world should navigate superintelligence.

In the Plan A scenario, we propose international coordination to develop AI safely. Keeping risk at an acceptable level requires substantial slowdowns in the pace of AI development. A central blocker for this type of coordination is the risk that one party develops AI in secret, thus gaining an advantage. Fortunately, it seems very hard to build and operate covert AI datacenters in a way that would be catastrophic. In the rest of this supplement, we’ll analyze how difficult it would be for China to defect from Plan A in secret, and the mitigations that can be put in place to prevent…

saved by

related reading