In a recent video, Curt Jaimungal explores a provocative question: can governments successfully classify AI mathematics, much like they treated nuclear physics during the Cold War? The discussion, featuring insights from Geoffrey Hinton, argues that such efforts are fundamentally doomed because mathematical ideas are globally known and continually rediscovered.
Officials claim they can restrict certain AI algorithms or mathematical principles as state secrets. However, the concept of a "zeitgeist"—the shared knowledge and intellectual climate of an era—makes it nearly impossible to prevent independent discovery. Math, unlike physical weapons, is an abstract system taught in universities worldwide. If one lab clasps a breakthrough, another, perhaps across the globe, is likely to reach the same conclusion. History supports this: encryption algorithms were once classified but eventually became public.
Jaimungal's video underscores that attempts to classify AI math may be not only impractical but also counterproductive, potentially stifling innovation and cooperation. The full podcast with Prof. Hinton delves deeper into these regulatory challenges.