Mathematicians' AI warning dismissed as 'protectionism' signals hollowing out of intellectual institutions

anshulkundaje · x · 2026-09-19

In a Substack essay, Ruxandra Teslo argues education should be treated as an AI safety area: the more mundane danger than AI escaping control is that humans willingly relinquish it as their own cognitive capacities atrophy. She points to mathematicians' letter about AI being dismissed as 'protectionism' as a canary in the coalmine for progressive hollowing out of intellectual institutions. Judgment, taste and purpose needed to steer AI are themselves products of intellectual work and cannot be cleanly separated from it. The essay contrasts Trump calling AI-escape concerns a 'hoax' with Dario Amodei's call to 'pace the frontier' via permanent independent evaluator access, a split Sam Altman publicly echoed. Her conclusion: if we count on human judgment to guide AI, we must maintain institutions that cultivate it.

Related event: Mathematician Warns Humans Voluntarily Ceding Control Is Worse Than AI Gone Rogue(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →