An Open Letter to Steven Pinker on AI

Context

Writing on 6 October 2026, Scott Alexander opens his letter with a tongue-in-cheek challenge to a duel. His target is the psychologist Steven Pinker, who has spent years dismissing the existential risks of artificial intelligence. Alexander zeroes in on a frustrating habit among prominent optimists, who often brush off safety concerns by attacking cartoon versions of what researchers actually believe. He lines up fifteen years of Pinker’s dismissals against the physical progress of machine learning, forcing a pointed question: how should public thinkers handle the messy, entirely real risks of new technology without resorting to denial?

Show notes