Sayash Kapoor

AI should be understood as normal technology with unpredictable societal effects rather than a special case, requiring adaptive policymaking and resilience frameworks instead of deterministic predictions.

“Co-author of the article and the 'AI as Normal Technology' essay”

“Signed a public letter on AI safety and embedded evaluations.”

“Co-author of an essay reframing 'rogue agent' incidents as security and governance problems rather than alignment research evidence.”

“Co-author of research finding agents can do engineering but struggle with critical parts of AI research lifecycle.”

“Kapoor collaborated with Narayanan on the theory that real-world tasks have irreducible error that limits AI performance advantages over humans.”

“Co-author articulating a vision of AI as normal technology, analyzing societal impacts and policy implications.”

“Co-authored essay analyzing AI's impact on software engineering employment and challenging mass layoff predictions.”

“Sayash Kapoor is a co-author of CORE-Bench and HAL leaderboard for agent evaluation.”