An Epistemic Audit for Existential Risks from AI

LessWrong is publishing a new framework for auditing your beliefs about existential risks from artificial intelligence. Created by Alexander Müller, the tool works through a Google Sheets template and asks you to map your uncertainty across key questions in AI development, safety, and alignment. It follows the causal chain from building capable systems through deployment and potential existential harm. The audit serves three purposes: first, understanding where your overall view rests on genuine analysis versus borrowed assumptions; second, identifying which uncertain questions would most shift your thinking and deserve deeper study; and third, creating a shared language for discussing AI risk disagreements through concrete, numbered cruxes rather than abstract conclusions. Müller recommends quarterly re-audits to track how your confidence changes over time.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton