concept
Veil of Ignorance
Choosing governing principles without knowing one’s own position or advantage.
The veil of ignorance is an impartiality device associated with John Rawls: choose governing principles without knowing how one’s own position will benefit. Using the Veil of Ignorance to align AI systems with principles of justice tests a limited version for selecting AI-assistance principles.[1]
It offers one approach to Social Alignment and value disagreement. Experimental support for a procedure in a bounded task does not make its chosen principle morally compulsory or show that an AI can implement it reliably.
Psychological evidence
Veil-of-Ignorance Reasoning Favors the Greater Good tests shifts in judgments after impartial-perspective exercises. A measured shift does not establish the moral correctness of the resulting judgment.
Sources
Pages that link here
- Artificial Intelligence, Values and Alignment paper
- Collective Good and Optimization in Socioeconomic Systems paper
- Engineering a social contract: Rawlsian distributive justice through algorithmic game theory and artificial intelligence paper
- Iason Gabriel person
- Using the Veil of Ignorance to align AI systems with principles of justice paper
- Veil-of-Ignorance Reasoning Favors the Greater Good paper
Last updated 2026-10-07