Bai and colleagues: Constitutional AI and reinforcement learning from AI feedback

What changed

Proposed training an assistant through principle-guided self-critique/revision and reinforcement learning from AI-generated preference feedback.

Scientific scope

Proposed training an assistant through principle-guided self-critique/revision and reinforcement learning from AI-generated preference feedback. Scope limitation: Experiments address the paper's harmlessness evaluations; they do not establish universal safety or remove human choices about the constitution.

Evidence

The retained primary-source abstract, bibliographic record and stated review scope support this provisional assessment. No independent experimental replication by WRS is claimed. Experiments address the paper's harmlessness evaluations; they do not establish universal safety or remove human choices about the constitution.

Contributors

• The named authors listed in the linked research record · SCIENCE: Complete source byline retained in the research record; human research direction, method development and evaluation. Individual labor shares are not measured.

• Bai and colleagues: Constitutional AI and reinforcement learning from AI feedback — original paper
• Bai and colleagues: Constitutional AI and reinforcement learning from AI feedback — company research page

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction