Collective Alignment

Collective Alignment: OpenAI Surveys 1,000+ People on Model Spec Defaults

OpenAI conducted a global survey of over 1,000 participants to gather public input on how AI should behave, comparing responses against its existing Model Spec. The initiative, called 'collective alignment,' aims to shape AI default behaviors to better reflect diverse human values. Results are being used to update or validate Model Spec guidelines. This represents a structured attempt to incorporate democratic input into alignment policy.