human preferences
Humans' expressed, revealed, inferred, or counterfactual comparative judgments about actions or outcomes; preferences are often used as evidence or training signals for alignment, but need not perfectly coincide with deeper human values.
Metadata
- aliases
- concept_id
- human_preferences
- preferred
- en
- human preferences
- ja
- 人間の選好
- relations
- relations
- broader
- related
- status
- active