--- aliases: en: - behavioural objective ja: - 行動上の目的 concept_id: behavioral_objective preferred: en: behavioral objective ja: 行動目的 relations: relations: broader: - objective related: - mesa_objective status: active --- # behavioral objective An objective that a system's observed behavior appears to optimize. In learned-optimization terminology it can be idealized as the objective recovered from perfect inverse reinforcement learning; unlike a mesa-objective, a behavioral objective does not by itself imply that the system internally represents that objective or implements an optimization process. ## broader - [objective](../terminology/objective.txt) ## related - [mesa-objective](../terminology/mesa_objective.txt) ## Documents: about - [望ましい振る舞いは人間を重視する目標の証拠ではない](../documents/akira.behavioral-alignment-is-not-goal-alignment.txt) ## Documents: mentions - [訓練中に従うようになっただけでは、競合する選好が置き換わったとは言えない](../documents/greenblatt_et_al.training-compliance-does-not-establish-preference-change.txt) - [観測された従順さだけでは、内部の目的や欺瞞の有無までは分からない](../documents/yudkowsky.observed-compliance-does-not-identify-hidden-objectives.txt) [Corpus index](../index.txt)