Conceptual analysis demonstrates persistent deployer answerability in autonomous AI systems, indicating that delegation does not relieve human principals of moral responsibility.
Key Points
To evaluate whether deploying autonomous AI agents generates a genuine responsibility gap that diminishes principal answerability for resulting harms.
Conducted conceptual analysis and normative argumentation regarding agency, delegation, and moral responsibility.
Distinguished among the concepts of attributability, answerability, and accountability using an established philosophical tracing framework.
Formulated and operationalized a principle of responsibility conservation based on the type-level foreseeability of harmful dispositions.
Refutes the responsibility-gap thesis as a delegation illusion that improperly conflates causal remoteness with reduced deployer answerability.
Shows that AI autonomy, opacity, and adaptivity affect action attributability while leaving deployer answerability intact through prior guidance control.
Determines that answerability is epistemically graded relative to a principal's ex ante access to the system's dispositional profile rather than specific post-delegation acts.