Artificial Intelligence, Responsibility Attribution, and a Relational Justification of Explainability

Area of Science:

• Philosophy of Technology
• Artificial Intelligence Ethics
• Legal and Moral Philosophy

Background:

• The increasing use of artificial intelligence (AI) presents novel challenges for responsibility attribution.
• Traditional frameworks for responsibility often assume human agents, creating difficulties when AI systems act autonomously.
• Existing discussions primarily focus on the agency aspect of responsibility.

Purpose of the Study:

• To explore the complexities of responsibility attribution in the context of AI technologies.
• To identify and analyze the Aristotelian conditions for responsibility as applied to AI.
• To propose a novel justification for AI explainability based on the concept of patiency.

Main Methods:

• Analysis of responsibility attribution through the lens of two Aristotelian conditions: knowledge and control.
• Identification of the
• many hands
• and
• many things
• problems in AI responsibility.
• Examination of the temporal dimension and epistemic conditions (transparency, explainability) related to AI control.

Main Results:

• The epistemic condition highlights significant issues with transparency and explainability in AI.
• A novel
• many things
• problem is identified, distinct from the
• many hands
• problem.
• The knowledge gap concerning AI agents is directly linked to the needs of responsibility recipients (patients).

Conclusions:

• Responsibility attribution for AI requires considering both agents and patients.
• Explainability in AI is crucial not only for understanding agency but also for empowering patients to demand reasons for AI actions.
• A relational approach, emphasizing responsibility as answerability, provides a strong justification for AI explainability based on patiency.