Delegated agency and moral responsibility in artificial intelligence

Area of Science:

• Ethics of Artificial Intelligence
• History of Technology
• Science and Technology Studies

Background:

• Current AI ethics discourse often focuses on technical autonomy and recent machine learning advances.
• This perspective overlooks the historical pattern of delegating moral authority to artificial agents.
• Ethical failures in AI are frequently linked to governance issues and human design choices.

Purpose of the Study:

• To reframe artificial intelligence (AI) ethics by analyzing the historical delegation of moral authority to artificial agents.
• To provide a normative framework for AI governance and accountability by situating AI within a longer history of artificial agency.
• To demonstrate that AI ethical risks originate from human responsibility displacement, not machine autonomy.

Main Methods:

• Qualitative, interdisciplinary approach integrating historical-thematic analysis.
• Comparative interpretation of technological artifacts, including mythological figures, automata, and early computational systems.
• Visual-conceptual synthesis to analyze delegated artificial agency.

Main Results:

• Historical artificial agents share structural features with contemporary AI: bounded autonomy, delegated authority, override mechanisms, and human oversight dependence.
• These features directly correlate with current AI ethics concerns like alignment failures and responsibility gaps.
• Consistent patterns emerge across historical contexts regarding the design and control of artificial agents.

Conclusions:

• Ethical risks in AI are fundamentally governance failures rooted in human design and accountability, not inherent machine capabilities.
• Situating AI within the history of artificial agency clarifies that moral responsibility remains with human actors and institutions.
• This historical perspective offers a robust framework for improving AI governance and ensuring accountability.