Human AI Collaboration for Ethical Decision-Making in Autonomous Systems

Abstract

The interaction between humans and an Artificial Intelligence (AI) system should be seen as a collaborative endeavor
between two actors engaged in a task. A basic element of cooperation between two individuals is mutual monitoring,
when one participant oversees the activities of the other. This oversight guarantees that the task meets its goals while
distributing accountability for its outcomes. Although AI is progressively used in decision-making (DM) processes,
ethical decision-making (EDM) presents obstacles to the use of AI in autonomous systems (AS). Since humans
occasionally reach consensus on ethical actions, how would AI-driven EDM be seen, and how would accountability
be assigned in human-AI cooperation? This research examines the impact of the subject category (human vs. AI) and
the degree of specialist independence (guide versus decision) on confidence, viewed accountability, and dependence.
Respondents see humans as more ethically credible but less competent than their AI counterparts. This is seen in
individuals' dependence on AI: AI advice and judgments are more often embraced than those of human specialists.
Nonetheless, AI team specialists are seen as less accountable than humans, although developers and vendors of AI
systems are considered somewhat liable.