AI and Ethics When Human Beings Collaborate With AI Agents

Area of Science:

• Human-Computer Interaction
• Cognitive Psychology
• Artificial Intelligence Ethics

Background:

• Human-AI interaction is increasingly a collaborative process between two agents.
• Effective collaboration necessitates co-supervision, where each agent monitors the other's actions.
• This co-supervision is crucial for achieving objectives and establishing shared responsibility for outcomes.

Purpose of the Study:

• To explore the concept of co-supervision in human-AI collaboration.
• To investigate how psychological research can inform the establishment of responsibility boundaries.
• To propose a methodology for assessing shared responsibility in human-AI activities.

Main Methods:

• Review of psychological research on supervision and evaluation of agent actions.
• Analysis of cognitive and non-cognitive factors influencing supervision effectiveness.
• Conceptual framework development for shared responsibility in human-AI systems.

Main Results:

• Co-supervision is essential for shared responsibility in human-AI collaborations.
• Psychological insights can guide the programming of AI for defined responsibility limits.
• Existing research on human supervision offers a basis for AI supervision models.

Conclusions:

• A methodology for assessing responsibility in human-AI collaboration is proposed.
• Understanding supervision dynamics is key to defining human and AI accountability.
• Further development is needed to operationalize responsibility assessment in collaborative AI tasks.