Ethical Decision Making in Autonomous Vehicles: The AV Ethics Project

Area of Science:

• * AI Ethics
• * Robotics
• * Transportation Technology

Background:

• * Autonomous vehicles (AV) necessitate ethical decision-making protocols, particularly in unavoidable accident scenarios.
• * Existing ethical frameworks struggle to provide universally accepted guidelines for AV behavior when harm is inevitable.

Purpose of the Study:

• * To propose the Ethical Valence Theory (EVT) as a novel computational approach for AV ethical decision-making.
• * To develop a flexible framework accommodating diverse moral positions and user expectations.

Main Methods:

• * Conceptualizing AV decision-making as a process of mitigating competing moral claims from various road users.
• * Quantifying potential harm and associated uncertainties in AV actions through a deliberative process.
• * Developing a computational model for ethical implementation in real-world AV scenarios.

Main Results:

• * The Ethical Valence Theory provides a structured method for evaluating and mitigating moral claims in AV decision-making.
• * The approach quantifies harm and uncertainty, enabling ethical decisions that are coherent with real-world complexities.
• * EVT offers a flexible computational tool adaptable to various ethical viewpoints and social acceptability criteria.

Conclusions:

• * Ethical Valence Theory presents a practical and adaptable framework for autonomous vehicle ethics.
• * This approach facilitates the development of socially acceptable AV decision-making algorithms.
• * EVT serves as a valuable tool for assessing the ethical performance of autonomous systems.