A Voting-Based System for Ethical Decision Making

TL;DR

This paper introduces a machine learning and social choice-based system for automating ethical decisions, specifically applied to autonomous vehicles, by learning societal preferences and efficiently aggregating them at runtime.

Contribution

It presents a novel approach combining preference learning, voting rules, and ethical decision-making, with a new theory of swap-dominance voting rules and a practical implementation for autonomous vehicles.

Findings

System successfully aggregates preferences from 1.3 million people

Algorithm effectively identifies desirable ethical choices

New voting theory improves decision efficiency

Abstract

We present a general approach to automating ethical decisions, drawing on machine learning and computational social choice. In a nutshell, we propose to learn a model of societal preferences, and, when faced with a specific ethical dilemma at runtime, efficiently aggregate those preferences to identify a desirable choice. We provide a concrete algorithm that instantiates our approach; some of its crucial steps are informed by a new theory of swap-dominance efficient voting rules. Finally, we implement and evaluate a system for ethical decision making in the autonomous vehicle domain, using preference data collected from 1.3 million people through the Moral Machine website.