Learning Tractable Probabilistic Models for Moral Responsibility and Blame
This addresses the need for autonomous systems to make moral decisions quickly, though it is incremental as it builds on existing formalisms and hybrid methods.
The paper tackles the problem of learning tractable probabilistic models for moral responsibility and blame from data, proposing a hybrid framework that automatically induces models and enables efficient reasoning, with experiments comparing system judgments to human ones in domains like lung cancer staging and trolley problems.
Moral responsibility is a major concern in autonomous systems, with applications ranging from self-driving cars to kidney exchanges. Although there have been recent attempts to formalise responsibility and blame, among similar notions, the problem of learning within these formalisms has been unaddressed. From the viewpoint of such systems, the urgent questions are: (a) How can models of moral scenarios and blameworthiness be extracted and learnt automatically from data? (b) How can judgements be computed effectively and efficiently, given the split-second decision points faced by some systems? By building on constrained tractable probabilistic learning, we propose and implement a hybrid (between data-driven and rule-based methods) learning framework for inducing models of such scenarios automatically from data and reasoning tractably from them. We report on experiments that compare our system with human judgement in three illustrative domains: lung cancer staging, teamwork management, and trolley problems.