Responsibility in a Multi-Value Strategic Setting
This work addresses responsibility for AI safety and ethics, but it appears incremental as it extends prior single-outcome models to multi-value settings.
The paper tackles the problem of responsibility attribution in multi-agent systems with multiple values, presenting a model that also covers responsibility anticipation and shows that non-dominated regret-minimising strategies reliably minimise an agent's expected degree of responsibility.
Responsibility is a key notion in multi-agent systems and in creating safe, reliable and ethical AI. However, most previous work on responsibility has only considered responsibility for single outcomes. In this paper we present a model for responsibility attribution in a multi-agent, multi-value setting. We also expand our model to cover responsibility anticipation, demonstrating how considerations of responsibility can help an agent to select strategies that are in line with its values. In particular we show that non-dominated regret-minimising strategies reliably minimise an agent's expected degree of responsibility.