Responsibility Gap and Diffusion in Sequential Decision-Making Mechanisms
This addresses foundational issues in AI ethics and law by providing complexity results for responsibility attribution in multi-agent systems, though it is incremental in its theoretical analysis.
The paper tackles the computational complexity of responsibility diffusion and gap in collective decision-making mechanisms, showing that diffusion-free mechanisms are Π₂-complete, gap-free mechanisms are Π₃-complete, and their intersection is Π₂-complete.
Responsibility has long been a subject of study in law and philosophy. More recently, it became a focus of AI literature. The article investigates the computational complexity of two important properties of responsibility in collective decision-making: diffusion and gap. It shows that the sets of diffusion-free and gap-free decision-making mechanisms are $Π_2$-complete and $Π_3$-complete, respectively. At the same time, the intersection of these classes is $Π_2$-complete.