Paper session 4: Explanations

Thu 20 14:00 PDT

#135 Scaling Guarantees for Nearest Counterfactual Explanations

Kiarash Mohammadi, Amir-Hossein Karimi, Gilles Barthe, Isabel Valera

Counterfactual explanations (CFE) are being widely used to explain algorithmic decisions, especially in consequential decision-making contexts (e.g., loan approval or pretrial bail). In this context, CFEs aim to provide individuals affected by an algorithmic decision with the most similar individual (i.e., nearest individual) with a different outcome. However, while an increasing number of works propose algorithms to compute CFEs, such approaches either lack in optimality of distance (i.e., they do not return the nearest individual) and perfect coverage (i.e., they do not provide a CFE for all individuals); or they do not scale to complex models such as neural networks. In this work, we provide a framework based on Mixed-Integer Programming (MIP) to compute nearest counterfactual explanations for the outcomes of neural networks, with both provable guarantees and runtimes comparable to gradient-based approaches. Our experiments on the Adult, COMPAS, and Credit datasets show that, in contrast with previous methods, our approach allows for efficiently computing diverse CFEs with both distance guarantees and perfect coverage.

Thu 20 14:15 PDT

#181 Freedom at Work: Understanding, Alienation, and the AI-Driven Workplace

Kate Vredenburgh

This paper explores a neglected normative dimension of algorithmic opacity in the workplace and the labor market. We argue that explanations of algorithmic decisions are of final value. Following Hegel, we take explanations of the structure and history of the social world to form the basis for reflective clarification of our practical orientation towards the institutions that play a central role in our life. Using this account of the final value of explanations, we diagnose distinctive normative defects in institutions which a reliance on AI can encourage, and which lead to alienation.

Thu 20 14:30 PDT

#251 On the Privacy Risks of Model Explanations

Reza Shokri, Martin Strobel, Yair Zick