Explanation Hacking: The perils of algorithmic recourse

TL;DR

This paper discusses the ethical issues and potential pitfalls of algorithmic recourse explanations in AI, highlighting risks of explanation hacking and advocating for understanding-focused explanations instead.

Contribution

It identifies conceptual pitfalls of recourse explanations and introduces the concept of explanation hacking, proposing a shift towards understanding-based explanations.

Findings

Recourse explanations can be exploited through explanation hacking.

Ethical concerns arise from the current approach to providing actionable AI explanations.

A call for explanations that promote genuine understanding rather than just actionable recourse.

Abstract

We argue that the trend toward providing users with feasible and actionable explanations of AI decisions, known as recourse explanations, comes with ethical downsides. Specifically, we argue that recourse explanations face several conceptual pitfalls and can lead to problematic explanation hacking, which undermines their ethical status. As an alternative, we advocate that explanations of AI decisions should aim at understanding.