Trust Explanations to Do What They Say

TL;DR

This paper emphasizes the importance of providing clear, contract-like explanations for AI decisions to ensure appropriate trust calibration, advocating for transparency about when explanations are reliable.

Contribution

It introduces the concept of trust contracts for AI explanations, proposing that explanations should specify use cases where they are trustworthy, enhancing responsible AI deployment.

Findings

Automated explanations can build trust if properly calibrated.

Trustworthiness of explanations should be explicitly communicated.

Contracts for explanations can improve AI transparency and user trust.

Abstract

How much are we to trust a decision made by an AI algorithm? Trusting an algorithm without cause may lead to abuse, and mistrusting it may similarly lead to disuse. Trust in an AI is only desirable if it is warranted; thus, calibrating trust is critical to ensuring appropriate use. In the name of calibrating trust appropriately, AI developers should provide contracts specifying use cases in which an algorithm can and cannot be trusted. Automated explanation of AI outputs is often touted as a method by which trust can be built in the algorithm. However, automated explanations arise from algorithms themselves, so trust in these explanations is similarly only desirable if it is warranted. Developers of algorithms explaining AI outputs (xAI algorithms) should provide similar contracts, which should specify use cases in which an explanation can and cannot be trusted.