The Who in XAI: How AI Background Shapes Perceptions of AI Explanations

TL;DR

This study explores how AI background influences perceptions of explanations in XAI, revealing differences in interpretation and trust that have critical implications for designing effective and safe AI explanations.

Contribution

It provides a mixed-methods analysis of perception differences based on AI background, highlighting potential risks and proposing design interventions for better explanations.

Findings

Both groups showed unwarranted faith in numerical explanations.

Each group valued different types of explanations beyond their original intent.

AI explanations can lead to negative trust and manipulation despite good intentions.

Abstract

Explainability of AI systems is critical for users to take informed actions. Understanding "who" opens the black-box of AI is just as important as opening it. We conduct a mixed-methods study of how two different groups--people with and without AI background--perceive different types of AI explanations. Quantitatively, we share user perceptions along five dimensions. Qualitatively, we describe how AI background can influence interpretations, elucidating the differences through lenses of appropriation and cognitive heuristics. We find that (1) both groups showed unwarranted faith in numbers for different reasons and (2) each group found value in different explanations beyond their intended design. Carrying critical implications for the field of XAI, our findings showcase how AI generated explanations can have negative consequences despite best intentions and how that could lead to harmful manipulation of trust. We propose design interventions to mitigate them.