Towards Better User Requirements: How to Involve Human Participants in XAI Research

TL;DR

This paper reviews current practices in involving human users in XAI research, identifies limitations in understanding user needs, and proposes methods for better capturing user requirements to improve XAI systems.

Contribution

It critically analyzes existing user involvement in XAI research and offers new suggestions for more effective requirement elicitation.

Findings

Current XAI research often overlooks comprehensive user requirement gathering.

Limitations include insufficient user involvement and unclear requirement identification methods.

Proposed suggestions aim to enhance user-centered design in XAI development.

Abstract

Human-Center eXplainable AI (HCXAI) literature identifies the need to address user needs. This paper examines how existing XAI research involves human users in designing and developing XAI systems and identifies limitations in current practices, especially regarding how researchers identify user requirements. Finally, we propose several suggestions on how to derive better user requirements.