What question did this study set out to answer?

The study aims to understand how ML practitioners perceive and implement explainability in AI systems.

February 8, 20260 citationsOpen Access

How ML Practitioners Perceive Explainability: An Interview Study of Practices and Challenges

UHUmm E Habiba MHMohammad Kasra HabibUniversity of Stuttgart CGChushu Gao

Key Points

The study aims to understand how ML practitioners perceive and implement explainability in AI systems.
Conducted qualitative interviews with 14 ML practitioners from 11 different companies.
Explored perceptions and practices related to explainability in AI-based systems.
Identified categories of explainability definitions and challenges faced.
Defined four main categories of explainability.
Revealed a lack of standardized approaches to address end-user needs.
Identified eight key areas of challenges in implementing explainability.

Abstract

Because of limited guidance on how to handle explainability requirements, we conducted a qualitative interview study with 14 ML practitioners from 11 companies on the perception and implementation of explainability in AI-based systems. The study identifies four main categories of explainability definitions, reveals that standardized approaches for addressing end-user needs are lacking, and categorizes challenges into eight areas. The findings highlight the need for a structured requirements engineering approach for explainability.

问 AI

Bookmark

View Full Paper