Paper Type

Complete

Abstract

The use of AI-based decision systems in high-impact domains highlights the need to balance explainability and privacy. While explainable AI (XAI) promotes transparency and accountability, privacy-preserving techniques restrict information disclosure, creating trade-offs rarely addressed in an integrated way. Existing studies often focus on technical or governance aspects separately, offering limited support for organizational decisions. This study proposes a governance-oriented framework that treats explainability as a sociotechnical capability embedded in organizational processes. Based on Design Science Research, it organizes decisions into four layers: context and risk; stakeholder needs; explanation design with privacy safeguards, and governance and accountability. An expert evaluation with Brazilian public administration professionals showed positive perceptions of usefulness and governance, while indicating the need for further refinement. By reframing explainability as a governance capability rather than a purely technical attribute, this study contributes to Information Systems research on responsible AI and supports structured decision-making in privacy-sensitive organizational environments.

Paper Number

1221

Comments

SIG SEC

Share

COinS
 
Aug 15th, 12:00 AM

A GOVERNANCE-ORIENTED FRAMEWORK FOR BALANCING EXPLAINABILITY AND PRIVACY IN AI-BASED DECISION SYSTEMS

The use of AI-based decision systems in high-impact domains highlights the need to balance explainability and privacy. While explainable AI (XAI) promotes transparency and accountability, privacy-preserving techniques restrict information disclosure, creating trade-offs rarely addressed in an integrated way. Existing studies often focus on technical or governance aspects separately, offering limited support for organizational decisions. This study proposes a governance-oriented framework that treats explainability as a sociotechnical capability embedded in organizational processes. Based on Design Science Research, it organizes decisions into four layers: context and risk; stakeholder needs; explanation design with privacy safeguards, and governance and accountability. An expert evaluation with Brazilian public administration professionals showed positive perceptions of usefulness and governance, while indicating the need for further refinement. By reframing explainability as a governance capability rather than a purely technical attribute, this study contributes to Information Systems research on responsible AI and supports structured decision-making in privacy-sensitive organizational environments.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.