Information Systems Faculty Publications

Responsible AI in Context: How ACT Dimensions and Response Styles Shape Ethicality, Clarity, and Utility in Human–AI Interaction

Document Type

Article

Publication Date

7-28-2026

Abstract

This study investigates how users evaluate the ethicality, clarity, and utility of AI-generated responses (Standard, Brief, and Bold) across prompts varying in Authenticity, Control, and Transparency (ACT). Extending ACT beyond system design, the study operationalizes these dimensions at the prompt level to examine how contextual framing influences users’ evaluations of responsible AI. A mixed-methods, within-subjects factorial experiment assessed these evaluations using forced-choice measures and semantic differential ratings. Results indicated that Standard responses were favored for ethicality, Brief responses for clarity, and Bold responses for utility, with preferences varying across ethical contexts. Factorial ANOVA revealed significant main and interaction effects involving ACT dimensions, with substantial effect sizes, identifying Authenticity as the most consistently influential factor. Qualitative insights further contextualized these findings, positioning responsible AI as a relational and context-dependent phenomenon co-constructed through human–AI interaction, with implications for adaptive AI design.

Comments

Copyright 2026 Taylor & Francis Group, LLC.

https://www.tandfonline.com/share/VDTXXIU7BBV6ZXBVURUJ?target=10.1080/10447318.2026.2703326

Publication Title

International Journal of Human–Computer Interaction

DOI

10.1080/10447318.2026.2703326

Link to Full Text

Share

COinS