Why should I trust you? Influence of explanation design on consumer behavior in AI-based services

Abstract

Purpose

This study explores how the format of explanations used in artificial intelligence (AI)-based services affects consumer behavior, specifically the effects of explanation detail (low vs high) and consumer control (automatic vs on demand) on trust and acceptance. The aim is to provide service providers with insights into how to optimize the format of explanations to enhance consumer evaluations of AI-based services.

Design/methodology/approach

Drawing on the literature on explainable AI (XAI) and information overload theory, a conceptual model is developed. To empirically test the conceptual model, two between-subjects experiments were conducted wherein the level of detail and level of control were manipulated, taking AI-based recommendations as a use case. The data were analyzed via partial least squares (PLS) regressions.

Findings

The results reveal significant positive correlations between level of detail and perceived understanding and between level of detail and perceived assurance. The level of control negatively moderates the relationship between the level of detail and perceived understanding. Further analyses revealed that the perceived competence and perceived integrity of AI systems positively and significantly influence the acceptance and purchase intentions of AI-based services.

Practical implications

This research offers service providers key insights into how tailored explanations and maintaining a balance between detail and control build consumer trust and enhance AI-based service outcomes.

Originality/value

This article elucidates the nuanced interplay between the level of detail and control over explanations for non-expert consumers in high-credence service sectors. The findings offer insights into the design of more consumer-centric explanations to increase the acceptance of AI-based services.