Paper Type
Complete
Abstract
Generative AI Health Assistants (GAIHAs) rely on generative AI to provide healthcare information in response to users’ queries, which shapes emerging forms of human-AI co-agency in health decision-making. Since users cannot directly assess algorithmic processes, they form trust and risk judgments based on AI responses. However, limited research has examined how such perceptions influence adoption. Using the Extended Valence Framework (EVF), this quantitative study investigates how perceived actionability (epistemic), perceived empathy (relational), and perceived risk disclosure (dual) shape trust, perceived risk, and adoption intention. Using survey data from 200 U.S. participants, we employed partial least squares structural equation modeling (PLS-SEM) for analysis. Results show that all three response styles enhance trust, while empathy and risk disclosure increase risk awareness. Trust and perceived benefit predict adoption intention, whereas perceived risk does not directly constrain adoption once trust is established.
Paper Number
1439
Recommended Citation
Al-lataifeh, Zainab; Harris, Mark; Bhagat, Sarbottam; Chin, Amita Goyal; and Smith, James N., "AI Response Styles and the Adoption of Generative AI Health Assistants" (2026). AMCIS 2026 Proceedings. 17.
https://aisel.aisnet.org/amcis2026/conftheme/conftheme/17
AI Response Styles and the Adoption of Generative AI Health Assistants
Generative AI Health Assistants (GAIHAs) rely on generative AI to provide healthcare information in response to users’ queries, which shapes emerging forms of human-AI co-agency in health decision-making. Since users cannot directly assess algorithmic processes, they form trust and risk judgments based on AI responses. However, limited research has examined how such perceptions influence adoption. Using the Extended Valence Framework (EVF), this quantitative study investigates how perceived actionability (epistemic), perceived empathy (relational), and perceived risk disclosure (dual) shape trust, perceived risk, and adoption intention. Using survey data from 200 U.S. participants, we employed partial least squares structural equation modeling (PLS-SEM) for analysis. Results show that all three response styles enhance trust, while empathy and risk disclosure increase risk awareness. Trust and perceived benefit predict adoption intention, whereas perceived risk does not directly constrain adoption once trust is established.
Comments
NEXTTRANS