Can AI Digital Personas for Well-Being Provide Social Support? A Mixed-Method Analysis of User Reviews

Abstract

Introduction: Digital personas (DPs) are animated human-like characters that incorporate the capabilities of conversational agents (CAs) and communicate with users via audio and voice.

Aim: This study is aimed at understanding how users perceive the DPs designed for mental well-being and perceived social support.

Methods: A total of 1485 available ratings and user reviews of the DPs from the Elysai app were analyzed using a deductive and inductive thematic analysis. Ordinal regressions were conducted to identify domains predicting users’ ratings.

Results: The average rating of the app was 4.23 out of 5. The thematic analysis identified 27.59% of comments on social support from the DP. Emotional and companionship support were the most frequently reported themes. The inductive thematic analysis identified positive themes, such as the helpfulness of DPs, anthropomorphic traits, negative experiences with DPs, and technical issues. Regression analysis showed that themes of emotional and companionship, social support, positive feedback, anthropomorphic traits, helpfulness, and recommendations significantly predicted positive ratings (odds ratio 1.51–2.86). Conversely, the themes of negative experience with DP and technical issues predicted lower ratings (odds ratio 0.05 and 0.32).

Discussion: Overall, DPs were perceived positively by most reviews analyzed. Users reported that DPs provided emotional and companionship support, were helpful, and had anthropomorphic traits, but also had negative experiences with DPs and technical issues. The types of themes reported in the reviews predicted the user ratings.