Exploring the psychological factors influencing intention to use AI virtual companions: an SEM-ANN study.

Authors

Yue Sun, Jun Liu, Ting Liu, Zhengqi Wei, Huajie Shen

Published in

Frontiers in psychology. Volume 17. Pages 1855345. Epub Jun 09, 2026.

Abstract

AI virtual companion products have shown promise in providing emotional support and interactive experiences, but their user acceptance and broader adoption still face challenges. Building on the TAM3 model, this study introduced AI trust, perceived anthropomorphism, and social anxiety to construct a theoretical model of usage intention.
Using survey data from 712 users in China, the study employed both PLS-SEM and ANN for empirical analysis.
The results showed that perceived usefulness, perceived enjoyment, AI trust, and perceived anthropomorphism were all significantly and positively associated with usage intention, whereas the direct associations of perceived ease of use and social anxiety with usage intention were not significant. In addition, gender showed significant differences in the paths linking AI trust and perceived enjoyment to usage intention, while age showed significant differences in the paths linking AI trust and perceived anthropomorphism to usage intention. Both SEM and ANN indicated that AI trust was the most critical predictor, whereas perceived usefulness had the lowest relative importance. However, the two methods differed in their ranking of the intermediate predictors: in SEM, perceived enjoyment was more important than perceived anthropomorphism, whereas the opposite pattern emerged in ANN.
The originality of this study lies in its theoretical extension of the user acceptance framework, its methodological demonstration of the advantages of integrating SEM and ANN, and its practical implications for optimizing virtual companion products. Specifically, the findings suggest the need to balance technical safety with emotional experience and to adopt differentiated design and promotion strategies for different user groups.

PMID:
42344986
Bibliographic data and abstract were imported from PubMed on 25 Jun 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.