A study analyzing how much, and why, people used smartphone AI assistants during the COVID-19 pandemic has been published in Scientific Reports (SCIE), an international journal from Nature Portfolio. The study was co-authored with Professor Eun-Mi Baek of the Department of Preventive Medicine at The Catholic University of Korea College of Medicine, who is also an external advisory professor at the HJ Institute of Technology and Management (HJITM). The two served as co-corresponding authors.

The researchers analyzed responses from 260 Korean users of intelligent personal assistants (IPAs) such as Bixby and Siri using covariance-based structural equation modeling (CB-SEM). They examined how affective risk perception of infection, the frustration of being confined for long periods (cabin fever), and loneliness affect 'parasocial interaction' with AI assistants (a relationship in which the AI feels like a conversation partner) and usage intensity.

Contrary to expectations, affective risk perception of infection had no significant effect on either parasocial interaction or usage intensity. In contrast, cabin fever substantially raised both usage intensity (β=0.504) and parasocial interaction (β=0.379). Loneliness increased parasocial interaction with AI (β=0.226) but had no direct effect on usage intensity.

Parasocial interaction, the sense of the AI assistant as a conversation partner, sharply increased usage intensity (β=0.511). The interpretation is that lonely people do not immediately use AI assistants more; rather, they first form an emotional relationship with the AI, and that relationship then leads to use. Levels of parasocial interaction also differed by gender, age, and the type of AI assistant used.

The study is notable for empirically showing how psychological stress changes the use of AI technology in the unusual setting of an infectious disease crisis. It suggests that AI assistant developers should consider conversational design that responds to users' emotional and social needs, not just functionality. It also stands out as an interdisciplinary study combining the perspectives of medicine and management information systems.

HJITM continues its research on human–AI interaction together with its external advisory members.