Researchers at the HJ Institute of Technology and Management (HJITM) have published a study analyzing the factors that determine ChatGPT use and word of mouth (WOM) among university students and office workers in Telematics and Informatics (SSCI), an international journal in the field of information and communications.

The researchers surveyed 645 university students and office workers in Korea and analyzed the data using partial least squares structural equation modeling (PLS-SEM). The integrated model comprises 13 variables in four stages: “basic attributes (perceived intelligence, system quality) → knowledge and service (knowledge acquisition and application, personalization, trust) → benefits and dispositions (utilitarian benefits, personal performance, satisfaction, innovativeness) → behavior (intention to use, use, word of mouth).”

The analysis showed that perceived intelligence — the sense that ChatGPT is smart — strongly increased knowledge acquisition (β=0.542), knowledge application (β=0.557) and personalization (β=0.472). Knowledge acquisition and application increased utilitarian benefits and personal performance but had no significant relationship with satisfaction. Personalization, by contrast, strongly increased satisfaction (β=0.424). In other words, it is the feeling of being catered to, rather than usefulness, that creates satisfaction. Personal performance (β=0.378) had the largest effect on intention to use.

Intention to use led strongly to actual use (β=0.592) and word of mouth (β=0.475), but there was no significant relationship between actual usage volume and word of mouth (β=0.007). People do not recommend the tool to others because they use it a lot; rather, the desire to keep using it and the benefits they have experienced drive recommendation. In a multigroup analysis comparing university students and office workers, the personalization → satisfaction and intention to use → word of mouth paths were stronger among students, while the utilitarian benefits → word of mouth path was stronger among office workers.

The study is significant in organizing AI tool use into a single flow that begins with intelligence and quality and ends with word of mouth. It suggests that AI service companies should focus less on increasing usage time and more on helping users experience gains and want to use the service again, and that differentiated strategies are needed — emphasizing personalized experiences for students and tangible work benefits for office workers.

HJITM continues to conduct research on the acceptance and diffusion of generative AI.