Exploring factors influencing the adoption and usage of ChatGPT: Internet usage patterns in South Korea

Abstract

ChatGPT, a large-language-model (LLM) chatbot, is reshaping everyday activity from productivity tasks to leisure pursuits. To illuminate its early diffusion, we analyze six months of panel-based internet-usage logs for 6,791 South-Korean users, covering 15 site categories. Logistic-regression results show that intensive engagement with job- and education-related platforms, entertainment sites, community forums, and news/media outlets significantly increases the likelihood of adopting ChatGPT, whereas frequent visits to finance/real-estate and horizontal portal sites exert a negative effect. Interpreted through the Technology Acceptance Model and Diffusion-of-Innovation theory, the positive categories align with high perceived usefulness, ease of use, relative advantage, and compatibility, while the negative categories highlight risk sensitivity and low task–technology fit. By marrying behavioral trace data with established psychological constructs, the study advances theory on generative-AI uptake, demonstrates the value of usage logs over self-reports, and emphasizes the importance of cultural context in global adoption research. Practically, the findings furnish evidence-based guidance for developers and marketers seeking to target early adopters and accelerate the diffusion of generative-AI services tailored to diverse user needs. • First real-world log study linking web habits to early ChatGPT adoption • Job, education, entertainment, community, news use predicts higher adoption • Finance and portal browsing predicts lower adoption due to risk concerns • Behavioral data confirms usefulness and compatibility drive generative-AI uptake • Insights guide targeted marketing of ChatGPT to accelerate global diffusion

Ähnliche Arbeiten