Generative AI did not simply appear as another technical upgrade. When tools like ChatGPT began to circulate, they quickly slipped into places where software had rarely been trusted before: drafting documents, shaping ideas, responding to customers, e...
Generative AI did not simply appear as another technical upgrade. When tools like ChatGPT began to circulate, they quickly slipped into places where software had rarely been trusted before: drafting documents, shaping ideas, responding to customers, even assisting creative work. What is striking is not only how capable these systems are, but how naturally people have begun to rely on them in situations that once depended almost entirely on human judgment.
Most existing studies have approached this shift by asking whether the technology works well—whether it is useful, efficient, or easy to operate. Those questions are necessary, but they leave something important out. People do not decide to use a technology only because it performs well. They decide because others around them are using it, because it feels socially acceptable, or because they believe they can handle it without embarrassment or failure. Someone who has already experimented with digital tools or understands how AI systems function will usually approach generative AI with curiosity rather than caution.
This is why the Theory of Reasoned Action offers a helpful way to think about generative AI adoption. It allows us to treat the decision to use these systems not as a purely technical choice, but as a social and psychological one. In this study, the innovative qualities of AI services are examined alongside two personal traits that are often mentioned but rarely explored in depth: innovativeness and self-efficacy. At the same time, the analysis asks whether people’s existing knowledge of AI and their resistance to new technologies change how strongly these traits shape adoption.
The empirical part of the study draws on survey data from South Korean users of generative AI services, a group for whom tools like ChatGPT have already become part of everyday work and communication. The survey was conducted in October 2024 and produced 324 valid responses. Standard Likert-type questions were used to capture attitudes and perceptions, and the data were analyzed with SPSS.
The choice to focus on Korean users reflects more than simple accessibility. As generative AI spreads, the gap between those who can use it effectively and those who cannot is becoming increasingly visible. This gap is not only technical; it affects who can work faster, who can learn more easily, and who gains economic advantages. Understanding how ordinary users make decisions about AI, therefore, has broader social implications.
In the analysis, adoption was shaped by both personal and social forces. On the personal side were how people felt about the service, how open they were to new ideas, how confident they were in their own abilities, what they already knew about AI, and how resistant they were to change. On the social side were the expectations and influences of others.
The findings suggest a clear pattern: people are far more likely to accept generative AI when they see it as innovative, feel positively toward it, understand it, and sense that its use is supported by those around them. Resistance existed, but it was not strong enough to stop most users from engaging with these services.
Taken together, these results point to a simple but often overlooked conclusion. The future of generative AI will depend not only on better algorithms, but on whether people feel capable, informed, and socially supported when they use them. Technologies that fit smoothly into how users think and work are more likely to survive than those that merely impress on technical grounds.