Explaining and Sustaining AI Assimilation in Higher Education: The Dual Roles of Engagement and Skeptical Engagement

In the classrooms of today's universities, a quiet revolution is reshaping how students learn to write. Artificial intelligence tools, capable of generating text and offering instant feedback, are no longer just futuristic concepts; they are embedded in the daily workflow of higher education. Yet, simply having these tools available does not mean students are using them effectively. The real challenge lies in assimilation: the process where a new technology stops being a novelty and becomes a natural, routine part of how a person works. For years, researchers have tried to understand why some people embrace new technology while others reject it. They have looked at what makes a tool useful and easy to use, and they have also studied the fears that hold people back, such as worries about privacy or the feeling of being controlled by a machine. A prevailing assumption has been that fear and risk are simply barriers that stop adoption, and that to get people to use AI, we must remove those fears. However, this view might be too simple. It treats the human mind as a switch that is either on or off, ignoring the complex, critical thinking that happens in between.

A team of researchers from North South University, Universiti Teknologi Brunei, and Curtin University Sarawak set out to investigate this more nuanced reality. They wanted to understand how students actually integrate AI into their writing process, particularly when they are drafting and revising essays. Instead of assuming that students either love or hate these tools, the researchers proposed that there are two distinct ways students engage with them. The first is straightforward engagement, where students are motivated and trust the system. The second, which the researchers call "skeptical engagement," is a state of cautious, critical use. In this state, a student does not blindly accept the AI's suggestions but instead questions them, verifies them, and selectively applies them. The study sought to determine if this skeptical approach, often seen as resistance, might actually be a productive way to learn and sustain the use of AI over time.

To uncover the truth, the researchers employed a two-step approach. First, they listened to the voices of 22 students who had used AI for their writing assignments. These students wrote detailed essays about their experiences, describing what helped them and what held them back. The researchers analyzed these stories to build a map of the factors at play. They found that students felt most engaged when they felt they had control over their work, when the AI seemed to understand their emotional needs, and when the feedback was clear and fair. Conversely, they felt stressed or cautious when the technology was overwhelming, when they felt they were losing control of their own writing, or when they worried about ethical issues like data privacy. This initial phase revealed that students were not simply rejecting the technology; they were actively negotiating with it, often adopting a stance of critical scrutiny.

Armed with these insights, the researchers then surveyed 303 students to test their ideas on a larger scale. They asked these students to rate how much they agreed with various statements about their use of AI. The results confirmed that both enthusiasm and skepticism drive the long-term use of AI, but they do so in different ways. Factors that supported a student's sense of autonomy and competence, such as the ability to choose their own writing goals and receiving clear explanations for the AI's suggestions, strongly increased their positive engagement. On the other hand, factors like the mental fatigue caused by constantly checking the AI, the fear of losing control over one's own ideas, and concerns about ethical fairness, did not cause students to stop using the tool. Instead, these risks led to a higher level of skeptical engagement. The students did not walk away; they stayed, but they stayed with their guard up, carefully examining every piece of advice the machine offered.

The study found that both of these paths lead to the same destination: assimilation. Students who were enthusiastically engaged used the AI more, and students who were skeptically engaged also used the AI more, provided they felt the partnership was working well. In fact, the data showed that when students perceived a high-quality collaboration with the AI—where the tool felt responsive and helpful rather than intrusive—their critical, skeptical approach became even more effective at turning into routine use. The skepticism did not disappear; it became a tool for better integration. The researchers discovered that students who questioned the AI's output were not failing to adopt the technology; they were adopting it in a way that preserved their own agency and critical thinking skills.

One of the most significant findings was that the traditional view of risk as a simple barrier was incomplete. The study explicitly showed that risk awareness does not necessarily suppress adoption. Instead, it can cultivate a mode of use that is more thoughtful and responsible. The researchers also found that not all factors were equally important. While the feeling of having control over the writing process was the strongest driver of positive engagement, the specific skill of learning how to use the tool did not significantly predict engagement in this context. Similarly, while students worried about bias in the AI, this specific fear did not significantly drive their skeptical behavior in the way that mental fatigue or loss of control did. This suggests that the immediate experience of the interaction matters more than abstract concerns about the system's internal logic.

The implications of this research are clear for anyone involved in education or technology design. It suggests that trying to make AI tools perfectly seamless and risk-free might not be the best way to encourage their use. Instead, educators and designers should create environments that support critical thinking. They should encourage students to question the AI, to explain why they accept or reject a suggestion, and to maintain control over the final product. By treating skepticism not as a problem to be solved but as a valuable part of the learning process, institutions can foster a deeper, more sustainable relationship between students and artificial intelligence. The study concludes that the future of AI in education does not depend on students blindly trusting the machine, but on their ability to work with it critically, ensuring that the technology serves the learner rather than replacing them.