Tailored to Thrive: How Customized Generative Artificial Intelligence Appropriation Shapes Academic Achievement in Higher Education
Drawing on social cognitive theory and data from 420 Chinese university students, this study reveals that customized generative AI appropriation enhances academic achievement by boosting academic self-efficacy, an effect that is further strengthened by students' academic resilience.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the landscape of modern learning, a new tool has arrived with the force of a tidal wave: generative artificial intelligence. These systems can draft essays, solve equations, and explain complex ideas in seconds, leading many to wonder if they will become the ultimate shortcut to success or a crutch that weakens the mind. The central question for educators and students alike is not simply whether these tools are used, but how they are used. If a student asks a machine for a direct answer, the result is often a finished product with little learning attached. But if a student treats the machine as a partner, shaping its questions and refining its answers to fit their own needs, the experience changes entirely. This shift from passive consumption to active shaping is the heart of a recent investigation into how university students interact with artificial intelligence and what that means for their grades and confidence.
Researchers from three Chinese universities set out to understand this dynamic by looking at a concept they call "customized appropriation." In plain terms, this describes the extent to which a student takes a generic technology and bends it to their own specific purposes. It is the difference between asking a search engine for a definition and a student who modifies their prompts, selects specific AI models for different tasks, and critically edits the output to match their unique learning goals. The team wanted to know if this active, tailored approach leads to better academic results. They also explored the role of two internal factors: a student's belief in their own ability to succeed, known as academic self-efficacy, and their capacity to bounce back from setbacks, known as academic resilience. The study suggests that the way a student engages with the technology matters more than the frequency of their use.
To find the answers, the researchers followed 420 university students over a period of time, surveying them in three separate waves spaced two weeks apart. This method allowed them to see how one factor influenced another over time, rather than just taking a single snapshot. At the first stage, they measured how much the students customized their use of AI and how resilient they were when facing academic pressure. Two weeks later, they asked about the students' confidence in their ability to handle difficult tasks. Finally, they assessed the students' academic achievement, looking at both their self-reported satisfaction with their grades and their actual grade point averages. The data revealed a clear path: students who actively tailored their AI interactions reported higher levels of confidence in their academic abilities, and this confidence, in turn, was linked to better academic performance.
The study found that the relationship between customizing AI and feeling confident was not the same for everyone. It depended heavily on a student's resilience. For students who were naturally better at persisting through challenges and adjusting their strategies when things went wrong, the act of customizing AI had a stronger positive effect on their confidence. These students seemed better equipped to handle the inevitable friction of working with AI, such as dealing with imperfect answers or the need to rewrite prompts multiple times. They viewed these hurdles as part of the learning process rather than as failures. In contrast, students with lower resilience did not gain as much confidence from the same level of customization, suggesting that the psychological strength to keep trying is a key ingredient in turning AI into a learning tool.
The researchers concluded that the value of artificial intelligence in education is not inherent in the software itself but emerges from the human interaction with it. When students take control, adapting the technology to their specific needs, they build a stronger belief in their own capabilities. This belief then drives them to engage more deeply with their studies and achieve better results. However, the study also noted that this process is not automatic; it requires a certain level of mental toughness to navigate the trial and error involved in customizing a tool. The findings suggest that universities should look beyond simple rules about whether AI is allowed and instead focus on teaching students how to actively shape their interactions with these tools. By guiding students to tailor prompts, verify information, and persist through difficulties, educators can help turn a powerful technology into a genuine partner in learning.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.