Chinese students and generative AI: confident in their prompts, but they hardly adapt to the type of duty

A survey of 234 students shows that they provide context and respond to their requests, but use almost the same strategies for creative exercise and research.

Having access to generative AI is not enough: you also need to know how to formulate, evaluate and refine your requests. Lei Fan and Fangxue Liu surveyed 234 university students in China about their perceived comfort with prompts and the strategies they report using according to academic tasks.

Students use AI for a variety of activities and report being quite proficient, especially at providing context and rephrasing after an initial answer. But their strategies vary little by type of task: only the precision of the expected format is more common for preparing a creative workshop than for a research assignment. Many agree with both concise instructions and giving everything from the start, suggesting that good brevity is about sorting the information rather than length. Ready-made prompt models are the least popular.

In the statistical model chosen, the perceived cognitive dimension is the most closely linked to the feeling of mastery, and the free comments point to a recurring difficulty: obtaining an answer that really corresponds to the initial intention.