A study links prompt-writing ability with students’ perceived academic gains and finds that structured prompts produce higher-rated AI responses.
The study examined whether students’ ability to guide AI systems helps explain differences in the benefits they report receiving from them. In a survey of 380 Chinese undergraduates, prompt-writing competence was strongly associated with perceived academic gains, while its association with perceived educational inequity was statistically significant but small.
The researchers also compared basic prompts with prompts organized around a student-oriented five-part structure: role, goal, input, process and output. Across 400 matched comparisons and 800 AI-generated responses, the structured prompts consistently received higher quality ratings, especially for using supplied materials, following the requested format, maintaining context and providing actionable responses.
What the study found
The survey found a strong association between students’ prompt-engineering competence and their perceived academic performance gains. Family digital resources, gaming experience and a tendency toward planning and independent work helped explain differences in this competence. The link between family digital resources and perceived gains was largely indirect through prompt-writing competence.
Prompt-writing competence was also associated with perceived educational inequity, but the association was small. In the experiment, each of 100 prompt instances was tested in both a basic and a refined form across four task types and four AI systems, creating 400 matched comparisons and 800 outputs. The refined prompts consistently received higher ratings on overall quality, with the largest gains involving use of provided materials, format compliance, role and context consistency, and actionability.
Why prompt skills matter
The findings suggest that giving students access to the same AI system may not lead to the same educational benefits if their ability to guide and assess its responses differs. Prompt-writing competence may be one pathway connecting students’ backgrounds and prior experiences with the academic gains they perceive.
The prompt structure tested here offers a concrete way for universities to help students formulate clearer requests. However, the study supports it as a way to improve rated AI output quality—not yet as evidence that it improves student learning or reduces educational inequality.
Evidence and caveats
The study used two kinds of quantitative evidence. The first was a one-time survey of 380 Chinese undergraduate AI users, recruited through university forums and student communities. Because the survey measured all variables through self-report at one time, its associations cannot establish cause and may partly reflect common measurement bias. The participants also may not represent students in other countries, institutions or social settings.
The second part was a within-student comparison of basic and structured prompts across four task types and four AI systems. Outputs were scored on seven quality dimensions by two human experts and an AI-based rater. This experiment tested the quality of AI responses, not student learning or equity outcomes. The researchers also note that future work should more fully account for differences among tasks, AI systems and raters, report rating reliability, and test the prompt structure in classroom interventions.
// Source
International Journal of Educational Technology in Higher Education · 2026 · DOI: 10.1186/s41239-026-00633-6
Authors: Zhengdao Li, Yezi Liu, Jianan Jiang, Zhenxin Zhang, Zhihao Li, Weitong Chen
Institutions: Guangzhou University, Guangdong University of Foreign Studies, Xidian University
