Improving AI prompt literacy: experimental evidence on the effect of an online tutorial among university students

Abstract

This study investigated university students’ use of generative artificial intelligence (AI) tools, focusing on prompt literacy, user behaviour, and satisfaction with AI‑generated outputs. 90 students completed a range of tasks using a generative AI tool of their choice, with half completing a tutorial on effective prompting and output evaluation. ChatGPT was used by most participants, with limited use of alternative tools and a reluctance to engage with additional features. Overall, students reported high levels of satisfaction with AI outputs, particularly for fixing codes, while satisfaction was lower for image creation and data organisation. Students who completed the tutorial were significantly more likely to produce well‑structured prompts aligned with the CLEAR framework and to critically evaluate outputs. However, self-reported confidence did not always align with observed behaviours. The findings suggest that prompt literacy is a learnable skill and that tutorials can support more effective and reflective use of generative AI.