Prompt engineering knowledge predicts ethical AI awareness and learning outcomes among tertiary students in Bangladesh

This study examines the relationship between prompt engineering literacy and academic learning outcomes among tertiary-level students in Bangladesh, with particular attention to the ethical dimensions of artificial intelligence (AI) integration in higher education. Using a concurrent mixed methods design, data were collected from students across public, private, and National University-affiliated institutions through online forms, telephone surveys, and messaging-platform surveys, complemented by key informant interviews with students and educators. The findings show that students with stronger prompt engineering knowledge achieved substantially larger gains on a post-intervention knowledge assessment, and that this knowledge, together with the frequency and perceived usefulness of AI engagement, was the strongest independent predictor of learning outcomes, ahead of institutional type. Ethical concerns, most notably academic dishonesty, algorithmic bias, and overreliance on automation, were widely recognized by students, yet formal institutional guidance on ethical AI use remained rare and was almost entirely absent among National University-affiliated colleges. These results carry direct implications for policy and practice. They suggest that prompt engineering merits treatment as a foundational academic literacy rather than a technical add-on, that AI ethics training needs to reach under-resourced institutions as a matter of priority, and that closing the gap between how quickly students adopt AI and how slowly institutions govern it is now an urgent task for Bangladeshi higher education.