Cognitive offloading in student–AI collaboration: A longitudinal analysis of prompting strategies

Prompting an AI is not a neutral act: it reflects how students think, plan, and offload cognitive effort. As large language models (LLMs) become increasingly integrated into higher education, understanding how students engage with these tools is essential. This study examines cognitive offloading through students’ interactions with LLMs when generating social network datasets. A total of 281 prompts from 122 submissions in four assignments were analyzed using qualitatively coded prompts and Co-Occurrence Network Analysis. High-quality submissions demonstrated cohesive prompting patterns, integrating contextual details, instructions, and polite language, leading to fewer disagreements and more effective task guidance. Low-quality submissions were characterised by disagreement and direct instructions, with limited contextualization and little constructive engagement. The findings show how cognitive effort is distributed in the student-AI collaboration and how prompting strategies develop in multiple assignments.