Students as strategic users: examining the calculating learner hypothesis applied to the use of generative AI for college coursework

Generative AI (GenAI) has intensified concerns regarding academic integrity. Guided by the calculating learner hypothesis, in which students maximise perceived benefit while minimising effort, this exploratory study examined when, where, and why undergraduates use GenAI.Undergraduates (N = 538) reported GenAI use and perceptions of value, difficulty, cost, and effort-minimisation orientation for two of their courses. Use was categorised as supplementary (learning-focused) or substitutionary (completion-focused, such as producing assigned work). Multilevel models separated within- from between-student variation.Most students (78.4%) used GenAI in at least one course, and supplementary use exceeded substitutionary use, which was reported in 8–11% of courses. Students used GenAI more in their more difficult course. Among users, use was more frequent in courses perceived as lower in value, higher in cost, and higher in effort minimisation. Substitutionary use was more likely in lower-valued courses and those with stronger effort minimisation; supplementary use was unrelated to effort minimisation, which did not mediate the value-use association.The hypothesis was partially supported. GenAI use varied more across courses within students than between students: the same student may use GenAI to support learning in one course and to reduce effort in another.