Integrating Generative AI into Teaching Linear Programming: Understanding Student-AI Interaction, Limitations, and Perceptions

Abstract

Generative artificial intelligence (AI) is rapidly reshaping the landscape of higher education. However, how students interact with these tools in quantitative learning contexts remains underexplored. This study examines student-AI interaction patterns, common sources of error in AI-generated solutions, and students’ perceptions of generative AI in the context of linear programming. In the assignment, students solved problems manually and then engaged with generative AI, critically reviewing AI-generated outputs through iterations. The results reveal that whereas AI-generated responses often correctly formulated decision variables, objective functions, and constraints, errors frequently arose during the solution stage, particularly in identifying feasible regions and corner points. Students frequently characterized generative AI as conditionally useful: generative AI was regarded as valuable when used appropriately, but insufficient for fostering deeper conceptual learning. Perceived learning benefits were strongly associated with students’ perceptions of reliability, overall experience, and intentions for future use. These findings highlight the importance of student-AI interaction in shaping outcomes and underscore the need to develop students’ ability to guide and critically evaluate AI-generated solutions. The study offers practical insights for instructors integrating AI into quantitative coursework and highlights directions for future research in AI literacy and instructional design.