Generative AI as Scaffolding for Operations Research Education: Framework, Classroom Evaluation, and Design Principles

Operations research (OR) education requires students to progress through a demanding sequence of cognitive transitions, from problem framing and mathematical formulation to computational implementation, validation, and interpretation, yet scalable instructional support for these transitions remains limited. This article proposes a task-centered framework that maps generative artificial intelligence (GenAI) support functions to the main bottlenecks of OR learning and specifies the instructor role needed at each stage. We illustrate the framework through six use cases, organized by implementation complexity and evaluated in terms of pedagogical role and risks. We then evaluate one classroom implementation in depth: a study in which 66 master’s students used a customized GPT companion to implement and solve a quadratic multiknapsack case in Python and Pyomo. Mixed-methods evidence from pre-post surveys, regression and mediation analyses, a compact path model, and qualitative coding shows that the intervention worked through manageability rather than task ease. Confidence gains were large and concentrated in the computational stages of OR modeling ([Formula: see text] overall). The findings suggest that GenAI is most valuable in OR education not when it removes difficulty, but when it makes difficult modeling and implementation tasks manageable enough for students to engage, persist, and learn. Supplemental Material: The online appendices are available at https://doi.org/10.1287/ited.2025.0167 .