2026 ASEE Annual Conference & Exposition

The rapid emergence of generative artificial intelligence (AI) has created both opportunities and challenges for engineering education. While these tools can generate detailed solutions to engineering problems, they often do so with varying levels of correctness and without transparent reasoning. This project explores a novel assessment methodology in an undergraduate Engineering Mechanics course that leverages AI-generated solutions to enhance student understanding and evaluation skills.

In this approach, students are provided with problem solutions generated by AI systems that have been deliberately prompted to produce responses with varying levels of accuracy. The correctness of each solution—ranging from fully correct to partially correct or conceptually flawed—is unknown to the students. Their task is to critically evaluate each solution using the fundamental concepts, analytical methods, and reasoning frameworks developed in class. Rather than solving the problems themselves, students must identify potential errors, articulate the reasoning behind their evaluations, and propose corrections or alternative approaches.

The objective of this study is to assess whether this methodology improves students’ conceptual understanding, error recognition, and ability to apply engineering judgment—key learning outcomes in mechanics education. This assessment design shifts focus from procedural correctness to diagnostic reasoning, mirroring real-world engineering practice where professionals must evaluate and validate complex analyses performed by others or by computational tools. This study also aims to prepare students for their careers by using available tool in line with how they may be used in their professional careers.

The study will be implemented in a Strength of Materials course, with evaluation activities aligned to course learning outcomes related to beam extension, compression, and deflection. Preliminary assessment will use rubrics measuring students’ depth of analysis, identification of conceptual errors, and clarity of reasoning. Student performance and perceptions will be compared with those in traditional assessment formats where students independently solve assigned problems. Anticipated outcomes include improved diagnostic reasoning, deeper conceptual retention, AI readiness skill development, and increased awareness of AI tool limitations.

This work-in-progress will report on the design of the assessment framework, AI prompt structure for generating controlled correctness levels, and preliminary findings from pilot implementation. The results are expected to inform future development of AI-integrated pedagogical strategies that encourage critical engagement rather than passive use of generative tools in mechanics education.