2026 ASEE Annual Conference & Exposition

Recent advancements in large language model (LLM)-based artificial intelligence have significantly expanded its capabilities in solving complex engineering problems. To explore its potential as a design aid, students in a senior-level Machine Design course at a medium-sized private institution were tasked with evaluating the effectiveness of AI tools in a mechanical design project.

The project required student teams (up to four members each) to design a reverted compound gear train that met specific gear ratio and power transmission requirements. Students were encouraged to use AI tools—such as LLM-based assistants—for calculations, design exploration, and iterative refinement. They were free to select any AI platform and were required to document their prompt strategies, refinements, and validation methods for AI-generated outputs.

The design process tasks included:
• Selection of catalog gears
• Calculation of gear forces
• Preliminary shaft geometry definition
• Determination of bearing reaction forces
• Construction of shear and moment diagrams
• Selection of shaft materials and diameters
• Specification of keys for power transmission
• Selection of appropriate bearings

The project was divided into two phases. In Phase 1, students completed gear selection, gear force calculations, selecting where components would be on the shafts, and shaft shear/moment diagram development. Upon completion, the instructor provided a MATLAB script to compute reaction forces and generate shaft diagrams (shear, moment, angle, and deflection), ensuring students used correct values for Phase 2.

Following Institutional Review Board (IRB) approval, students were surveyed regarding their experience using AI in the design process. The survey assessed perceived effectiveness across various tasks and overall satisfaction. Results indicated that AI tools were particularly helpful for basic calculations and conceptual guidance. However, students noted limitations in AI’s spatial reasoning capabilities, especially in tasks involving shaft layout and component placement.

This study highlights both the promise and current limitations of AI as a design assistant in mechanical engineering education. It provides a foundation for future integration of AI tools into design pedagogy and encourages critical evaluation of AI-generated content in engineering workflows