EXPLORING THE POTENTIALS AND LIMITATIONS OF AI-BASED CAD DESIGN FOR MECHANICAL ENGINEERING

This paper investigates the integration of large language models (LLMs), specifically ChatGPT, with CAD software to support the generation of 3D models for mechanical engineering applications.A literature review of recent AI-to-CAD research identifies current capabilities and gaps regarding CAD integration, parametric modeling, and adherence to engineering norms.A proof-of-concept case study demonstrates how ChatGPT can generate parametric Python scripts for Fusion 360 across four mechanical test cases with varying complexity.The results show that AI-generated models can reproduce simple standardized components, support parametric design, and aid in educational settings by prompting users to explicitly define design intent.However, challenges persist with spatial reasoning, complex geometry, and integration depth.The study concludes that while current LLMs offer value for design automation and CAD education, significant development is needed to support complex and norm-driven engineering design workflows.