Integrating AI Technologies in Product Design: A Market Analysis and Conceptual Application for CADdrive Software

Abstract

This thesis explores the integration of generative AI technologies into early-stage product design, focusing on their reliability and stability within a CAD-based environment. Four AI models—ChatGPT, GitHub Copilot, Gemini, and DeepSeek—were evaluated through a multi-method approach combining repeated prompt testing and expert surveys. Quantitative analyses using cosine and Jaccard similarity scores were complemented by qualitative evaluations from twelve design and engineering professionals. The results indicate that ChatGPT and Gemini provided the most stable and contextually relevant design suggestions, particularly in structural domains such as material selection and glass specification. In contrast, GitHub Copilot, while demonstrating creative potential, occasionally deviated from technical standards. DeepSeek delivered consistent but less detailed outputs. The findings suggest that generative AI tools can enhance ideation, decision-making, and educational support in CAD platforms like CADdrive, provided their integration is managed with expert oversight and domain-specific evaluation criteria. This study contributes to responsible AI adoption in product design and engineering education by proposing a practical framework for assessing AI-generated outputs in iterative and collaborative design tasks.

Date of Award	2025
Original language	English
Supervisor	Christian Zehetner (Supervisor)