Design automation has been an ultimate goal, which is to develop and create designs or design concepts that are superior to products currently available or are not available yet. However, the early-stage system design for concept generation and solution ideation is still dependent on heuristics, experience, and intuition, resulting costly trials more often than not. The outcome from this research thrust, AI for Systems Architecting, can transform the current practice of systems design to a more informed one with the recently developed AI tools. By incorporating big data and knowledge of past designs into large language models (LLMs) along with design ontology and principles, we demonstrated domain specific GPT (Generative Pretrained Transformer) is feasible. Design specific GPT will provide design advice and know-hows for industrial problem solving. This will accomplish the concept of Virtual Design Assistant (VDA) platform which complements human’s creativity with the information processing power of AI. This will benefit to most US manufacturing industry to achieve true smart manufacturing and to bring the manufacturing leadership back.
Kim, Sang-Gook Sang Min Yoon, Sang, Yang, Maria, Choi, Jungwoo and Akay, Haluk and Burnell, Edward. (2019). AI for design: Virtual design assistant. CIRP Annals. 68 (1) 141
