October 25, 2023

Photo from google

Since the emergence of OpenAI's ChatGPT, using natural language to describe and generate programming language or algorithmic drawings has become a hot topic in the tech industry. However, applying natural language for "Computer-Aided Design" (CAD) in engineering and product design is a challenging endeavor. At a recent (2023) user conference, Manish Kumar, CEO of the 3D CAD brand SolidWorks, acknowledged that they are indeed working on developing such technology. He also mentioned that there are still many challenges to overcome in the development stages.

Currently, Generative Design is a computer-aided system that integrates AI algorithms. In this system, designers input specific details of their requirements, and the algorithm generates various design solutions that meet the specified criteria. This process is expected to significantly reduce the time required for design. However, compared to software like Midjourney, the key difference with current Generative Design is that designers still need to have a concrete understanding of the characteristics of various materials and physical constraints. Current AI drawing software, on the other hand, can achieve desired results using natural language, highlighting a clear distinction between the two approaches.

Manish Kumar proposed that before a fully developed ChatGPT in the CAD field, Generative Design still needs to go through three stages.

In the first stage, AI models based on parameters provided by designers and optimizes materials and structural content within set conditions. Currently, some products in the industry are already operating with technology at this stage, such as Ansys' Discovery 3D from the CAE brand.

In the second stage, AI is capable of generating and adjusting different designs based on design intent. For example, a designer could request AI to provide mouse designs that adhere to ergonomic characteristics and save materials, then choose from the numerous design versions suggested by AI. According to Manish Kumar, the "reverse design analysis" based on visual imagery also falls within the technological capabilities of this stage.

The third and final stage involves designers further requesting AI to incorporate existing mass-produced components into the design considerations, ensuring that the final design is more aligned with the requirements of commercial production.

Indeed, whether it's ChatGPT or Midjourney, the essence of such generative AI applications is to allow users to express their requests to the software in natural language, then iteratively adjust and refine until achieving the desired textual or visual content. As AI technology continues to advance, it is anticipated that similar AI applications will emerge, enabling "non-expert" users in product design to interact with AI in a continuous feedback loop, refining their requests until they obtain industrial design outcomes that meet their satisfaction.

—

Reposted this article from：Amag（原文標題：向AI提需求就能產出設計？SoidWorks CEO透露正開發設計界ChatGPT）

#ai #productdesign #productdevelopment #solidworks #chatgpt