Generative Ai for CAD

Feeling: Inspired. This past Tuesday I attended a Next-Gen Ai Design Techniques seminar facilitated by Dorian Gorski with leading Ai design guru Greg Aper It has really had my head spinning this week. I would like to comment directly what I took away from the seminar soon. Though I couldn't help putting a couple of my "future of Ai" ideas down in Figma. I expect to see more robust integration of Ai in the designer's work flow, so I took a stab at visualizing what could (and likely will) be.

Generative Ai for CAD is in its infancy, but soon I imagine the ability to generate parametric geometry from imagery and materials databases from text/image prompts. Google’s DreamFusion, Nvidia’s Get3D, and OpenAI’s Point-E do some of this now, and Vizcom's Sketch to 3D is coming soon, but we haven't seen an integrated tool that builds editable parametric, uv mappable geometry. I don't think we will have to wait very long though.

The tools don't even need to be native to a certain CAD or rendering platform. The below could be standalone generators that output our preferred files types. Imagine being able to snap a picture of an object with your phone and have a simplified object generated that you can modify and add to an assembly. Then when it is time to explore colorways and visualize, find real-world material samples to prompt the creation of digital materials that can be tweaked to your needs. What else would be a good use of Ai in design that would save time, yet leave the creative keys in the hands of the designer?