"In 2022, design researchers at Toyota Research Institute in California began experimenting in vehicle design with Stable Diffusion, a generative-AI text-to-image model that produces photorealistic results. The AI is capable of working broad and often imprecise cues - sleek, masculine coupe, for example - to get human designers to the rough sketch (of a vehicle) stage," Neil wrote in a recent Wall Street Journal article.

Since 2022, "TRI's team has been teaching the AI to analyze the imagery it sees in terms of aerodynamic efficiency," Neil wrote.

"The next steps are to understand what other engineering quantities can be inferred; and to figure out a way to incorporate all of Toyota's design resources and engineering archives," Neil wrote, quoting Avinash Balachandran, director of TRI's Human Interactive Driving division.

"Our work is only to amplify that of the designers and engineers, not to replace them," Balachandran said.

Read the full article here.