Artificial intelligence can now aid in the design of more visually appealing cars, according to a recent paper co-authored by John R. Hauser, a professor of marketing at MIT Sloan, and professors Alex Burnap from Yale School of Management and Artem Timoshenko from Kellogg School of Management. In the automotive industry, the aesthetic appeal of a product is known to impact roughly 60% of purchasing decisions. Car manufacturers, who spend over $1 billion on average to design a new model and up to $3 billion on major redesigns, are now using machine learning models to generate new and aesthetically pleasing designs that can be quickly evaluated.
Hauser and his co-authors developed two models in partnership with General Motors. The first is a generative model that creates new car designs based on prompts from designers about viewpoints, colors, body type, and image. The second is a predictive model that forecasts how consumers will rate designs in terms of aesthetic appeal or innovativeness. The models are not intended to replace human designers, but rather to augment the design process.
Carmakers have traditionally relied on theme clinics to determine the aesthetic appeal of their designs, where they bring targeted consumers to a single location to judge designs. Theme clinics can cost $100,000 each, and carmakers need to hold hundreds each year to ensure they select the right designs for production. Using predictive modeling, carmakers can weed out designs that are unlikely to score well on aesthetics, thereby shortening development timelines and reducing costs.
A vehicle that highlights the importance of aesthetics in the industry is the Pontiac Aztek, a crossover SUV released by GM in 2000. Although it had high customer satisfaction scores for its multiple outdoor features, the Aztek’s aggressive and “in your face” design resulted in poor sales, making it one of the ugliest cars of all time. On the other hand, the Buick Enclave, which was redesigned and re-released as a more attractive vehicle, sold 30% higher than the Aztek.
Hauser points out that when two cars are equally reliable and effective, consumers will buy the one that is more attractive. By using AI to generate new designs and predict consumer preferences, carmakers can improve the aesthetic appeal of their models and increase sales. The generative model can also be used in other industries where aesthetics play a prominent role, such as furniture design and fashion.
The paper shows that AI can not only predict the appeal of new aesthetic designs but also generate designs that are aesthetically pleasing or innovative. The deep neural network predictive model achieved a 43.5% improvement over the baseline and over more conventional machine learning models. The generative model produced images that were deemed aesthetically appealing by consumers and even suggested designs that were later introduced to the marketplace.
Hauser emphasizes that the models are a tool for designers to get new ideas and try them out. With machine learning, designers can quickly evaluate new designs, reducing development timelines and costs. The AI models may offer a new way for carmakers to stay ahead of the competition by creating more attractive models that resonate with consumers.