A new study has revealed how machine learning models can be used to optimize the spending of car manufacturers, allowing them to reduce their costs by as much as $3 billion on a single design and also lessen the number of failed products that make it to market.
Consumers tend to go for items that appear attractive; this is particularly true in the car industry, where product aesthetics are thought to play a role in around 60% of buying decisions. This trend can be seen from real estate’s curb appeal to smartphones with smooth edges.
Professor says:
“People buy cars based on aesthetics. Styling can make a difference.”
A professor of marketing at the Sloan School of the Massachusetts Institute of Technology has reported that designing a car model typically costs more than one billion dollars, while major redesigns can be as expensive as three billion.
Hauser’s recent co-authored paper has revealed that machine learning models can foresee the attractiveness of novel aesthetic designs and fabricate attractive or imaginative designs. Furthermore, these models can be run on an ordinary office laptop after they have been trained.
Alex Burnap, a professor at the Yale School of Management, and Artem Timoshenko, a professor at the Kellogg School of Management, collaborated on the paper.
Hauser says:
“The models are a tool for designers to get new ideas and try them out.”
“They are capable of generating new images that are highly aesthetically pleasing and that can be evaluated quickly.”
Low Demand For Unappealing Cars
In the automotive industry, product aesthetics are a major factor in purchasing decisions, as evidenced by the notorious reputation of the Pontiac Aztek.
In 2000, General Motors designed the Aztek, a crossover SUV on the same platform as the Buick Rendezvous. It had plenty of features attractive to fans of outdoor activities and was generally well-received by customers — with one exception: its exterior styling.
The Aztek was not a success. It had an overly bold design that wasn’t appreciated by many and became known as one of the most unattractive cars ever made. Consequently, General Motors discontinued production in 2005.
Approximately half the sales of the Aztek were not enough to keep it in production; it was subsequently redesigned and rebranded as the Buick Enclave. The Enclave has been a consistent success, with a 30% higher manufacturer’s suggested retail price than its predecessor. It has now been over 15 years since the launch of this model, and it is still being manufactured today.
Hauser commented that the Aztek provides a meaningful lesson.
Hauser continues to say:
“If two cars are equally reliable and effective, consumers will buy the one that’s more attractive.”
Using AI To Design Aesthetically Pleasing Car Models
Today’s car manufacturers are investing heavily to prevent the launch of a successor to the Aztek. Historically, this undertaking has relied on theme clinics which involve getting hundreds of potential purchasers together at one place to evaluate designs.
The price tag for a theme clinic can be as much as $100,000, and car manufacturers have to invest in many of them annually to guarantee that the right models are created and released.
Predictive modeling provides an attractive option for car manufacturers: they can identify designs that are unlikely to be well-received in terms of aesthetics and save time by not progressing them further.
Testing fewer designs in theme clinics will result in shorter development timelines and lower costs.
Hauser and his colleagues designed two models while collaborating with GM as their research partner.
– A generative model which produces novel car designs in response to commands from designers regarding perspective, color, form, and image.
– A predictive model that anticipates how customers evaluate designs regarding their aesthetic charm or originality.
The predictive model, constructed on a deep neural network, was the starting point of the research. It showed remarkable outcomes compared to the baseline, with a 43.5% enhancement, and even outdid more traditional machine learning models.
Hauser continues to say:
“Our model was able to indicate the designs that were good and the designs that were bad,”
“But as we got more and more into the process, we realized the real leverage was in creating new designs.”
The generative model created visuals that shoppers found attractive and even created designs that were eventually sold. Additionally, the researchers discovered that this method could be applied to items other than vehicles.
Transform Your Design Process with Augmented Experiences
Despite the success of artificial intelligence in various applications, it is not intended to replace human designers. According to Hauser, a professional designer must set the parameters before the generative model can produce designs.
Designers work through a repetitive and staggered process while creating automotive designs. From the generation of the concept to testing, evaluating, and reworking, thousands of decisions are made before resulting in the final product. Consumers and critics then rate this product using words like sporty, luxurious, or rugged.
Hauser and his colleagues consider artificial intelligence to enhance the design process, similar to how computer-aided modeling is used in furniture design, fashion, and other areas where appearance is essential.
Hauser went on to say:
“There are a number of different ways you can cut a dress.”
“A machine learning model can give designers ideas about what customers will think is aesthetically pleasing, but a designer isn’t going to produce exactly what the machine puts out.”
Artificial intelligence (AI) is a game-changer in the automotive industry, enabling designers to create more appealing and attractive cars. By using AI tools to analyze customer preferences, market trends, and design principles, car designers can gain valuable insights into what features and styles will most likely resonate with consumers.
This can lead to the creation of cars that not only look great but also meet the needs and desires of customers. However, it’s important to note that AI is just one tool in the designer’s toolkit and should be used with other methods and techniques to create truly exceptional cars.
Source: MIT Sloan
