4 Ways AI Will Change Design and Manufacturing
Although technology has taken over many aspects of our lives, our product design and manufacturing processes are still largely stuck in the industrial age. Companies struggle to efficiently create better-performing products and keep costs low. After extensive experimentation, they arrive at the best designs they can. Then, they feed instructions into manufacturing machines, which churn out thousands of identical products or parts, leaving little room for customization.
All of this is about to change. We’re on the cusp of a revolution in designing and manufacturing products. In my “AI for Computational Design and Manufacturing” course at MIT, I work with business professionals to help them understand how artificial intelligence (AI) and machine learning (ML) will soon affect products, as well as how they can embrace these changes within their own companies.
Specifically, here are four ways AI and ML will alter product design and manufacturing:
1. Optimizing several variables. Product designers generally have a good sense of what the results they will get with different materials. But when designers must balance several desired outcomes, things can quickly get complicated. For instance, when designing a car, designers try to optimize not only performance, but also cost, durability, safety and fuel efficiency. By leaning on AI and ML tools, design teams can rapidly iterate through thousands or even millions of different potential designs, and then spend their scarce time focusing only on those the algorithms have identified as having the most potential.
The word “design,” in this instance, typically refers to performance design, rather than aesthetic design. While humans are still better than computers at creating beautiful products with consumer appeal, AI and ML can calculate how minor changes to a product will influence several different aspects of performance. This will be an invaluable improvement for design teams, as it will let engineers spend time on more creative aspects of their jobs rather than spending countless hours of laborious and inefficient trial-and-error experimentation. Plus, it will lead to better products.
2. Unprecedented customization. Product customization calls for lots of manual labor. Even fairly standard products, such as athletic shoes, typically require assembly lines with dozens of human workers. But AI and ML will soon open the door to much more automated product customization.
For example, in keeping with the sneaker example, emerging technologies will let each pair of athletic shoes be fully customized, improving the shoe’s performance for an individual athlete. Shoe buyers will soon use new input devices, such as sensors that create pressure maps of their feet and capture information that will lead to uniquely tailored designs. Then based on high-level specifications, generative design tools will automatically synthesize designs and convert them into machine-readable assembly instructions.
Recent advances in AI and computation have already set us on the road to a whole new world where each product is one-of-a-kind with unprecedented levels of complexity.
3. Automated experimentation. For many products, it can be difficult—even impossible—to predict performance without first carrying out experiments. For instance, there aren’t any numerical models that help product designers determine how effective a given drug will be in alleviating a patient’s symptoms, or how efficient a solar cell will be at generating electricity.
Although AI and ML do not eliminate the need for experimentation, they can help researchers efficiently plan and even conduct experiments. In the near future, we will see completely automated workflows in which designers set parameters for desired outcomes, and then robots conduct experiments and assess the results.
4. Smart manufacturing. Today, most manufacturing systems are extremely “dumb.” Manufacturing equipment might be able to turn out standardized products at a consistent rate, but it cannot assess and react to changing conditions. However, adding sensors to manufacturing facilities, along with layering AI and ML algorithms onto equipment, will let companies use smart manufacturing processes that are much more dynamic, responsive and resilient.
For example, imagine that the temperature in a manufacturing plant spikes overnight, or that a machine is fed a batch of materials with slightly different properties from standard materials. Without sensors and smart systems, machines will simply continue to operate as normal and not take variations in the environment or materials into account. This can lead to delays, machine degradation and ruined products.
By contrast, smart manufacturing systems detect when something is off and automatically adjust to changing conditions. This, in turn, can improve quality control, as well as reduce costs and increase reliability.
It’s likely we can’t even imagine some of the most powerful ways AI and ML will change product design and manufacturing. (After all, many of the ways we use smartphones were completely unforeseen just a decade ago.) But by learning how to use these technologies in their operations, business and IT leaders can position themselves at the leading edge of their industries and ensure they are as ready as they can be for whatever the coming years may bring.
Wojciech Matusik is a professor of electrical engineering and computer science in the Mechanical Engineering Department and the Computer Science and Artificial Intelligence Laboratory at MIT, where he also leads the Computational Fabrication Group.