Artificial intelligence has been used in design and manufacturing for some time, at least since the deep learning revolution of the early 2010s.
However, designers, engineers, and architects are discovering that the power of newly emerging generative AI – like the core models that power ChatGPT – is potentially even more transformative.
The principle is simple. With a natural language interface, designers can simply explain what they want to create, what materials they have, and how it should work.
Then, just as ChatGPT finds the best way to put words together to form sentences, it finds the best way to bring your ideas to life.
This isn’t just an exciting vision of how designers will work in the near future: it’s already being implemented today.
So here’s a look at how generative AI is changing the face of design, covering what we’ve seen so far and some ideas for how it will evolve in the future.
How is generative AI used in design?
One of the main use cases is for the design of physical objects, components or products. Factors such as material efficiency and production speed can be optimized by AIs that have learned the qualities of different materials as well as the pros and cons of each manufacturing process.
Siemens, for example, envision a future where a manufacturing technician can get instant answers to questions about how changes to manufacturing processes impact the finished product.
Of course, there are problems caused by hallucination – the tendency of the current generation of generative AI models to confidently make mistakes.
Errors in the design or manufacturing process of objects could lead to dangerously defective products. It is fair to say that human skills are currently needed to mitigate this problem, not only in design, but in all areas where generative AI is used. This is one of the reasons why I believe human jobs will be changed by AI rather than replaced by it in the near future.
Also in the field of graphic design, generative AI is quickly proving useful. Nutella used algorithms to create millions of different products unique packaging designs for its Nutella Unica range. Each jar is stamped with a code, allowing it to be identified by collectors of unique packaging art.
And it also has potentially transformative applications in architecture and urban planning. According to a McKinsey Report on the impact of generative AI, it is used to design buildings, shape urban landscapes and augment the skills of human designers to integrate their work with the natural environment.
Architects can use it to manipulate things like room layouts and features like stairwells and facades. Then they can simply set the parameters they need and let the AI create multiple prototypes and candidates that match their expectations.
It also proves to be a useful tool in fashion design. Hong Kong-based computer scientist Calvin Wong created an AI fashion design assistant which generates outfits in seconds based on specific requirements and information on available materials. Aida – the AI-powered design assistant – creates finalized design images for clothing in around ten seconds. The process typically takes weeks for human designers.
One company confidently betting on generative AI as the future of design and manufacturing is longtime provider of industry-standard tools, Autodesk.
I recently spoke to Autodesk’s head of research, Mike Haley. He told me that AI has played a role in generative design for some time, but has traditionally posed challenges, particularly around the enormous amount of computational resources required.
The development of cloud-based generative AI platforms, however, has removed these barriers, allowing more creatives to understand the benefits for themselves.
Haley said: “I’m very optimistic that we’re now seeing the emergence of these kinds of tools that can take this information from the real world, help us reason about it, and produce better designs for the world. . »
The future of generative design?
It is clear to me that generative AI has great potential to increase capabilities and streamline design work processes.
But there are also challenges. One of the most important will be balancing the continued need for true human creativity with the desire to create efficiencies through automation. Sure, AI can create 100 designs per second, but can we be sure that they all express the creative and technical talent that companies want to present to their customers?
And, of course, there are ethical concerns around data ownership, authorship, and intellectual property rights that still need to be resolved.
But overall, I see a future in which the tools available to designers and creators in all fields continue to become more intuitive and more useful in helping them with tasks involving human creativity.
In the future, I think we’ll see tools that can do more than create to order, and more capable of understanding and anticipating the nuanced needs of individual designers and brands.
They will also be more geared towards large-scale customization. It could even begin to upend long-held expectations of uniformity and the cookie-cutter approach we have toward mass-produced goods and products. Which certainly has the potential to make the world a more interesting place.