Design is both an art and a science. Form and function must exist in tandem. So, when it comes to the adoption of artificial intelligence, fields like fashion, architecture and interiors have found themselves at a crossroads.
For those who view design as a largely creative endeavor, AI appears to undermine the intrinsic humanity underpinning their work. But for those looking to improve efficiency — whether it’s in idea generation, reducing waste or streamlining manufacturing — the technology can help to enhance the creative process.
Regardless of where you stand on whether (or how much) designers should embrace generative AI, this much is clear: it is having a seismic impact on the built environment, the products on our shelves and the clothes in our closet.
Below, three of the most prominent early adopters of AI-driven design, from across the creative industries, discuss how the technology is transforming their practices:
Philippe Starck, industrial designer
In 2019, my “AI chair” became the first mass-produced piece of industrial furniture ever created with the help of artificial intelligence. I launched the series alongside an Italian furniture company, Kartell, at Milan Design Week and it went on sale that year.
Generative design was in its infancy then, but the chair’s story goes back a few years further. In 2017, I had been growing a little bored with myself and my creations. I wanted to explore new territories, devoid of human, cultural or sentimental influences, which is why I decided to work with AI.
I started to collaborate with Autodesk, one of the earliest generative design software companies. They allocated me a dedicated computer, and I asked it a simple question: “Can you help me rest my body, using minimum matter and energy?”
The AI had to understand what a human body was — and it knew nothing about my culture, my origin, my desires or my gender. The machine struggled for more than two years, going back and forth with me and Autodesk’s engineers on the designs. It was stuck.
The solution came to me, in a roundabout way, when my great friend Henri Seydoux — the CEO of French drone firm Parrot — explained to me that the fingers of a fetus’ hand in its mother’s womb are not formed by adding matter but by removing it. (The hand is initially a paddle-like shape, and the fingers are only revealed when cells in the areas between the digits die off). I told the machine about this via the engineers, to help reorientate the AI’s thinking, and within minutes, it produced a chair.
In those days, it took three years for the machine to come up with a chair. Today, it would probably take a few seconds.
Artificial intelligence is, by definition, artificial. It doesn’t exist; it is speed. AI is a fantastic tool, because it can save time, it can connect the dots. It does the fastest calculation of a tremendous mass of information. It delivers a cocktail.
The technology has already changed the way we work. We see its amazing power every day. But it is a power that doesn’t yet have a heart or a brain. What we do, as designers, is talk to AI, embed AI in technology and control AI. It is incredibly intelligent and increasingly autonomous, but it remains rooted in what we program — in what we choose for it. We still need that little magic lying inside the human brain. And when I sketch with my pencil and pad, I am still faster than AI.
I have doubts about the level of creativity it can ever produce or replicate. There is magic and poetry in real creativity, which I think is still the prerogative of the human subconscious. AI is not yet the great creator that we are. But tomorrow, who knows.
For now, while AI is revolutionary, it is a means to an end, not an end in itself. It is another tool. When we invented the telephone, it was not the telephone that mattered, but the fact we could talk to each other. So, I would advise designers to beware of the pitfalls (it is a minefield) and, above all, stick to your own vision, develop your subconscious and believe in your intuition and your ideas. Then — and only then — should you let AI help you a little. At least, for now.
Norma Kamali, fashion designer
I’ve always been very curious about technology. In the 1960s I got a job at Northwest Orient Airlines and found myself sitting at a UNIVAC computer all day. It would give me information about flights, arrivals, issues with the aircraft. From then on, I felt so comfortable with computers. I just thought, “Oh, I get this.”
A few years ago, I was at a meeting in Abu Dhabi where someone pitched an e-commerce site that was run, designed and merchandised by AI. A scientist at this meeting asked if — because I was a designer — they could “download my brain.” I thought it was such a great idea that I should do it myself.
I have a fashion archive, spanning my 57-year career, which is incredibly valuable to my label. I thought, “What if my experience and vision could be part of an AI experience? What if my archive could extend the life of the brand after I’m gone?” When Karl Lagerfeld was hired for Chanel, he religiously repurposed the label’s archive. He brought it back to life. I thought AI could do that for me.
I took a course at the Massachusetts Institute of Technology on AI, generating content and writing really great prompts. It went way over my head. There were 150 students, all way smarter than me, and I hoped none of them knew enough about fashion to recognize me. But I’m grateful I hung on. By the time I finished, I was so excited by the opportunities. It was like a whole world opened up.
Now, I have an AI program that has categorized my archive, using tagged descriptions based on silhouettes, necklines and other design elements. When I write a prompt, AI searches my archive for similar garments and responds with design ideas that are consistent with my previous work.
For example, I recently used AI to recreate a new version of a popular semi-sheer black and beige striped dress from my brand. I was getting tired of it, because I’d already created so many iterations. I had already done it as a mini-dress, a long gown, a jumpsuit. But when I saw Jessica Biel photographed in the dress last summer, I thought “Oh God, I have to do it again.” She just looked amazing.
But I really could not think of what else to do. So, I pulled a picture of the original design and gave AI a prompt, asking for variations. AI reconfigured the stripe pattern, making some thicker and moving the bands around. Rather than the striped skirt, AI suggested a sheer beige panel with a slit up the side. In one design, it added a collar. I picked a few of my favorites, and each one looked like I’d created it. We put a collection — some designed by me, some by AI — up on our website, letting customers know which was which, but it didn’t make a difference. They sold at the same rates.
That is the difference with AI. I was emotionally connected to that picture of Jessica — I don’t think the dress had looked better on anybody, ever. It would have been difficult to tear up the design and change it. But AI could go anywhere with it and come up with something new.
Although I agree with a lot of the anti-AI rhetoric — that it will take jobs and replace human actions — I also strongly believe that it’s going to create new jobs, a new world and a new way to live, too. Learning as much as you can about AI, to see how your creativity comes to life in a new way, is really a good thing to do.
I still sketch, I still make patterns, I still use real-life models. All of that is part of the process. But I can’t imagine not enjoying this new opportunity. Why would you pass that up?
Tim Fu, architectural designer
People think about AI in a very shallow way. But in architecture, it can assist with everything from mood boards and rendering to exploring different floor plan configurations.
Let’s say we’re designing a backyard layout. We would present AI with the main objectives, visual references and some reasonable building materials. The AI then provides us with tons of the best solutions, which we critically analyze and curate. We still need to connect the door to the porch and so on — so the overarching goals of the architect remain intact.
At my firm, Tim Fu Studio, we’ve developed AI toolsets for architectural and spatial optimization. We outline factors that affect our designs — such as maximizing square footage or maximizing sunlight — and let machine intelligence explore different configurations based on those parameters. Evolutionary algorithms keep testing options until they get better and better, then we can cherry-pick the best ones.
We recently unveiled designs for a new resort on Lake Bled, Slovenia, featuring seven luxury villas. We believe this is the world’s first fully AI-driven architectural project.
We used our AI tools for the floor plans, maximizing views of the lake in each property, for instance. But we also instructed a diffusion model to analyze an image database of the region’s architecture to generate designs ideas in harmony with the local heritage and style. The resulting outputs helped us reimagine traditional Slovenian design motifs, such as timber “rizalit” — vertical enclosures often found projecting from the facades of the region’s houses — which we have reinterpreted as glass-fronted atriums flooding the villas with light.
What we really mean by “style,” visually speaking, is correlation. AI can do that: Diffusion works by picking up the correlation between pixels, which is essentially the same way humans see patterns.
We treat AI like a calculator sometimes. We just let it process because it does it so much faster than we can. But architects are slow to adapt to anything they feel threatens their creative agency. It’s very difficult to lay down your ego and realize, ‘Hey, there could be aspects of my job that machines do better.’ Our company’s ethos is to remain humble in the face of technology. AI expedites everything we do. Our delivery is faster; our quality is better. But technology will only assist, not replace, architects. When hiring for the studio, we not only test candidates’ AI-prompting and problem-solving abilities, but also their sketching. We sketch on paper in the office every single day.
Like everything revolutionary, people are both reluctant to use AI and actively outspoken against it — especially those who don’t really understand it. People also have a lot of misconceptions about AI because there’s so much slop on the internet. Some architecture offices even ban it completely out of fear, which is great news for us. It means our competitors aren’t jumping on the train.
Architecture’s end result is the places we make and the experiences we build, not just the renders we produce. The biggest milestone will be when, instead of arguing for the use of this technology, we just let people walk through a space, to experience it first-hand without any knowledge of AI’s involvement. We’re still waiting for a lot of our bigger AI-driven projects to be announced — some of them in the Middle East, including complete urban master plans — and it’s exciting.
As told to CNN’s Oscar Holland and Leah Dolan. The interviews have been edited for length and clarity.