In January, what had previously only been pixels made a real-world splash on the catwalks of Paris Fashion Week.
The models’ futuristic-looking clothes, designed in a collaboration between fashion house Acne Studios and artist and programmer Robbie Barrat, were designed by an artificial intelligence (AI).
‘When you design a collection, you have an idea of what a jacket looks like, or a pair of trousers,’ says Jonny Johansson, creative director of Acne Studios.
‘The computer doesn’t know what a jacket is. It tries to learn from the images we gave it, and then creates its own idea. It was freeing, because we could then design a jacket that’s like from a parallel universe.’
Barrat is well known in the world of AI, or generative, art. Two years ago, code he devised was used to create an artwork that sold at Christie’s for £337,000 – the first of its kind.
Acne Studios used designs generated by a GAN to inform their real-world clothes. (Photo: Twitter/@videodrome)Since then, Barrat’s method, a form of machine learning called a Generative Adversarial Network (GAN), has been used in a range of creative pursuits, from helping write albums to automating CGI.
But high fashion design is just a small part of the total fashion industry, a trade spanning supply chains to style suggestion to large scale manufacturing.
AI, and its subset of neural network-powered machine learning, is set to transform how we wear and consume clothes and in the process, might even help improve the planet.
Though Acne Studios’ futuristic clothes were the first examples of AI on the catwalk, startups have been playing with GAN-generated clothes for far longer.
What is a Generative Adverserial Network?
Machine learning is a kind of artificial intelligence that uses vast quantities of data to predict, or learn about, certain outcomes, without being preprogrammed.
A neural network is a series of algorithms that works a bit like the human brain. They’re used in machine learning to spot patterns by comparing a template to new data.
A generative adverserial network (GAN) is a kind of machine learning system, where one neural network trains another. It typically consists of a “generator” network, which creates random versions of an image, for example, and a second “discriminator” network, which judges how accurate the generated images are, according to its own criteria.
After students at MIT experimented, with the help of IBM, to produce an artificial intelligence’s idea of what Tommy Hilfiger should look like, they realised the applications could be far larger.
Last year, two graduates of the same course launched an AI-powered brand called Coven.ai, producing unconventional, unique designs powered by the same GAN-techniques that Barrat had used to design Acne Studios’ catwalk show.
Another fashion startup has used Google’s generative AI to churn out t-shirts made from doodle data.
On a larger scale, Indian fashion business Myntra used their in-house AI-powered software Ratatouille to suggest new clothing designs, cutting the design process from six months to one.
‘It throws up questions around what the relationship is between designer and neural network,’ says Matthew Drinkwater, head of London College of Fashion’s Innovation Agency.
‘Is it the computer designing, or do you use it as a collaborative tool?’
Amazon’s Echo Look was the first mass market device to use artificial intelligence to suggest style ideas. (Photo: Getty)It’s unsettling to think your clothes may have been designed by something not human, but it’s far from the only aspect of the fashion industry to be affected by algorithms.
When the Amazon Echo Look was announced three years ago, with its ability to algorithmically decide between outfits you wear, many observers envisioned a new era of computer-chosen style choices.
It wasn’t inconceivable to imagine a world in which an AI-powered recommendation advice, like the Look, suggested a piece from an AI-designed clothing range, like Acne Studios’ new offering.
Amazon’s drones would then carry out the delivery, and the only human involved in the process will be the postman carrying it to your door.
Even the models that showcased those AI-generated clothes were threatened with becoming obsolete—virtual influencers like Miquela attract millions of followers and tens of millions of pounds.
Lil Miquela is a digitally created avatar that models clothes on Instagram, with more than a million followersBut in reality, the uptake by the fashion industry at large has been much slower.
‘There’s a huge universe of [AI] categories that are being created at the moment, and the fashion industry is the last one to adapt to it,’ says Hasna Kourda, CEO and founder of Save Your Wardrobe.
‘The industry has been very skeptical about the digital world as a whole – not just AI,’ Kourda tells Metro.co.uk.
‘It’s taken them decades to just have a website, an e-commerce or even a landing page.’
However, the pace of take-up for AI and machine learning-powered tech has begun to increase.
‘We crawl the web for fashion photography very much like Google does for information,’ says Kostas Koukoravas, CEO and founder of Intellistyle.
‘We focus specifically on fashion imagery, looking at images put together by human stylists on social media, in catalogues and retailers’ websites, and then use that imagery to analyze and understand what style means.’
Intellistyle uses the masses of fashion based-imagery online and uses it to catergorise and suggest new combinations. (Photo: Intellistyle)Concepts of style and what’s considered ‘in fashion’ aren’t just human anymore.
Instagram likes, Twitter hashtags, and Google-distributed display advertising are just as, if not more, influential on what people buy than a magazine editor or fashion designer.
So if taste has always been arbitrary – often decided by royalty or an cadre of magazine editors – what difference would it make if it was decided, and designed, by a machine instead?
‘The AI’s understanding of customers’ styling preferences quite often outperforms human stylists,’ Koukoravas tells Metro.co.uk.
‘We were at London Fashion Week last year and pitted our AI against the choices of human stylists and asked the audience to select which one they like best.’
‘About 70% of the audience preferred combinations that the AI provided.’
But if you’re preparing yourself for getting your next clothing recommendation from a machine, then you might want to think again.
Your most likely interaction might not be with the machine itself, but with a person assisted by the AI.
Intellistyle claims that its suggestions are better than human-powered suggestions. (Photo: Intellistyle)‘‘When you have a human recommending something to a customer, it has much better results,’ says Kourakas.
‘AI still doesn’t have the empathy or the communication abilities that humans have.’
Better picked clothes could help the fast fashion industry be more sustainable, too.
A study by Marks and Spencer found that adults in the UK only regularly wear 44 percent of the clothing they own.
In theory, better chosen clothes and suggestions on how to wear them might reduce the 3.6 billion unworn clothes left in the nation’s wardrobes.
‘We’re looking at it from the perspective of “Can AI help you make the right decisions”’, says Kourvakas.
‘As opposed to going out and buying something mindless and coming back home and thinking “Hey, I haven’t got anything to match it with.”’
As Christopher Wylie, the whistleblower turned H&M data scientist, told a fashion conference last year, ‘producing less will get you more.’
In an industry where retailers regularly destroy millions of pounds of worth of unused stock or send the mostly fossil fuel-built clothes to landfill, it’s a huge problem to solve.
‘You walk into stores on the high street across the country and typically they’ll have identical products,’ says Matthew Drinkwater, head of London College of Fashion’s Innovation Agency.
‘But it doesn’t make sense that consumers in one location, let’s say in the West End of London, would be buying the same things as perhaps on the south coast or in Cornwall.’
Products like Intellistyle and Save Your Wardrobe’s AI-powered suggestive technology, based on clothes that you already own, aim to help reduce the crisis.
‘More and more people are aware of their wasteful consumer behaviors,’ says Save Your Wardrobe CEO Hasna Kourda.
‘They’re taking a stand against that and are looking for tools to help them make better decisions.’
Though the fashion industry may have been slower than others to take up AI, experts are predicting enormous growth in the coming years.
‘It really allows people to focus more on the creative aspect [of fashion], as opposed to some of the more mundane and repetitive work,’ says Intellistyle’s Kourvakas.
‘Humans will always have a need to be creative, but AI will support that.’
While warnings of large job redundancies have loomed large, many in the industry, like Kourvakas and Kourda, are more optimistic about the potential for change.
‘Maybe one day it will be everybody’s tool. I don’t know where it’s going, we just wanted to try it and see if it works,’ says Johnasson.
Do you have any tips about the way technology is transforming industry?
Leave a comment below about what you’d like us to look at next.