For and against the AI stylist

Fashion experts debate the pros and cons of retailers using artificial intelligence to recommend products and outfits online

If we only wore clothes for warmth and modesty, then fashion as we know it wouldn’t exist. We hold on to dresses that no longer fit for the memories they conjure. We project our moods by the cut of a shirt. We judge and admire the style of others. We blush at compliments. So, what happens when technology invades the emotional space? Fashion retailers increasingly see personalisation as key to reaching consumers online and in-store, notably in the form of artificial intelligence (AI) stylists, dressed-up algorithms that can trawl through enormous amounts of customer data to make personalised styling suggestions. But do they limit intuitiveness and individuality? Or can AI stylists help to nurture self-expression?

AI stylists: the pros

Imagine trawling through 3.7 trillion outfits to suggest a mere handful to one individual. No stylist, however committed to the fashion cause, would take on such a task. But an AI stylist would relish it.

“AI can instantly look through millions of options, remember your preferences and learn from your behaviour,” explains Kieran O’Neill, co-founder and chief executive of menswear styling service Thread. “But our focus has always been on developing AI that can scale and improve the expertise of a [human] stylist, not replace it.” In other words, to do the “heavy-lifting”.

According to Simon Leesley, managing director of Stitch Fix UK: “They serve as a recommendation engine for our stylists so they can be curators rather than scroll through and pick from thousands of pieces. The algorithms understand fit nuances that humans can’t and remove any selection bias. They provide the science and our stylists provide the art.”

Leaders in machine-learning

Thread and Stitch Fix are pioneers of this complementary model where human stylists, with their flair and creativity, are supported by the data-crunching of technology. But with 11 humans, in Thread’s case, to 3.7 trillion AI-generated outfits, can these services really help individuals to express themselves, to nurture a unique sense of style?

“It’s rare that two people get the same suggestions,” says O’Neill. “Our stylists are constantly adding edits to the pool that AI picks from, so we keep things fresh and avoid stale, repetitive, self-reinforcing ideas. We have the biggest range of menswear brands in the UK – 871 – so we introduce our customers to [designs] that they wouldn’t otherwise be aware of.

“Unlike other AI styling services, we’re not a subscription box that sends a limited, surprise selection of clothes. We give our customers inspiration, tips, advice, all tailored to them, but ultimately they’re in control, which allows the nuances of individuality to come out.”

We give customers inspiration, tips, advice, all tailored to them, but ultimately they’re in control

But if AI can only make recommendations based on existing and historical data, how can the algorithm ever push customers out of their comfort zones? “We’ve chosen to accumulate and decode data at a granular level,” says Jessica Murphy, co-founder and chief customer officer of global personalisation platform True Fit, whose clients include H&M and Levi’s.

“But we look at stretch, fits, fabrics – one customer only buys non-iron shirts, for example – not just colour. We’re trying to create an understanding of you as an individual, not just in one moment.
So, when a customer said, ‘I’d never pull off white jeans’, we suggested a pair that really fit and she loved it.”

Leesley agrees, referring to humans as “creatures of habit”, who gravitate towards the same styles in the same colours from the same brands. “Many of our customers tell their stylist, ‘I never would have picked this Baum und Pferdgarten jumper, for example, but I absolutely love it’. In this sense, it can be a very fun and creative way to shop.”

Democratising fashion styling services

AI also puts an elitist industry on a level playing field. “A personal stylist has traditionally been a privilege reserved for higher-income classes,” says Kostas Koukoravas, founder and chief executive of Intelistyle, adding that before algorithms consumers used fashion magazines for style inspiration.

“AI stylists offer the same service, but outfits are a lot more personalised to the customers’ needs. And they can see a lot more products as opposed to just the ones promoted by the magazine’s editorial team. The experience becomes a lot more fun and individual.”

Not flamboyant, maverick or trend-setting, though. Yes, fashion has the ability to give individuals a wonderful, visual platform for big self-expression. But for most of us, with the occasional, stand-out moments, we just want to look and feel good. “We surveyed people across the UK and found that one in two were looking for styling advice when shopping and 75 per cent would buy personalised recommendations from a styling service,” says Koukoravas.

“The majority of individuals find it difficult to articulate what they like, what suits them,” adds Murphy. “That moment when a customer turns around, looks in the mirror and says, ‘I love it’, we want to win at that.”

AI Stylists: the cons

If ever there was an industry that feared being left behind, it’s fashion. Much in the same way that it sets style trends, fashion wants to keep up with the latest technological innovations, too. But not every retailer is ready for them.

Personalisation is most definitely a way to exceed the expectation of today’s shopper. How this is achieved, however, is more about digital evolution than AI revolution,” says Olga Kotsur, co-founder and chief executive of digital retail technology company Mercaux.

“Retailers can be blinded by the hype of the AI revolution without addressing the foundations of in-store digitalisation that service basic customer needs, such as alternative recommendations. The real challenge for retailers is simply knowing where to begin. That’s why we closely advise our customers on their transformation journey, so AI styling is considered for deployment when the store, and crucially the staff, is ready to use it.”

Similarly, depending on where they shop, customers have different expectations of the type of service available to facilitate their styling choices. “We must make a distinction between top-end luxury brands and mainstream high street brands,” says Mario Coletti, UK managing director of AI data analysts iCoolHunt, which worked with shopping centre Westfield on its launch of The Trending Store.

“AI will never reach the high level of creativity and sophistication that is provided by human stylists. So, at the very top level, human stylists will always be in demand and potentially their role will grow, also thanks to the growth of tools and solutions that technology will provide such as the AI stylists.”

Limits in creativity of AI-driven services

While businesses have made a strong case for AI’s ability to nurture, rather than restrict, self-expression, industry futurist Doug Stephens, founder of Retail Prophet, argues that there is a limit to AI’s creative capabilities. “All AI can deal with is known data, so it develops a predictive model that doesn’t expand a world of possibilities; it shrinks it,” he explains. “AI never gives you a radical departure, so it will never offer a highly creative solution.”

The ability to talk with someone, share our biggest fears and worries, but also have fun with them, is key

Stephens also points to the emotive experience of buying and wearing clothes, to the “human intuition and rapport with customers”, a viewpoint supported by other experts.

“While AI is incredibly efficient, it lacks empathy and communication skills. What I wear when I need to feel comforted might be different from what I wear on a day I’m gleaming with confidence,” says Koukoravas. “Human stylists play a huge part in understanding those associations, helping us find the right clothes in a more organic way. The ability to talk with someone, share our biggest fears and worries, but also have fun with them, is key.”

Leesley doesn’t see a time when AI will be good enough to style people without humans. “AI will continue to become more sophisticated, but our aim is not to have technology replace humans. We are in the relationship-building business and the world needs more meaningful interaction between people, not less,” he says.

“We generally think people are uncomfortable with the idea of a ‘robot’ as their personal stylist. So, we’ve built a human-centric service to solve a very human problem, ‘How do I find clothes that I love, that fit me and make me feel great?’ It’s natural that we need stylists at the core of such a business.”

Even when AI is at the very core of a shopping experience, like Westfield’s AI-powered pop-up store, stocked exclusively with 100 trending items based on real-time social media data, the human touch is crucial.

“The technology used to create the store was innovative and relevant in the digital age we live in. But it still required a stylist pulling the right product from the retailers that encompassed the key trends in a relevant way consumers would by, and staff were on hand to communicate how it all worked,” explains Myf Ryan, Westfield’s chief marketing officer for Europe, and group director of brand and strategic marketing. “The human element was essential.”