A US clothing company is selling itself as AI-led. But what does that mean? Josie Adams finds out for IRL.

I have started getting ads for clothing that doesn’t exist. There are digitally-rendered dresses all over my Instagram timeline that look like they’ve been plucked out of the next season of Euphoria, but according to the website that sells them they haven’t actually been made. Finesse, a US-based fashion company, is using artificial intelligence to make clothing at exact demand – no overstocking.

Finesse uses AI to predict trends in the fashion industry, then makes 3D models of the designs it thinks are trending. These designs are placed on the website as a showcase of the next drop – customers can pick the design they’re most excited about, and the designs that get the most clicks are basically voted into existence. 

This isn’t the first time a fashion house has employed AI. H&M uses it to predict demand and thereby improve its supply chain, and Zara uses it in robots that work with its inventory. Many industries use AI to streamline processes: Jetstar has an AI chatbot to help you resolve problems, and Netflix uses it to throw up terrible film suggestions. What Finesse seems to represent is a conceptual shift: AI isn’t a background assistant to the company, it’s a selling point. It’s fashion, but apparently more sustainable – all thanks to machines. But how sophisticated are these machines, really?

Alex Bartley Catt, the managing director of Auckland-based AI company Spacetime, says using AI as a selling point could work because it’s more of a concept than a specific technology. “AI is understood as more of a cultural and marketing term,” he says. “And it’s a banner term for a lot of different technologies that do cool things with data.” That’s what Spacetime does: cool things with data, sold under the umbrella term “AI”. Spacetime’s clients use chatbots to reduce call centre load, natural language search engines to create company Googles, and robotic process automation to do data entry work.

Recently, Spacetime’s client Fine Wine Delivery asked for something more customer-facing: a “flavour wheel” that selects your perfect bottle of wine. Catt says New Zealand companies are mostly still focused on using AI at the backend of their businesses. “They’re looking for efficiency, they’re looking to automate things,” he says. “It’s only when you get really visionary entrepreneurs, like Jeff Poole from Fine Wine Delivery, that you start actually putting this stuff in front of customers.”

While Fine Wine Delivery is open to using AI at the customer level, fashion houses are still using it in the backend; it seems like Finesse isn’t doing anything differently, it’s just talking about it more. Finesse’s AI is all about predicting trends. You might think predicting trends in a subjective field like fashion would be difficult, but apparently not. Founder and CEO of Finesse Ramin Ahmari has described the stock market as “much more unpredictable” than fashion.

A daily deluge of #OOTD on social media means AI technology can crunch what the cool kids are wearing and spit out predictions: butterfly tops are on the rise. A human mind can look at these predictions and decide whether or not they make sense. It seems Finesse is, to a degree, outsourcing this step by getting customers to vote on their favourite designs.

“It might also sound a little bit cliche, and like I’m trying to cover my ass, but the best result today is always going to be a collaboration of human and machine,” says Catt. “AI can be way better than humans, particularly when reading radiology scans and looking for that sort of thing, but context is where AI always loses out.” This is where humans remain superior: context. An AI can tell you there’s a chair in the room, but only a human can tell you why the chair is there and who might sit in it. And at a fashion house, a human needs to be there to say yes, butterfly tops really are a cool idea.

This is good news: Finesse is not proof robots are taking jobs away from our fashion designers. While Spacetime hasn’t yet dipped a toe in the fashion industry, they’ve done plenty of other projects. “No one’s ever lost a job,” says Catt. All that happens is upskilling. “That person goes from having to do manual data entry themselves to watching over their army of robots doing it for them and solving problems as they come up.”

But will the robots ever solve their own problems, and will humankind be considered a problem? We won’t lose our jobs, but could we lose our way of life? “It’s a matter of time,” says Catt. “We’re just nowhere near that.”

We’re still in a relatively early stage of AI in the workplace. We can ask Alexa to give us a pop quiz, but we can’t have a conversation with it. We can ask DALL-E2 to paint us a picture, but it can’t tell you whether the picture is good or bad art. And although futuristic-looking clothing brands might be using AI as a marketing ploy, they’re not actually using any new-fangled technology that could wipe out fashion designers or life as we know it. Human and machine still need each other – and probably will for some time.

I watch Catt fiddle with his airpods over the Google Meet screen. “I think we’re kind of already cyborgs in the sense that we’re always on computers, always looking at screens, always have our phone with us and it controls us in a way we don’t even comprehend and understand.”

There may come a tipping point when the machines no longer need us, but the fashion industry isn’t it.