Three fingers, six fingers, and a prompt: AI meets Australian Fashion Week

The experimenter

Karla Spetic is not a technophobe. The Croatian-born, Melbourne-based designer has built a label on romantic silhouettes with a whimsical undertone, the kind of clothes that look like they were sketched by hand at a cafe table, which they probably were. But ahead of her showing at Melbourne Fashion Festival this year, she opened a new tab and started prompting.

She used generative AI for slogan text on a T-shirt range. The image generator, though, was a different story.

“The ideas come from my heart,” she told the ABC’s Jane Rocca, “and AI came back with affirmations I could use, but sometimes the designs had human hands and fingers with either three or six fingers. It was really bizarre and not always accurate.”

The hands thing is almost a cliche at this point. AI image generators have been adding and subtracting fingers since DALL-E 2. But Spetic’s point is not really about the fingers. It’s about the distance between what a prompt can summon and what a designer knows in her bones. “AI didn’t replace my creativity,” she said, “but it did demand clarity.” She had to get specific about what she wanted in a way sketching never required. The machine made her a sharper thinker by being worse at intuition.

Here’s what gets me about this. Nobody in this conversation, not Spetic, not the people I’ve spoken to, not the ABC’s sources, is actually arguing that AI should design the clothes. The argument is smaller than that, and more interesting. It’s about which parts of the process are sacred and which are just work.

The convert

Vince Lebon is on the other side of this question, and he is unambiguous about it. The founder of footwear label Rollie Nation cut his teeth at Asics and Nike and has a background in coding. He built his own AI tool. He uploads a sketch and it generates an entire footwear range with realistic product imagery.

“I just presented my first collection using the tool to my team,” he said, “and normally we’d share outlines of sketches, but with AI I was able to present full photos of new footwear complete with realistic imagery and it was incredible.”

Lebon has gone further than most. He told his assistant designer that no shoe goes to sampling until it has been through the AI pipeline, a posture that feels less radical when you remember that some Australian labels are already going from sketch to Shein in three days. The logic is hard to argue with: sampling is expensive, physical prototyping burns time and materials, and if a machine can show you what a shoe looks like before you cut leather, you’d be reckless not to look.

“As a business owner, I’m looking at it as a tool that can streamline processes, sell my ideas efficiently and also reduce costs,” he said. The fear that AI eliminates jobs, he thinks, is misplaced. “It’s a fear that workers won’t be needed, and that AI will replace them.” He sees it as leverage, a way for a small Australian label to move at a speed that used to require a much bigger team.

I find Lebon’s position compelling, and also a little unnerving. Not because he’s wrong. He isn’t, on the economics. But “no shoe goes to sampling without AI” is a gate, and gates, once installed, tend to stay. What happens when the AI’s taste, trained on every shoe ever made, starts to feel more reliable than the designer’s instinct? What happens when skipping the AI check feels like negligence?

I don’t know the answer. I’m not sure anyone does.

The sceptic

Prabal Gurung has dressed Michelle Obama and Lady Gaga. He was in Melbourne for the Fashion Festival in February, and his take on AI lands somewhere between Spetic’s cautious pragmatism and a flat refusal.

“My creative process is my lived experience; AI can’t be that,” he said. “It can try to blend cultures and tell a new story, but to be honest, it’s nothing extraordinary. That comes from the human touch, the human experience. You could never rely on an entire collection to be built by AI as it would be missing something very important.”

“Nothing extraordinary.” That’s the rudest thing anyone in this piece says about AI, and the most important. Gurung is not arguing about efficiency or cost or whether a generated image has the right number of fingers. He is arguing about whether the output is any good. And from the position of someone who has actually looked at what AI produces in fashion, he is saying it is not.

Like the others, he concedes AI is useful on the margins. Spreadsheets, not sketchbooks. “It helps streamline the administrative side of the business, and that’s where AI is most efficient.” He draws a hard line at the creative core. Then he says something that sounds less like a fashion designer and more like someone who has been doing this long enough to stop caring about quarterly numbers: “In a fractured world, fashion is needed more than ever to foster a community, and when you remain true to your identity and voice, your collections will survive a long time.”

It’s generous and mostly right. But the uncomfortable question underneath it is not whether AI can make good art. It’s whether the economics of this industry will leave enough room for designers to spend time at the “lived experience” end of the spectrum in the first place. If the margins keep tightening, Gurung’s argument doesn’t fail on its merits. It fails because nobody can afford to live it.

The wider frame

This conversation is not happening in a bubble. In January, Alexis Mabille put an entirely AI-generated couture show on a Paris runway. Every look, every fabric drape, every invitation image. February was worse. Gucci copped a wave of backlash for AI marketing images, readers calling them “cheap, chintzy, lazy.” Then Miuccia Prada and Raf Simons, at their own show in Milan, named AI and inequality as fashion’s two greatest challenges. Not one or the other. Both, in the same sentence. There’s an essay in that pairing but the short version is: the people at the top of this industry see the same thing from their perch that a Melbourne designer with a laptop sees from hers.

Closer to home it gets messier because the stakes are less abstract. Last year Melbourne label Atoir used AI-generated models to sell clothes on The Iconic and customers felt deceived. Not because the clothes weren’t what they ordered. Because the person wearing them didn’t exist and nobody mentioned it. Mid-2025, Vogue ran a Guess campaign with an entirely generated model and the same thing happened. Technically flawless image. That was the whole problem. It raised the bar on what consumers think they should look like without raising it on anything that breathes.

Fourteen designers are debuting at Australian Fashion Week this year. Van Ermel Scherer, a proud Larrakia woman, is showing luxury swimwear named after her late grandmother. Paris Jade Burrows wants people to “feel disgust.” Gloria Chol walked across a country to survive and is now putting South Sudanese-Australian streetwear on a runway. These are labels built on perspectives a prompt cannot fake, histories a training set does not contain, and hands that have actually cut fabric. The question AI poses to them is not the abstract one about creativity versus efficiency. It’s simpler and harder: how long can you afford to stay human-scale, and what breaks when you can’t.

What I keep coming back to

The thing I can’t shake from Spetic’s account is not the three-fingered hands or the slogan prompts. It’s that she said using AI “demanded clarity.” A machine that cannot feel anything forced her to think harder about what she wanted to say.

That’s not a failure. It might be the best argument for the whole enterprise.

But there’s a difference between using a tool that sharpens your thinking and one that replaces it. Lebon’s AI pipeline sharpens his process. The risk is that it slowly dulls his instinct, and he won’t notice because the shoes will still look good. They might even look better, by the metrics a machine understands. Whether they’ll feel like anything? Different question. Not one a prompt can answer.

I’m less sure than Gurung that AI cannot make something extraordinary. The history of technology in creative fields, synthesisers, CGI, digital cameras, is a history of machines making things their inventors did not anticipate. But I am certain he’s right about the thing that actually matters. What counts in fashion is not the output. It’s the person who made it, and what they meant by it, and whether anyone on the other side of the runway felt it.

Australian Fashion Week runs through Thursday. Somewhere in Sydney right now a designer is pinning a hem by hand and another is typing a prompt. They are both making clothes. The difference is not in the fabric.