All Articles
Tech & Culture

I Let an AI Dress Me for a Month. I Have Phased Partially Out of Reality.

By Vogue Vapor Tech & Culture
I Let an AI Dress Me for a Month. I Have Phased Partially Out of Reality.

I Let an AI Dress Me for a Month. I Have Phased Partially Out of Reality.

An op-ed by Vogue Vapor contributing writer Sloane Merritt, who is fine, probably.

It started, as most catastrophic personal decisions do, with a very reasonable-sounding pitch meeting.

"The future of fashion is AI," my editor said, sliding a list of apps across the table with the casual confidence of someone who would not personally be wearing any of them. "Spend a month letting algorithms design your outfits. Write about it. It'll be fun."

She said fun. I want that on record.

That was thirty-one days ago. I am writing this from my apartment, wearing what three separate AI systems collectively agreed was a "transitional earth-tone coordinated set" and what I can describe to you as: a beige garment with four and a half pockets, a neckline that begins at my left collarbone and ends somewhere near my right elbow, and what appears to be a decorative panel on the back that one of the apps labeled, with complete sincerity, as a "heritage gill reference."

I do not have gills. The AI felt otherwise.

How This Started: Optimism, Hubris, and a $12.99/Month Subscription

Week one was almost convincing. The AI styling app — I won't name it, but its logo is a geometric brain wearing a beret — asked me seventeen questions about my "aesthetic identity," my "color story," and whether I considered myself more of a "structured romantic" or an "intuitive minimalist." I said intuitive minimalist. It generated an outfit.

The outfit was a slate gray wide-leg trouser, a fitted cream blouse, and a long cardigan. It was perfectly normal. It was something a very put-together woman named Claire who works in publishing and owns a French press might wear. I was delighted. I told my editor this was going to be a very boring article.

By day four, the algorithm had learned enough about me to get ambitious.

The second outfit featured a blazer with three sleeves. Not in the fashion-forward, asymmetrical-runway sense. Three sleeves. Symmetric. Evenly spaced. The app's style notes described it as "a reimagining of bilateral silhouette norms that invites the body into a conversation about assumption." My body's response to that conversation was: where does my third arm go?

I wore it to a coffee shop. A child stared at me for six unbroken minutes. I told her it was fashion. She said, "But where does your other arm go?" I said I didn't know. She nodded like this was acceptable and went back to her hot chocolate. Children, I've found, are more adaptable than adults.

The Trench Coat Incident

I need to address the trench coat.

On day eleven, I fed my wardrobe data into a second AI system — a more advanced one that sources described to me as "trained on 40 million garments and the complete works of Balenciaga." I asked it to design me a classic trench coat. It generated detailed specs. I had them made by a sample studio in the Garment District, because I am a professional and I was committed to this bit.

The coat arrived on day fourteen.

It was, technically, a trench coat. It had the belt. It had the epaulettes. It had the double-breasted buttons. It also had a collar structured in a way that the AI's design notes described as "referencing the architectural confidence of a baguette," a lining pattern that could only be described as "aggressively food-adjacent," and — I am not making this up — a label sewn into the interior that the AI had auto-generated, which read: CATEGORY: SANDWICH (FORMAL).

I called the AI company's customer support line. A representative explained, with admirable composure, that the model had likely encountered a dataset overlap between fashion photography and food styling imagery during training, and that this was "a known edge case they were actively working to resolve."

"So the coat thinks it's a sandwich," I said.

"The coat," she said carefully, "exists in a liminal categorical space."

I have worn it four times. It is extremely warm. My doorman calls it "the lunch coat" and greets it separately from me every morning.

Weeks Three and Four: The Unraveling

By week three, I had stopped using one AI and started cross-referencing three simultaneously, feeding each system's output into the next as a prompt, creating what I can only describe as a fashion telephone game played between machines that have never seen a human body in person.

The results were: a pair of trousers with a third leg (not a design choice — the AI simply could not resolve the question of whether humans have two legs or three and split the difference), a turtleneck described in the style notes as "aggressively bilateral," a skirt that had been generated with twelve pleats on one side and zero on the other because two AIs disagreed and the third abstained, and a hat.

I can't describe the hat. I've tried. My therapist said I should stop trying and just feel what the hat made me feel. What the hat made me feel was that I had made a series of choices that led me here, and I should sit with that.

The hat had what I can only describe as sentient lapels. It was technically a hat. It had lapels.

What I Learned (Besides the Obvious)

Here is the thing about AI fashion that nobody in the breathless tech press coverage wants to say out loud: the tools are impressive, the outputs are fascinating, and wearing them in public is an experience that will fundamentally alter how strangers perceive your grip on reality.

Silicon Valley has decided that fashion is a problem to be optimized — an inefficient human behavior driven by irrational preference and emotional bias that can be systematized, personalized, and delivered via algorithm. And they're not entirely wrong that the current system is broken: fast fashion is an environmental disaster, personal style is increasingly homogenized by social media, and most of us are drowning in clothes we don't wear.

But the solution to that problem is not a neural network that classifies formal outerwear as a sandwich. The solution probably involves humans, taste, and the irreducible weirdness of wanting to look like yourself.

AI fashion tools are, at their best, a very enthusiastic intern who has read every fashion magazine ever published but has never left the office. They know the vocabulary. They know the references. They will generate something technically coherent with the confidence of a person who has never once had to walk through a subway turnstile in a three-sleeved blazer while a child stares at them.

Where I Am Now

I am back in my own clothes. Mostly. The lunch coat has stayed — I've grown attached, and it is genuinely warm, and my doorman would be devastated if it disappeared.

My editor called the experiment "a triumph of commitment journalism." I called it "a month I will never fully recover from." We have agreed that both things are true.

The AI apps, for their part, have continued to send me weekly outfit recommendations. This week's suggestion is a "deconstructed power suit in anticipatory taupe" with what the style notes describe as "one and a half sleeves, for balance."

I have not opened the email.

I am not ready.

The coat, though. The coat I'm keeping.


Sloane Merritt is a contributing writer at Vogue Vapor. Her wardrobe is currently 60% human-designed and 40% in a liminal categorical space. She can be reached via email, but not on Tuesdays, when she wears the hat.