- Fashion {NUT}
- Posts
- Can AI Learn Good Taste?
Can AI Learn Good Taste?
Luxury brands are training algorithms on decades of design to keep their identities intact in an AI world.
Tam Sood
May 29, 2025
Hey there,
This week, we’re talking about good taste — and whether a machine can fake it.
As AI floods fashion — from trend reports to Social media fits — one corner of the industry is taking a quieter, smarter approach: training AI to behave like a legacy brand.
Forget “what’s trending.” These brands want algorithms that know the difference between minimalism and meh.
If you’re in fashion, here’s why your next intern might not be a design student — but a well-fed neural net trained on your runway archives.
Here’s what’s inside:
✅ How LVMH and Norma Kamali are building AI with brand identity
✅ Swipe File: What a taste-trained model looks like in the wild
✅ Tool of the Week: Adobe’s generative moodboard — cheat code for creative decks
✅ Fun Fact: Karl Lagerfeld, iPads, and the gloves stayed on
⏱️ Total read time: 2 minutes, 51 seconds
Feature Story
The AI intern who knows your brand better than you do
Luxury fashion is teaching algorithms how to “have taste” — and it’s working
TL;DR
Most brands use AI to speed things up.
But some fashion houses are using it to slow down — to protect what makes their style theirs.
Instead of relying on generic prompts, they’re feeding AI their archives, aesthetics, and attitude.
The result? A machine that doesn’t just make clothes — it understands the vibe.
The What — AI with Fashion Degree
AI is great at generating stuff. But good taste? That’s harder to code.
So brands like LVMH and Norma Kamali are building custom-trained AI models.
Not to chase trends — but to preserve identity.
They’re uploading decades of campaigns, sketches, and runway shows to teach machines what makes a “Dior” jacket a Dior jacket — not just a jacket with buttons.“
1. LVMH Built an AI Lab That Knows Its Brands Better Than Most Interns
At LVMH’s Paris-based “AI Factory,” engineers work hand-in-hand with brand teams to build tools that speak fluent fashion house.
These AIs aren’t recommending trends — they’re learning what not to do.
Why it matters:
If you’re Dior, looking like Prada is a business risk. LVMH’s AI helps avoid that.
2. Norma Kamali Cloned Her Creative Brain (Kinda)
Fashion icon Norma Kamali trained an AI on 50 years of her own work.
Now it spits out new designs in her signature silhouette — structured, future-forward, very Kamali.
Why it matters:
It’s not about replacing the designer. It’s about turning her archive into a living, designing, remixing assistant.
3. Videofashion Is Letting AI Mine 18,000 Hours of Runway History
What do you do with decades of footage from every major runway show since 1976?
If you’re Videofashion, you feed it to an algorithm and let it spot patterns humans missed.
Why it matters:
Fashion moves fast. History keeps receipts. AI helps connect the dots.
Who Wins?
Heritage brands: That have a “look” worth protecting
Independent designers: That want to scale without selling out
Archivists + platforms: That want to turn style memory into market insight
TL;DR: If your brand has taste, train your AI before someone else’s generic model eats your lunch.
🧵 Key Takeaway
Taste used to be intangible. Now it’s a dataset.
The brands doing this right aren’t replacing their creatives — they’re giving them superpowers.
In an AI-saturated world, your aesthetic might be your last competitive edge.
Swipe File (Real tactics used by Fashion brands)
Steal-This Play
Brand & Move
Norma Kamali x MetaMaison – Trained a Private AI on Brand Archives
What They Did
Norma Kamali collaborated with AI startup MetaMaison to feed 50 years of her design work — runway shots, sketches, patterns — into a proprietary model. The AI now generates new designs in her signature silhouette, helping her ideate faster while staying on brand.
Steal This Play
Have a distinct design identity? Train an AI on your own product photos and brand assets (yes, even using Midjourney or Claude with the right tuning). Use it for moodboarding, concepting, or even internal critiques.
Bonus: It keeps junior designers aligned without killing creative flow.
The Runway Reel
This week, i created this time capsule about Fashion in Italy in 1972. Made using AI. have a look
Tool Time
AI Tool of the Week: Adobe’s Generative Moodboard
Your creative director’s Pinterest board — but on steroids
What it is:
Adobe just rolled out a generative moodboarding tool in Firefly (its AI suite) that builds entire visual concepts from a single prompt — no dragging, cropping, or collaging required.
You type “futuristic tailoring in muted neutrals,” and boom: layout, vibes, references — all done.
Why it matters:
Moodboarding used to take hours of pinning, swiping, and convincing your team that “offbeat Brutalism meets cozycore” was actually a thing. Now it’s a one-click starting point for your next campaign, drop, or lookbook.
Bonus:
You can drag in your own assets to train the vibe. Want every board to feel like your brand? Just feed it a few of your best shots — the AI adapts.
A Final Note
Fun Fact
In the 1990s, Karl Lagerfeld refused to use computers — he sketched every Chanel collection by hand.
But when Apple released the first iPad?
He started using it to draw his collections. In full gloves.
Why?
Because, according to him, “I’m not a child of the digital age. I’m its stepfather.”
Lesson:
Even the most analog legends eventually meet their algorithm.
In fashion, the tools change — but the taste better not.
Let me know what you thought of this edition.Until next time,
