I trusted AI to dress me – it was really ageist

Chat GPT fashion
Chat GPT fashion

From creating shopping lists, replying to school correspondence or deciding the best wine to drink at supper, AI has infiltrated every aspect of our lives. As growing numbers of women outsource life admin to ChatGPT, I wondered whether it could solve the ever-present question, “What shall I wear?”

AI’s use in fashion is hotly debated. Some believe AI steals creative jobs, encourages dupes and promotes an algorithmic monoculture on social media, yet management consulting firm McKinsey & Company reported AI could add $275 billion profit to the fashion and luxury industry over the next five years.

Aistetic, a University of Oxford spinout, uses AI to build “sizing confidence”, while Amazon launched a “Fit Insights” tool this month. If AI can remove the painful experience of being a small in one store and a large elsewhere – along with the financial and environmental drains of inconsistent sizing (and subsequent hassle of returns) – I’m on board.

But what about styling advice? Several apps, including Cladwell, Whering and Style DNA, use AI to suggest outfit options once you’ve uploaded pictures of your own pieces. If you’re blessed with time on your hands, go for it, but I am time-starved so deploy ChatGPT’s guidance instead.

Tentatively, I begin to type…

I’m presented with four outfit options. “Casual but chic” appeals and I’m given a checklist of pieces to build my look. I need “a fitted blouse or a stylish sweater.” My colouring can’t take the suggested “jewel tones” so I steer towards “blush” (babygirl pink is having a moment for spring 2025.) You can amend your question to get further recommendations, skip certain items, add your size, shape, skin tone and weather - the more you share, the more specific the response.

My bottom half options include “sleek” black skinny jeans and I instantly spot a gap in ChatGPT’s contemporary style knowledge. Skinny jeans haven’t been fashionable for 15 years! But I accept the AI advice and squeeze into skinny denim for old times’ sake. Footwear suggestions include “ankle boots, ballet flats, or heeled sandals.” They’re all so different!

I pick ballet flats because there is no guidance on heel heights to suit the jeans. I could request more ideas, but I’m not confident ChatGPT can identify an in-proportion look. Finishing touches? “A delicate necklace… fun scarf or lightweight jacket.” I’ll definitely need a jacket, it’s January! I waver over the “fun scarf” and choose a necklace instead.

Alexandra Fullerton for The Telegraph
ChatGPT’s “date night” option for Alexandra - Andrew Crowley

Blouse, £100, Sezane; Jeans, £45, River Island; Jacket, £49.99, Zara; Necklace, £150, Otiumberg; Belt and shoes, Alex’s own

Adding my age identifies me as millennial and seems to colour ChatGPT’s answers so I repeat my question, changing the age to 65. My suggestions are now “based on style, comfort, and a touch of elegance” along with the condescending intro, “How wonderful that you’re planning a date night!” I flip the question again. What if I was 20? Options include a “form-fitting jumpsuit,” “fun, flirty mini dress” or a “bodycon dress” which can be “both sexy and classy.” It seems ageism is alive and well in the AI realm.

For “a formal dinner” my options include, “a knee-length wrap dress or sleek, form-fitting sheath dress – something that highlights your shape.” I baulk at the thought of “highlighting my shape” and muse that the repeated suggestions of fitted garments aren’t ideal for a lunch date. Again, I could request roast-dinner appropriate outfits but as each ChatGPT query uses 2.9 watt-hours of electricity (10x more than a quick Google) and could power 21,602 homes for a year, I shelve my concerns.

My rating: 6/10

ChatGPT’s fashion advice is incredibly generic and while serviceable, the outfit ideas are dated. I’m bewildered at the visibility of my legs in spray-on denim. I’d forgotten the weird hopping dance required to hoick up skinny jeans and lose confidence in AI.

On to my next question:

“You need to look polished… but also express your creativity.” Spot on! But how? I’m given five scenarios and pick “edgy professional.” I immediately meet a block. Along with a “tailored leather skirt” I need “a blouse or shirt in… a bold, geometric pattern.” This is hard to source and the specific design detail redacts my choice of brands. The few options I find look naff. Am I searching in the wrong places?

I ask: “What fashion brands are appropriate to wear in a creative office environment?” The answers include Reiss, Theory and Banana Republic. All go-tos for smart City-office workers. I roll my eyes. Obviously they’re in a different price bracket, but what about Marni, Stine Goya or even Me+Em? Everything AI suggested so far feels basic. Is it my age again? I abandon “edgy professional” and switch to “powerful and elegant.” I need “a well-tailored dress in a solid colour,” a “structured cardigan,” “neutral-coloured heels…” and “a structured bag.” On the shoot, someone describes my look as very “Tory party conference.” I feel like I’m cosplaying corpcore, not heading into a buzzy creative office.

Alexandra Fullerton for The Telegraph
ChatGPT’s “creative” office wear option for Alexandra - Andrew Crowley

Dress, £255, Jigsaw; Cardigan, £35.99, Zara; Bag, £550, DeMellier; Shoes, £185, Bobbies

My rating: 4/10

ChatGPT didn’t acknowledge the “creative” part of my question or have any understanding of creative and contemporary fashion brands. This look is wildly off brief, although I do like the individual components.

For my final query:

Realising ChatGPT doesn’t have a fashion editor’s knowledge of contemporary garments, I ask how it would style a fashionable piece. I disagree with the opening statement “with your height and size, there are plenty of ways you can make (barrel-leg jeans) work for you.” I’ve tried countless options that are usually too long or ridiculously baggy on my frame.

However the styling tips, such as “balance volume,” do make sense so I decide to channel “effortlessly cool.” My outfit requires: “a neutral turtleneck sweater tucked in… tan or beige barrel-leg jeans… tan block-heeled ankle boots… statement handbag… layered gold necklaces.” It’s tricky finding barrel-leg jeans to suit, so reducing the search to tan or beige limits my choice. I need to roll the hem of the jeans – unapproved by ChatGPT, I get an illicit thrill from styling off-piste.

Alexandra Fullerton for The Telegraph
ChatGPT’s suggestion for styling barrel leg jeans - Andrew Crowley

Jumper, £175, Brora; Jeans, £120, Seventy + Mochi; Boots, £295, Russell & Bromley; Pre-loved Chanel bag, £5515, Sign Of The Times

Wearing one colour head-to-toe is a dressing trick used by The Princess of Wales to look polished so this is an outfit suggestion I approve of, even though the overall effect is rather beige. I skip the layered gold necklaces. They don’t sit well over the high-neck jumper (something ChatGPT didn’t clock) but I am thrilled to show off a Chanel handbag. Well, you can’t get more “statement” than that, can you?

My rating: 7.5/10

With fashionable direction you can find a decent outfit suggestion from ChatGPT, but it requires the user to steer the options and use discretion when it comes to colour mixing and proportion play.


Why ChatGPT can’t be your stylist (and what to do instead)

Unless you are hyper focused and already have a stylist’s knowledge of on-trend pieces and independent brands, ChatGPT will likely give you a deathly dull outfit idea. I don’t believe anyone is so busy they need ChatGPT (and its hefty carbon footprint) to pick their outfits.

Duncan McKay, founder and CEO of Aistetic believes that “AI needs help. It needs to understand what you’re looking for.” I admit it took a while to finesse my questions and perhaps my expectations were too high… McKay continues, “style is incredibly subjective” and points out that one may visit a store several times before the shop assistants understand our taste and suggest pieces we actually want to buy.

Beyond ChatGPT, when your fashion brain is flummoxed, what else can you do? Phone a friend who knows your wardrobe, browse Pinterest, or pre-empt the fashion emergency and consult a stylist… Sara Parker Bowles founded Wrapp10, along with Florrie Thomas and Lucy Coldharbour, and their clever service connects shoppers to stylists for outfit suggestions that are distinctly bespoke.

Wrapp10 founder Sara Parker Bowles and Lucy Baxter
Wrapp10 founder Sara Parker Bowles and Lucy Baxter, founder of Coldharbour Digital, the agency responsible for building the original Wardrobe Revivals site - Dunja Opalko

“Whilst everyone was launching AI fashion advice, we went the other way by putting emphasis on human advice,” she says. “Even though the whole idea is digital [you can book a video consultation online] it’s very much based on real life. We can use AI as a great tool but some things can’t be farmed out to computers.”

Personal style, it seems, is one such thing.