An increasing number of people are turning to artificial intelligence for skincare advice, from personalized routines to rash identification.
Users often share before-and-after photos on platforms like Reddit, seeking cheaper alternatives to professional dermatological consultations.
However, experts warn that the advice provided by these chatbots is frequently inaccurate and can lead to adverse skin health outcomes.
AI's Limitations in Skincare
Dr Michelle Wong, a cosmetic chemist and science educator, compares chatbot output to "a blurry Jpeg of all the text on the web."
She notes that users rarely see what sources the AI uses, which could include user-generated forums or ill-evidenced blogs.
Wong has observed instances where chatbots instructed users to apply multiple products with the same active ingredients, like vitamin A, or recommended unusual application sequences.
In some cases, AI suggested nonexistent products or falsely claimed items were free of specific allergens.
"They're very confident liars," Wong says. "Try to remember it doesn't have a brain."
She points to previous research where an image-classification AI mistakenly associated rulers with skin cancer because training data frequently featured rulers next to malignant lesions.
Real-World Consequences
Dr Anita Lasocki, a dermatologist in Melbourne, has treated patients with skin issues triggered directly by chatbot recommendations.
"Most commonly I see irritant contact dermatitis or a flare-up of rosacea," she says, where patients layered too many steps or active ingredients hoping to solve their problem, but instead fueled it or created new complications.
Lasocki adds that AI chatbots cannot provide accurate diagnoses, leading patients to waste time and money on unindicated treatments.
Associate Prof Deshan Sebaratnam, a dermatologist at the Skin Hospital in Sydney, emphasizes that dermatology involves over 3,000 distinct conditions.
"AI is not good at separating these at present," he says.
A dry, scaly rash could indicate eczema, psoriasis, ringworm, scabies, lupus, or cutaneous lymphoma, each requiring distinct medical tests.
Sebaratnam also notes that teaching materials primarily feature white-skinned individuals, leaving AI poorly equipped to diagnose skin of color due to biased reference data.
The shift toward AI tools is accelerated by high costs and long wait times for specialists.
In Australia, the average dermatology fee is $230, and appointments often involve long waits.
For a basic skincare routine, Lasocki advises simplicity: a morning routine with SPF 50+ sunscreen, and an evening routine with a gentle cleanser and moisturizer.
"Not every patient needs active ingredients," she says.
Wong recommends applying products from thinnest to thickest consistency and always prioritizing packaging instructions over chatbot advice.
Sebaratnam sums it up: "No one understands your skin like a dermatologist."
