The rise of artificial intelligence (AI) in skincare advice has sparked growing concerns among medical professionals. As we increasingly rely on apps and chatbots for tailored regimens and condition identification, experts warn that the guidance provided can be untrustworthy – and potentially damaging to our skin health.
According to Dr Michelle Wong, a cosmetic chemist, AI's outputs often resemble a "blurry JPEG of all the text on the web." This is because its information sources are frequently unclear, spanning reputable dermatology websites to unknown blogs. She has witnessed instances where chatbots recommend using multiple products with identical active ingredients, suggest incorrect application orders (e.g., cleanser after serum), or even advise using non-existent or allegedly allergen-free products.
Dermatologists have already seen the consequences of relying on AI-generated skincare routines. Dr Anita Lasocki, a dermatologist in Melbourne, has treated patients who developed new skin problems – such as irritant contact dermatitis or exacerbated rosacea – after following chatbot advice. These issues often arise from layering too many active ingredients or steps into a routine, hoping to solve one problem but inadvertently creating another.
A fundamental limitation of AI is its inability to provide an accurate diagnosis. Associate Professor Deshan Sebaratnam, a dermatologist, highlights that there are over 3,000 distinct skin conditions. A dry, scaly rash, for example, could be eczema, psoriasis, ringworm, scabies, or even more serious conditions like lupus or cutaneous lymphoma. AI cannot perform the necessary diagnostic tests – such as skin swabs, scrapings, or biopsies – required to differentiate these conditions accurately.
The high cost of professional dermatological consultations can be a barrier for some individuals, who may turn to cheaper and quicker AI alternatives. However, experts recommend consulting reputable dermatology websites like DermNet, which offer extensive image repositories, as a more reliable approach than relying on AI chatbots.