Google Unveils AI Skin Analysis Tool Amid Clinical Concerns
Google introduces a new artificial intelligence tool that analyzes photos to identify potential skin conditions. The app faces scrutiny from medical professionals regarding diagnostic accuracy and racial bias in training data.
Google unveils an AI-powered dermatology tool that analyzes user photos and asks questions to produce a list of possible skin conditions. The web-based app, known as dermatology assist, draws from a database of 288 hair, nail, and skin conditions to address 90% of commonly searched dermatology questions. Although the tool remains in the pilot stages in the U.S., European regulators approve it as a low-risk medical device, allowing Google to integrate it into search results later this year.
External researchers and clinicians express a mix of cautious optimism and concern regarding the new technology. A major worry involves the AI's training data, as the vast majority of dermatology images feature people with white skin, raising questions about whether the tool works equally well for all patients. Additionally, medical professionals question how everyday users interpret the app's results without formal medical training.
Google executives emphasize that the tool does not actually deliver a medical diagnosis. Instead, Chief Health Officer Karen DeSalvo states that the application serves to help patients feel more engaged with their own health before visiting a doctor. To date, no AI-powered dermatology device has received approval from U.S. regulators.