Research into how AI can help users understand skin conditions

2026-06-12 · Source: The latest research from Google · Field: Health & Wellbeing — Health & Medical Research, Medical Devices & Health Technology, Clinical Care & Medical Practice · Depth: Intermediate, medium

Summary

Google Research scientists Rory Sayres and Yun Liu published findings on June 12, 2026, detailing how AI tools can help laypeople understand skin conditions. A large-scale quantitative study involving 2,345 participants, published in JAMA Dermatology, demonstrated that AI assistance significantly improved users' ability to name conditions, with accuracy nearly tripling from 8% (control) to 23% (AI arm). Participants also showed increased willingness to guess (62% vs 41%) and higher confidence. However, this study found no significant improvement in determining appropriate next steps. A separate mixed-methods study, published in ACM CHI, involved 110 diverse participants using a translated AI app for their own skin concerns. This study reported a 260% increase in condition naming ability, with clinicians finding the app's predictions 86% consistent with their assessments and deeming it helpful 92% of the time. Both studies highlight the importance of visual matching and human-centered design for effective AI health tools.

Key takeaway

For AI Scientists developing consumer-facing health tools, prioritize human-centered design that extends beyond mere condition identification. While AI significantly improves users' ability to name skin conditions, your focus must also include robust guidance for appropriate next steps. Integrate diverse visual examples and personalized, actionable information to ensure users can effectively interpret and act on AI insights, thereby supporting safer healthcare journeys.

Key insights

AI tools substantially enhance laypeople's skin condition identification, but require more human-centered design for accurate next-step guidance.

Principles

• AI significantly boosts condition identification.
• Visual matching is critical for user comprehension.
• Human-centered research ensures effective health AI.

Method

Evaluated AI's impact on skin condition understanding via a large-scale survey (2,345 participants, three arms including "Wizard of Oz" control) and a mixed-methods real-world study (110 diverse participants using a translated app).

In practice

• Offer diverse "textbook" examples for visual matching.
• Integrate actionable, personalized user guidance.
• Employ multimodal image and text AI approaches.

Topics

• Dermatology AI
• Consumer Health
• Human-Centered Design
• AI Model Evaluation
• Skin Condition Diagnosis
• Multimodal AI

Best for: Computer Vision Engineer, AI Product Manager, Research Scientist, Domain Expert, AI Scientist

Related on AIssential

• Humanising Healthcare - AI and Innovation at Optum — Human-centered innovation and responsible AI are crucial for transforming healthcare and improving patient outcomes.
• AI identifies early risk patterns for skin cancer — AI models can predict melanoma risk with high accuracy using routine healthcare data, enabling targeted screening.
• How people use Copilot for Health — Conversational AI serves diverse health needs, from personal symptom checks to caregiving and system navigation, with usage patterns varying by device and time.
• Good News & Bad News: AI is better than most therapy for some people. You need to understand some nuance, but its genuinely extraordinarily valuable. — AI provides uniquely personalized, accessible, and nuanced mental health support, often surpassing traditional therapy for specific needs.
• Health-care AI is here. We don’t know if it actually helps patients. — Despite rapid AI adoption in healthcare, its impact on patient outcomes is largely unmeasured.
• Vivek Natarajan of Google DeepMind at RAAIS 2026 — Building useful AI in expert domains requires moving beyond benchmarks to real-world clinical interaction and new knowledge generation.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The latest research from Google.