Skin of Color Still Underrepresented Across Major Dermatology Image Searches

Key Takeaways

• Skin of Color (SOC) remains underrepresented in dermatology-related image search results across major search engines, a research letter suggests. 
• Google showed improvement in SOC representation since 2020. Gains were modest and fell short of proposed equity benchmarks.
• Limitations in image sourcing and search algorithms continue to affect representation and create disparities across Google, Yahoo, Bing, and DuckDuckGo.

Skin of Color (SOC) remains significantly underrepresented in image search results across major search engines, according to a recent Letter to the Editor published in the Journal of Drugs in Dermatology.

Investigators building on earlier research in the area evaluated the first 50 images returned for 74 dermatologic conditions across Google, Yahoo, Bing, and DuckDuckGo. They then categorized the images according to skin tone using Fitzpatrick skin types, with Fitzpatrick V and VI classified as dark skin. Researchers excluded duplicates, non-photographic images, and photographs in which skin tone could not be reliably assessed.

Across all four search engines, only 5.88% of images depicted dark skin. Google demonstrated the highest proportion of dark-skin representation among both the top 50 images (6.81%) and top 10 images (7.03%), outperforming Yahoo, Bing, and DuckDuckGo.

Compared with findings published in 2020, Google showed a statistically significant increase in dark-skin representation of 1.14% (P = .04338). However, investigators noted that the improvement was modest. The number of conditions meeting the previously proposed benchmark of 12% dark-skin representation increased from nine conditions in 2020 to 11 conditions in 2024.

Notably, 19 dermatologic conditions still returned no dark-skin images in Google searches, a figure unchanged from the earlier analysis. The numbers were even higher for Yahoo (31 conditions), Bing (32), and DuckDuckGo (26).

The authors noted that Google has publicly committed to improving skin-tone representation and has incorporated resources such as the Monk Skin Tone Scale and partnerships with organizations including the American Academy of Dermatology and VisualDx. Yet the limited gains observed suggest broader challenges extending beyond search algorithms alone.

The findings point to a continuing shortage of diverse dermatologic imagery available online and highlight the importance of increasing representation of darker skin tones in educational resources, image databases, and source websites that inform search engine results.

"While there is still significant under-representation, we remain hopeful that future results can further the inclusivity of all skin tones," the authors wrote.

Source

Adler R, et al. Journal of Drugs in Dermatology. 2026. 2026;25(7):9527.