When AI and Experts Agree on Error: Intrinsic Ambiguity in Dermatoscopic Images
Loris Cino, Pier Luigi Mazzeo, Alessandro Martella, Giulia Radi, Renato Rossi, Cosimo Distante · 2026
Read When AI and Experts Agree on Error: Intrinsic Ambiguity in Dermatoscopic Images by Loris Cino, Pier Luigi Mazzeo, Alessandro Martella, Giulia Radi, Renato Rossi, Cosimo Distante (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13302442.txt.
Description
Loris Cino and colleagues present a 2026 study of skin lesion images that both artificial intelligence models and human experts get wrong. Several neural network architectures were trained on dermatoscopic images, and a subset of images was misclassified by every model far more often than chance would explain. Expert dermatologists then judged those hard images alongside a control set, to test whether the errors came from the algorithms or from the images themselves.
On the images the models had missed, expert agreement with the true labels collapsed, with a kappa score of 0.08 against 0.61 for the controls. Agreement among the physicians themselves also fell sharply on that subset. The team traced much of the shared failure to poor image quality, and released its data, code, and trained models publicly so others can probe where diagnostic images become genuinely ambiguous.
Catalog record
| CREATOR | Loris Cino, Pier Luigi Mazzeo, Alessandro Martella, Giulia Radi, Renato Rossi, Cosimo Distante |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |