November 9, 2026
The Russian doctor and pathologist, Arina Lehman, warned of the dangers of relying entirely on artificial intelligence systems to diagnose medical diseases, stressing the necessity of subjecting the conclusions of these technologies to direct scrutiny by doctors before confirming the final diagnosis.
Lehman explained, in statements reported by the “Russia Today” website, that research was conducted on 200 tissue scans of biopsy samples and surgical materials, comparing the results of two artificial intelligence models and the reports of pathologists. The experiment, which evaluated the samples using digital images taken with three different scanners, showed that agreement with doctors’ conclusions reached only 8% in the absence of patient information, and rose to 30% when identifying the organ from which the sample was taken, reaching 43% when clinical data was included.
Lehman emphasized that providing additional information, such as the source of the tissue and the reason for the study, significantly improves the performance of the models, but neural networks still make mistakes in several cases, which necessarily requires comparing the outputs of artificial intelligence with the clinical picture and examination data accurately.
She also pointed out that the digitization technology itself affects the results; The form may provide different answers for the same sample if it is scanned using different devices.
The Russian specialist concluded that artificial intelligence currently plays the role of an effective assistant in suggesting diagnostic options and verifying results, while emphasizing that the final decision must remain exclusively with the treating physician.
