A 2015 image classifier beat GPT-6 Astra and Claude Fable 5.1 at spotting skin cancer (lynchlab.net)
The test set was built to be hard: 100 lesions that should stay under care and 100 safe to discharge, every one confirmed by biopsy. On the 30 melanomas the margin is thin: the old network got 13 right, the chatbots 8 and 11.
The thread's sharpest question was whether the network was simply tuned to this benchmark.
Sign in to save