A 10-year-old CNN outperforms frontier LLMs for skin cancer diagnosis
Researchers evaluated a ten‑year‑old convolutional neural network against the latest large language models for detecting skin cancer from dermoscopic images. The older CNN achieved higher accuracy and lower false‑positive rates than the frontier LLMs, demonstrating that purpose‑built vision models can still outperform more generalist AI systems in specialized medical tasks. The findings highlight the continued relevance of established image‑analysis architectures.