AI & ML · Oct 26
GIGAZINE: Dr. Magnus Lynch's 10-Year-Old CNN Beats Frontier LLMs on Skin Cancer Images
The surprise is not nostalgia. It is a lesson in why domain-specific training still matters when general models bring bazookas to a biopsy.
- Benchmark frontier models against tuned specialist baselines before trusting general-purpose performance on narrow visual tasks.
- Use domain-matched data and clear label spaces when evaluating medical image AI.
- Treat this as an engineering signal, not a clinical verdict, until broader replication arrives.