I tested GPT-5.2 and the AI model's mixed results raise tough questions

I tested GPT-5.2 and the AI model's mixed results raise tough questions ...
Microsoft

MedFuzz: Exploring the Robustness of Large Language Models in Medical Question Answering

Large language models (LLM) have achieved impressive performance on medical question-answering benchmarks. However, high benchmark accuracy does not imply that the performance generalizes to ...