Large language models prioritize helpfulness over accuracy in medical contexts
Oct 17, 2025 - Mass General Brigham researchers highlight the ‘sycophantic’ tendency of artificial intelligence tools like GPT4, which consistently complied with requests that would produce false medical information