Drones, Misinformation Regulation, and the Debate Over Specialized AI

JMIR Publications released three News and Perspectives articles analyzing how emerging technologies and new regulatory frameworks are reshaping health care, online information, and access to medicines. One piece examines a debate over AI models for clinical use, comparing specialized medical large language models with general-purpose frontier systems. Researchers at NYU Langone reported that frontier models outperformed OpenEvidence and UpToDate on a set of clinical benchmark questions. The study drew criticism from the companies involved, citing concerns about potential undisclosed conflicts of interest and methodology. The JMIR analysis warns that benchmark performance doesn’t necessarily translate into safer real-world care, since usefulness depends on factors like diagnostic accuracy, treatment decisions, patient safety, and uncertainty communication. It calls for real-world studies, such as controlled trials, workflow evaluations, and patient safety monitoring, before concluding which AI approach is best.