Train derailed, rattling Quebec town | CTV News Montreal at SIX for July 6, 2026

AI systems, including ChatGPT and Gemini, demonstrate significant limitations when used for medical purposes, with research showing ChatGPT making approximately 30 errors per medical note (including 10% errors of commission) and achieving only 60-70% accuracy on structured medical questions about complex topics like menopause, while Gemini performed even worse at only 30% accuracy; these systems are better suited for specific factual queries and bureaucratic tasks rather than nuanced medical diagnosis or treatment recommendations.