Core Insights - The integration of AI in healthcare is progressing, with large pre-trained language models being implemented in hospitals across China, indicating a shift from theoretical applications to practical use in clinical settings [1][2][3] - The market for medical large models is expected to grow significantly, with projections estimating a market size of nearly 2 billion yuan by 2025 and an annual growth rate of 140% [3] - Despite advancements, the transition from medical large models to fully autonomous AI doctors faces challenges, including technical limitations, data availability, and societal acceptance [4][8][10] Group 1: Medical Model Implementation - Major hospitals in China have begun using AI models for diagnostics, with examples of successful applications in pediatric care demonstrating improved diagnostic accuracy [2][3] - Policies supporting AI in healthcare have been introduced, including guidelines for AI-assisted diagnosis and integration into medical service pricing [2][3] - The AI system "智医助理" has been deployed in over 75,000 grassroots medical institutions, providing over 1 billion diagnostic suggestions, thereby alleviating the burden on healthcare professionals [3] Group 2: Challenges and Limitations - The definition of AI doctors remains ambiguous, with a distinction made between medical large models and AI doctors, the latter requiring practical clinical experience [6][7] - Technical challenges persist, such as the "black box" nature of models and the risk of incorrect diagnoses, which can lead to serious consequences for patients [8][9] - Data scarcity and fragmentation hinder the development of effective medical large models, particularly in rare disease diagnostics where accuracy is often below 60% [9] Group 3: Societal Acceptance and Future Directions - Public skepticism towards AI in healthcare remains, with patients often preferring human doctors despite the capabilities of AI models [10] - Experts suggest that enhancing the credibility of AI through clinical outcomes, research publications, and transparent methodologies is essential for gaining acceptance [14] - Recommendations for policy adjustments include simplifying AI product registration processes and integrating AI services into health insurance systems to foster collaboration between medical institutions and technology companies [15]