Researchers reviewed studies about using AI language models to help train primary care doctors in eye care. These studies examined whether AI could help with education, clinical reasoning practice, and patient support tasks. The review focused on research involving primary care physicians, though specific study sizes weren't reported.
The main finding was that AI models could act as practice partners for doctors, helping with skills like patient triage reasoning, virtual interviews, and creating educational materials. However, nearly all the evidence comes from controlled tests using written medical scenarios, not from real clinics with actual patients. Some reports even noted that adding real patient photos to the AI's analysis sometimes made it less accurate.
There are important reasons to be cautious. The studies varied widely in how they tested the AI, making results hard to compare. The AI technology changes very quickly, so today's findings might not apply tomorrow. Significant safety risks exist, including the AI 'hallucinating' or making up incorrect medical information, showing bias, or causing doctors to rely on it too much without checking its work.
Readers should understand this research shows early potential, not proven effectiveness. For now, these AI tools might one day help make eye care training more accessible, but they should only support—not replace—a doctor's expert judgment. Much more testing in real-world medical settings is needed before we know how safe and useful they truly are.