Rehan Ahmed

When the pieces leave the room, who owns the patient? Photo generated by AI We teach the eye exam as a litany. Vision, pupils, pressure, motility, fields, slit lamp, fundus. We sort it into tiers—intermediate and comprehensive—and we bill it that way: 92002 and 92004 for new patients, 92012 and 92014 for established ones, each defined by a checklist of “elements” a payer can audit. To a trainee, the comprehensive exam looks like a clinical unit, a single indivisible act of medicine.

Photo Credit: Dreamstime Photos Imagine you’re evaluating an autonomous diabetic retinopathy screening platform for your practice. You ask AI to summarize the benefits. You get a clean list: improved access, earlier detection, reduced specialist burden, better consistency. The list isn’t wrong. But it isn’t enough. We’ve spent the last two years teaching clinicians how to prompt AI — be specific, give context, define the audience, ask for a table, ask for a patient-friendly summary.

Photo Credit: Dreamstime Photos If you’ve felt a little dizzy keeping up with AI lately, you’re not alone. In the past year alone, we’ve seen a flood of new large language models—GPT-4o, Claude 3, Gemini, Llama, Mistral and more—each arriving with breathless announcements about what it can do. New terminology is multiplying just as fast. Agentic AI. Retrieval-augmented generation. Multimodal models.