Researchers recently took random sample of 200 patient questions from the AskDocs forum on Reddit.
The asked both human doctors and ChatGPT to provide answers to the questions. A clinical team assessed the responses, focusing on the quality of the answers and the level of empathy demonstrated.
ChatGPT emerged as the clear winner, with evaluators favouring its responses nearly 80% of the time over the human doctors.
The insight here is not that doctors aren't empathetic. It's that ChatGPT has unlimited "time" to craft a proper response. Unlike humans, generative AI isn't limited by fatigue or effort perception. Patients often equate the amount of time spent explaining their diagnoses with the quality of care, as it fosters a feeling of empathy and clarity. Human doctors, used to running on tight schedules and billing on a time basis – are trained to be more blunt and straightforward with their bedside manner. The outcome is a patient perceived quality gap.
This offers up an interesting thought experiment – what other things in our day to day lives will benefit from more time? The ability to craft thoughtful, polite employee reviews from bullet points is an interesting use case. Or a CFO, leveraging GPT-4 to instantly generate and chart financial trends from a data set, so she can spend more time on diagnoses and interpretation.
The inverted way of looking at this is: What parts of my work would improve if I had twice or even ten times the usual time to spend on the delivery?