ChatGPT Accurately Diagnoses Psychiatric Patients in Study

Researchers at the University of Melbourne in Australia set out to explore whether ChatGPT 3.5 could assist psychiatrists. They tasked the neural network with analyzing patient case histories to devise treatment plans.

Two seasoned psychiatrists, both university instructors, evaluated the AI’s responses. They graded ChatGPT on a four-point scale, from A (excellent) to D (poor).

In 61 out of 100 cases, ChatGPT earned the highest mark, A. It received a B in 31 cases and a C in 8. Notably, not a single diagnosis from the neural network was deemed incorrect. The model shone brightest in crafting treatment plans, identifying symptoms, and making diagnoses.

Lead researcher Russell Frank D’Souza emphasized the findings: «ChatGPT demonstrates substantial knowledge and interpretive skills in psychiatry.» He believes the neural network holds transformative potential for the field of medicine.