Conducted with 50 medical experts, the research revealed that OpenAI's chatbot achieved a 90% diagnostic accuracy rate. This was higher than the 76% accuracy of doctors using ChatGPT and the 74% accuracy of doctors working without the AI tool.

Dr. Adam Rodman, an internal medicine specialist at Beth Israel Deaconess Medical Center in Boston and one of the study participants, expressed surprise at the findings. While he anticipated that AI tools like ChatGPT could aid doctors in diagnosing diseases, he did not expect such a small improvement in accuracy for doctors using the AI compared to those without it.

The study involved six anonymized case histories based on real, previously unpublished patient cases. Doctors were tasked with diagnosing these cases, with their conclusions evaluated by independent medical experts who only saw the participants' answers. This setup allowed comparisons between diagnoses made by ChatGPT alone, doctors without AI assistance, and doctors using ChatGPT.

Challenges in Using AI Tools:
One significant challenge was the difficulty many doctors faced in fully leveraging ChatGPT's capabilities. Many medical professionals disregarded the chatbot's suggestions when they disagreed with them, often believing their own expertise surpassed the AI's insights.

Additionally, most doctors treated ChatGPT as a basic search engine, asking it direct questions rather than exploring its full diagnostic potential. Only a few participants realized they could upload entire case histories and request comprehensive analyses from the AI.

The study highlights ChatGPT's potential as a diagnostic tool while also emphasizing the importance of training medical professionals to use AI effectively. With proper integration and usage, tools like ChatGPT could revolutionize diagnostic accuracy and support doctors in providing better patient care.