In the study cited by Quartz, a group of 50 doctors were asked to try to make six different diagnoses of medical conditions with the same medical history. In the study 26 attending physicians and 24 residents participated. Some participants received OpenAI’s ChatGPT to help them make decisions. Another group made diagnoses without artificial intelligence support.
Surprising research results
The test result was quite surprising. Doctors who completed the project without AI achieved an average score of 74%.., while doctors who used AI, their average result was only 2 percentage points higher. higher (76 percent.). As it turned out, artificial intelligence achieved a much better result. Alone ChatGPT achieved an average score of 90%.
Dr. Rodman, who helped design the study, told the New York Timese that he was shocked by the study’s results. Rodman expects that chatbots will significantly help doctors make the right decisions. Meanwhile, the study found that AI did little to improve doctors’ scores. Rodman was shocked again when he learned about the results ChatGPT. When making its own diagnosis, artificial intelligence had much better results than real doctors.
Doctors did not listen to AI suggestions
The study found that artificial intelligence did not help doctors as much as expected. The reasons were twofold. First, doctorsThey didn’t listen to the AIwhen she told them things they didn’t agree with,” Rodman said. He also added that most of the doctors were there attached to her own diagnoses and it was impossible to convince that the chatbot knew more than them.
Another problem that came up was that many doctors did not know how to take full advantage of artificial intelligence.
