After ChatGPT's AI platform successfully answered test questions on the US Medical Licensing Exam, Josh Tamayo-Sarver, who works in an emergency room, put it to the test.

As Presshub reports, if the doctor had fully trusted artificial intelligence, the diagnosis could have ended fatally for one of his patients.

The doctor, who is vice president of clinical innovation at Inflect Health, entered into the program the case histories of 40 people he had treated himself over time with symptoms that brought them to the emergency room.

“The chatbot did a good job of analyzing common diagnoses as long as everything I gave it was accurate and very detailed.

For about half of my patients, ChatGPT suggested six possible diagnoses, and the diagnosis I believe to be correct after a full evaluation and testing was among the six it suggested," noted Josh Tamayo-Sarver.

At the same time, artificial intelligence could not determine the woman's ectopic pregnancy, which the patient did not know about.

If this condition is diagnosed too late, it can be fatal.

In a real situation, the doctor operated on a young woman in time and saved her.

The AI ​​failed to correctly identify the situation.

“I'm afraid countless people are already using ChatGPT to self-diagnose instead of seeing a doctor.

If my patient in this case had done that, ChatGPT's response could have killed her," warned Josh Tamayo-Sarver.

After testing the effectiveness of this AI platform in real life for patient diagnosis, the doctor concluded that until this technology becomes more powerful, people should have realistic expectations about its limitations.

We will remind you that in the USA a model of artificial intelligence passed the exam for obtaining a medical license.

This is a test that must be taken by doctors before starting practice.

Read also:

• World news

• War in Ukraine

• News of Ukraine: video