Are robots capable of managing the emotions of their human interlocutors more effectively? This is suggested by a study conducted by researchers at the University of San Diego in California (USA). His work has been published in the specialized magazine JAMA Network, says Axios.
To verify which of the humans or ChatGPT was the best doctor, more than 200 medical questions were sent to the doctors and the OpenAI chatbot, powered by artificial intelligence. The clinical teams then analyzed the responses for both their quality and the empathy demonstrated.
“Sorry for your headache”
The artificial intelligence was hailed during the evaluations. In nearly four out of five cases, responses from ChatGPT were preferred over responses from a human doctor. But for research director John Ayers, this preference is explained by the very nature of the software.
So, whenever a patient mentions a headache, ChatGPT immediately responds, “sorry about your headache.” Epidemiologist John Ayers believes that a doctor does not have time to address this phrase to his patient. And this, even though he feels empathy for his patient.
A narrowly failed exam
In February, ChatGPT narrowly failed a tough medical exam in the United States. If the replacement of health professionals by artificial intelligence is therefore not on the agenda, it is still a very effective assistant. The technology has already proven successful in predicting sudden death cases.
Artificial intelligence was also able to detect the origin of a man’s cancer by 90%. Until now, his medical team has not been able to confirm this origin with certainty despite more than two years of follow-up.
Source: BFM TV
