Khaberni - Researchers at Harvard University evaluated the efficiency of artificial intelligence systems and their ability to sort symptoms, request appropriate tests, assess new patient information, and reach the correct conclusion about their problem.
It was found that advanced artificial intelligence models struggle to update judgments in response to new and uncertain information, and often fail to recognize when some information is completely irrelevant.
According to "Medical Express", the researchers said artificial intelligence may excel in selecting appropriate medical tests, but it still stumbles when faced with changing clinical information.
Comparison of Artificial Intelligence and Human Thinking
Interestingly, medical students who perform well in multiple-choice exams do not always achieve the same good results in assessments based on textual information because it is a completely different skill.
Dr. Liam McCoy, the co-researcher, stated: "It's important to recognize that performance in a task like clinical reasoning is very complex and task-specific."
This does not imply that artificial intelligence models cannot be improved to enhance their performance. In fact, McCoy sees the technology as enduring, so it falls upon researchers like him to continue working on improving it.
He explains: "This technology is coming one way or another, so I think as doctors, we need to make sure that it's effective, fair, and compliant."
He continues: "We need to develop artificial intelligence that meets the needs of patients, instead of letting it be managed by external parties."
McCoy said: "Artificial intelligence models perform exceptionally well on multiple-choice questions, but we are still far from the stage where a patient can safely enter the room, activate an artificial intelligence model assistant, and complete the whole visit."
