Are Med Students Cheating if They Use AI?
[post_content]
Disclaimer: This article has been automatically aggregated from
LOS ANGELES — Despite assertions to the contrary, students who use artificial intelligence (AI) in medical school assignments or elsewhere are not cheating, according to one professor here at the Association of Health Care Journalists Fall Summit.
“AI is not plagiarism,” said Graciela Gonzalez-Hernandez, PhD, who is also vice chair for research and education in computational biomedicine at Cedars-Sinai Medical Center in Los Angeles. Working with AI “is doable, but it has to be done properly.”
She said she makes sure the academic program she runs is AI-friendly. When giving an assignment, she tells students, “‘Here is how I want to see your prompts constructed … Let’s see the paragraph that comes out and let’s work on it together.'”
“If students spend their time, their energy, and their intellectual capacity on how I’m going to cheat my professor and make it look like I didn’t use AI, we’re just wasting our time,” she added. “They’re going to use AI. You can teach these things. It’s going to be part of education. Bring that back to the classroom [discussion] instead of having 250 people just sitting there.”
One criticism of using AI in medical school is that it might cause students to lose their ability to perform clinical reasoning on their own, said session moderator Eli Cahan, MD, MSc, an investigative journalist at Rolling Stone and a fellow in neonatal intensive care at Stanford University in California. The question is, “is clinical reasoning one of those things we need doctors to be good at?” he said.
For example, “for Christopher Columbus, being able to navigate was a really important skill, but now we have GPS,” he continued. “Are there forms of clinical reasoning we can cede to AI for being much more accurate? … I’m withholding judgment until we see whether and what kind of clinical reasoning actually is relevant for doctors to do. There is a balance between valuing that as a core property for a doctor and being precious about it.”
Tina Hernandez-Boussard, PhD, a professor of biomedical informatics also from Stanford University, warned about the downside risk of de-skilling from AI use. She mentioned that a group of medical residents recently went to Uganda to help out at a particular hospital’s cardiac ward. However, they lost electricity and there was no AI to read EKGs.
“The residents were not able to synthesize the information in a way they were confident to make decisions on the patients,” Hernandez-Boussard noted. “That’s a real issue to think about … So we still have to ensure our trainees still know how to synthesize that information.”
De-skilling is definitely a potential issue with using AI in medical school, Gonzalez-Hernandez agreed. “We have to have the capacity of training our physicians better by removing technology,” she said. “We have to make sure there are scenarios that are run with paper and pencil … I have heard stories of new residents not knowing how to interpret the heart directly.”
On the other hand, AI can help physicians-in-training by adding a layer of objectivity, she said. “In the [chart] notes, you can tell how in favor or against or doubtful the physician” is by the way the note is written. For example, a note might say either “The patient reports having constant back pain” or “The patient has had constant back pain for a week.”
“When they detach themselves from the expression there, it means something,” said Gonzalez-Hernandez, referring to the first note. “The doubt goes down the line and will affect outcomes eventually.” Critically, “this sort of language varies and is not used as often for one population as compared to another,” she added. “There is doubt implicit from the get-go in treating a patient of a certain demographic.”
And that is where the AI could provide some balance, according to Gonzalez-Hernandez. “Just as we should question AI when we get answers from a chatbot …. AI can push back: ‘Have you asked about this?'” she said. “The questions missed will be [added] to re-balance the equation from a neutral point of view.”
“We have the real possibility of shaping the way our physicians benefit from the use of AI,” said Gonzalez-Hernandez. “Physicians shape [AI] knowledge with their own training. … We want to reach a balance where AI allows physicians to look at all these things quickly, and aim at things they might have missed otherwise.”
for informational purposes only. We do not claim ownership, accuracy, or liability for the content provided. All rights belong to the original publisher.