AI Tools May Increase Doctors’ Workload, Dartmouth Study Finds
A recent study from Dartmouth College suggests that incorporating artificial intelligence into medical practice might actually lead to a higher workload for some doctors. This is surprising, given the common belief that AI tools are designed to streamline tasks.
According to the research, AI can complicate doctors’ tasks by introducing errors and adding irrelevant details in communication with patients. The findings were shared at the Computational Linguistics Association Annual Meeting in San Diego. The study reveals that physicians might be spending more time correcting AI-generated responses than they would take to craft their messages from scratch.
Sarah Preum, an assistant professor at Dartmouth, pointed out that while AI can mimic a doctor’s writing style, it lacks true understanding. Preum is one of the co-authors of the study, collaborating with Parker Seegmiller, a doctoral candidate.
The researchers analyzed responses produced by various AI platforms like Claude, Google Gemini, and ChatGPT, noting discrepancies between AI-generated content and actual physician writing styles. Common errors included overly long responses, missing essential follow-up questions, and irrelevant or inaccurate medical information.
“We found gaps in the existing literature about large-scale evaluations of AI’s efficiency in medical settings,” Preum remarked. Their goal was not just to assess accuracy but to evaluate whether AI actually lightens doctors’ responsibilities, focusing particularly on editing time.
As AI technology evolves, it has increasingly been adopted in the healthcare sector. A report from the American Medical Association indicates that 81% of physicians are now using AI tools in their practices.
However, concerns persist; a separate survey revealed that a significant portion of U.S. doctors is wary of how AI is affecting prior authorization processes, fearing it may lead to greater patient harm and inefficiencies. Public sentiment also reflects this trend, with a KFF study showing that 32% of adults turn to AI for medical advice and information.
