Dependence on AI in Medicine Raises Concerns Over Loss of Critical Skills: Study

Artificial Intelligence (AI) is rapidly finding its way into routine medical screenings, with gastroenterologists in Poland being one of the latest groups to adopt the technology. However, a recent study published in Lancet Gastroenterology and Hepatology raises concerns about the potential for excessive reliance on AI by medical professionals.

The research found that after using an AI-assisted system for polyps detection during colonoscopies, the participating gastroenterologists showed a decrease in their ability to identify polyps and other abnormalities when the AI was deactivated. The study’s lead author, Dr. Marcin Romańczyk from H-T Medical Center in Tychy, Poland, expressed surprise at this finding.

While some experts are alarmed by these results, others question whether they truly indicate a loss of critical skills due to AI dependence. Johan Hulleman, a researcher at Manchester University in England, who has studied human reliance on artificial intelligence, suggests that statistical variations in patient data might explain the apparent drop in detection rates.

As AI becomes increasingly common for various medical scans, it is likely that you will encounter its use during your next eye, breast cancer, or colon disease scan. Many doctors are still playing catch-up as learning how to utilize this technology was not part of their initial training.

“We’ve been taught from books and from our teachers,” says Dr. Romańczyk, “No one told us how to use AI.”

A few years ago, four Polish clinics trialed an AI system for detecting polyps during colonoscopies. The AI works in real-time, analyzing video from a camera inside the colon and highlighting potential areas of concern for doctors.

However, when Dr. Romańczyk and his colleagues reanalyzed the data collected during this trial, they discovered an unexpected trend: after the introduction of the AI system, doctors showed significantly lower detection rates for potential polyps when the AI was deactivated. The researchers found that detection rates dropped from 28.4% to 22.4%. This suggests a quick dependency on AI systems to identify polyps.

One theory as to why this may be happening is that doctors are subconsciously waiting for the AI’s markings to indicate potential polyps, causing them to pay less attention when the system is deactivated. Supporting evidence can be found in a similar study examining mammogram scans.

While the study raises valid concerns about potential AI dependency among medical professionals, some experts, like Johan Hulleman, remain skeptical. He suggests that statistical variations in patient data might contribute to the apparent drop in detection rates. Additionally, he points out that it is unclear how many of the potential polyps missed by doctors were medically significant.

Despite these concerns, Dr. Romańczyk believes that the decrease in detection rates is a real phenomenon and emphasizes the need for more studies examining the impact of AI on medical practice. “We have AI systems that are available, but we don’t have the data,” he says. This lack of information about how doctors interact with these new technologies is crucial to understanding their potential impacts on patient care.