The Hidden Cost of Automation or The Silent Loss of Competence by Dr. Timothy Smith

Photo Source: Picryl

Last week, I learned firsthand about the dangers of deferring judgment to a machine, and this made me think about the implications beyond my own kitchen. I used a new digital thermometer while preparing a roast, one that I have made many times before. The timing, color, and feel of the roast looked perfect, but the thermometer indicated the roast needed more time in the oven. Instead of following my better judgment and experience, I deferred to the thermometer. The roast came out overdone due to a calibration issue with the new thermometer, which I neglected in my rush to try out the new device. My kitchen mishap revealed a troubling willingness on my part to surrender my judgment to an algorithm despite my years of experience.

Not just amateur cooks with uncalibrated thermometers face the predicament of trading experience for machine certainty. We are witnessing a growing phenomenon of deference to machines, particularly as artificial intelligence becomes increasingly embedded in high-stakes professions where human expertise literally means the difference between life and death. A groundbreaking study published recently in The Lancet Gastroenterology & Hepatology provides an initial set of data supporting the finding of what researchers call "deskilling." (thelancet.com)

Deskilling refers to the erosion of human expertise when we become too dependent on automated assistance. The study titled "Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multicentre, observational study", conducted in Poland, tracked what happened when doctors who had grown accustomed to AI assistance during colonoscopies suddenly had to perform the procedures without AI assistance. The results revealed a marked drop in adenoma detection rates, from around 28% to 22%, when clinicians worked unassisted by AI. The finding represents approximately a 20% decline in clinicians' diagnostic capability to detect precancerous polyps. Over thousands of patients, the 20% drop in adenoma detection would put many patients at risk of a missed diagnosis and a chance at life-saving therapy. Moreover, the deskilling occurred over the course of two years, not decades, of dependence on AI. What makes this particularly concerning is the speed at which this skill degradation occurred. The doctors weren't relying on AI for years; they experienced measurable skill loss after just months of assisted practice. It suggests that our cognitive abilities, like muscles, can atrophy remarkably quickly when we stop using them actively.

Not just the medical field grapples with dependency on automation. Aviation has its own challenges, as autopilot systems have become increasingly sophisticated and widespread. The tragic 2013 crash of Asiana Airlines Flight 214 in San Francisco serves as a case study. The Boeing 777 pilots relied heavily on the aircraft's automated systems, including the auto-throttle, to maintain airspeed. They failed to notice the plane was flying too slowly and lost too much altitude on the approach to the runway. Planes approach the San Francisco airport over water, and the plane hit the seawall before it could land on the runway. The aircraft crashed and caught on fire, resulting in three deaths and hundreds injured. Post-accident investigations revealed that the pilots had become so accustomed to automated flight systems that their fundamental manual flying skills had significantly eroded. The aviation industry has since implemented policies requiring pilots to practice manual flying techniques regularly; however, the temptation remains to rely more on automation as it continues to improve.

These examples point to a "confidence paradox" of automation. The very systems designed to enhance our capabilities can diminish our confidence in our own judgment and erode the expertise they were meant to augment. The AI system in the colonoscopy study has improved diagnostic accuracy overall; autopilot systems have improved aviation safety. However, the problem arises when we mistake these tools for replacements rather than supplements to human judgment. We need to find ways to keep our judgment and skills sharp in an environment of increased automation. Just as people now regularly visit the gym to maintain their physical strength and health, we need to find ways to keep our minds and skills sharp, even in the face of the ever-growing quality and sophistication of automation, to know when to step in and take control confidently.

Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high-level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.

You can buy his book on Amazon in paperback and in kindle format here.