0
0
A recent large observational study published in The Lancet Gastroenterology & Hepatology has found that routine use of artificial intelligence (AI) to assist endoscopists in colonoscopies may inadvertently cause a decline in their skill levels when performing the procedure without AI assistance. This 2025 study in four Polish centers showed a 6% absolute decrease in adenoma detection rates (ADR) in non-AI procedures after clinicians had prolonged exposure to AI tools, highlighting a potential risk of over-reliance on AI technology with implications for patient outcomes.
Key Findings and Developments
The study analyzed 1,443 standard (non-AI-assisted) colonoscopies and compared adenoma detection rates before and after the introduction of AI-assisted colonoscopy at four centers. Experienced endoscopists, each having performed more than 2,000 colonoscopies, displayed a significant drop in ADR from 28.4% before AI exposure to 22.4% after AI introduction—a 20% relative decrease and 6% absolute decline in detecting precancerous polyps. Interestingly, the reduction was more pronounced in centers with initially higher ADRs and certain patient groups such as women under 60 years.
Despite AI-supported colonoscopy previously showing improvements in ADR by 5%–20% in randomized trials, this study raises concerns about “deskilling,” where clinicians’ core skills may erode due to overdependence on technology.
Expert Commentary
Krzysztof Budzyń, MD, and Marcin Romańczyk, MD, lead authors from the Academy of Silesia in Poland, underscored that the decline in skill was unexpected given the clinicians’ high experience level. They emphasized that AI should supplement, not replace, clinician vigilance, encouraging practitioners to consciously maintain active engagement even when AI tools aid detection.
Omer Ahmad, MD, a gastroenterologist at University College London, noted in a related editorial that this study “compels us to carefully consider the effect of AI integration” in medicine. While AI has tremendous promise to improve clinical outcomes, there is a critical need to protect against erosion of fundamental diagnostic skills required in endoscopy and other medical fields.
Rajiv Bhuta, MD, from Temple University, described the findings as a “signal” consistent with human tendencies: when machines perform skill-based tasks, human ability may decline through reduced attention and active participation. He pointed out the need to define boundaries of AI’s role to balance gains and losses in skill across healthcare.
Context and Background
Colonoscopy is essential for detecting and removing adenomas, which if left untreated may progress to colorectal cancer. AI tools, including FDA-cleared systems like Medtronic’s GI Genius and Fujifilm’s AI endoscopy technology, use computer-aided detection (CADe) to highlight potential polyps in real time. This has led to widespread optimism about AI’s role in improving cancer prevention.
However, most prior studies focused on immediate AI-assisted improvements, lacking data on the long-term impact of AI use on clinician skill retention when AI is not present.
Implications for Public Health
This study signals a critical challenge as AI tools become entrenched in clinical practice across specialties. While AI can boost the quality of care and efficiency, overdependence risks impairing clinicians’ independent diagnostic skills essential for situations where AI may be unavailable—such as system failures or emergencies.
Healthcare systems and educators must develop training and monitoring strategies to preserve endoscopists’ core capabilities alongside AI adoption. This may include regular audits of non-AI procedure outcomes and reinforcing hands-on skill retention to safeguard patient safety.
Limitations and Counterarguments
The study’s retrospective observational design and limited timeframe (several months post-AI introduction) mean causality cannot be firmly established. Factors such as increased workload after AI adoption and new clinical workflows might have contributed to ADR declines. Additionally, some doctors maintained better performance, indicating variability in adaptation.
Furthermore, the magnitude of decline relative to natural fluctuations in detection rates remains to be clarified with larger, longer-term research.
Practical Takeaways for Readers
For patients undergoing colonoscopies, AI-assisted technology represents a valuable tool to improve polyp detection and cancer prevention. However, clinicians and health systems should ensure that AI functions as a partner rather than a replacement for human expertise.
For medical professionals, conscious effort to stay engaged and attentive during procedures, regardless of AI assistance, is vital for maintaining high standards of care.
Medical Disclaimer:
This article is for informational purposes only and should not be considered medical advice. Always consult with qualified healthcare professionals before making any health-related decisions or changes to your treatment plan. The information presented here is based on current research and expert opinions, which may evolve as new evidence emerges.
References
