On August 31st， a study published in The Lancet Gastroenterology & Hepatology has garnered significant attention.
The research indicates that doctors who have become long-term reliant on Artificial Intelligence (AI) assistance may experience a notable decline in their independent diagnostic skills within a short period. This finding serves as a cautionary note for the over-reliance on new technologies in the medical field.
This collaborative study， conducted by several European institutions， primarily took place in four endoscopy centers in Poland that participated in the “AI for Cancer Prevention in Endoscopy – ACCEPT” trial. The research team collected data from 2177 colonoscopies performed between September 2021 and March 2022. All procedures were carried out by 19 experienced endoscopists who had each performed over 2000 colonoscopies. Among these， 1433 examinations were performed without AI assistance.
Colonoscopy is a crucial method for detecting and removing colon polyps， thereby preventing the occurrence of intestinal cancer. Previous studies have confirmed that AI assistance can improve polyp detection rates and reduce the risk of colorectal cancer.
To investigate the impact of AI use on physicians’ skills， the research team compared the quality of colonoscopies conducted in the three months prior to AI implementation with those performed in the three months afterward. The examinations were conducted randomly， with and without AI assistance. In the examinations without AI assistance， 795 procedures were performed before the routine use of AI， and 648 were performed after the introduction of AI tools.
The experimental results showed that in the three months before AI introduction， the adenoma detection rate (ADR) was approximately 28.4% (226/795). Three months after the introduction of AI， when doctors performed examinations without AI assistance， the ADR dropped to 22.4% (145/648)， representing a relative decrease of 20% and an absolute decrease of 6% in adenoma detection rates.
This suggests that while AI can enhance a doctor’s detection rate when used as an aid， physicians’ diagnostic skills may actually deteriorate once the AI assistance is removed.
Researchers analyzed that this phenomenon might be attributed to a reduction in doctors’ proactivity， focus， and sense of responsibility after prolonged reliance on AI assistance. This could lead to decreased confidence in their own judgment during cognitive decision-making processes without AI support， consequently impacting their diagnostic capabilities.
This situation has been described by the research team using the “Google Maps effect，” analogous to how individuals who consistently rely on navigation software may lose their ability to navigate independently when the software is unavailable.
These findings stand in contrast to the views of Google CEO Demis Hassabis， who stated， “All diseases will be cured in the next decade，” and to the optimistic projections by some that AI will be instrumental in finding cures for cancer.
This study prompts new considerations for the healthcare industry. Alongside the vigorous promotion of AI applications in medicine， a critical question that urgently needs addressing is how to prevent doctors from becoming overly dependent on AI， ensuring the continuous improvement and stable performance of their own skills.
In the future， medical practitioners and relevant institutions may need to re-evaluate the role of AI in medical processes and strike a balance between technological application and the development of physicians’ professional competencies by establishing scientific training and regulatory mechanisms.
