AI让医生6个月“废功”20%，癌症检出率崩盘

Core Insights - A recent study published in The Lancet reveals that over-reliance on AI tools by doctors can lead to a significant decline in their skills within a few months [1][10][22]. Group 1: Study Findings - The study involved 19 experienced doctors who had performed over 2000 colonoscopies, indicating that even seasoned professionals are not immune to skill degradation when relying on AI [4][30]. - The research showed that the adenoma detection rate (ADR) dropped from approximately 28.4% to 22.4% after three months of using AI, representing a relative decline of 20% and an absolute decline of 6% when AI assistance was removed [19][20][21]. - The study was conducted across four endoscopy centers in Poland, collecting data from 2177 colonoscopy procedures between September 2021 and March 2022 [12][11]. Group 2: Implications of AI Dependency - The authors of the study suggest that the primary reason for skill decline is the human tendency to overly rely on decision support systems like AI, which can dull pattern recognition abilities [24][28]. - This phenomenon has been termed the "Google Maps effect," where individuals lose traditional skills due to dependence on technology [25][26]. - The study raises concerns that regular use of AI may lower diagnostic confidence and degrade manual skills in endoscopy, particularly affecting less experienced doctors more severely [29][30]. Group 3: Broader Context and Reactions - The findings have sparked discussions on platforms like Reddit, where medical professionals express concerns about AI's limitations in replacing human judgment and the potential risks of skill degradation [31][33]. - The analogy of "autonomous driving" is drawn, suggesting that as AI performance improves, human vigilance may decrease, leading to critical oversights in medical practice [39].