Artificial intelligence (AI) is rapidly evolving in the field of GI endoscopy. This systematic review aims to summarize the current perspectives of endoscopists on AI in endoscopy and to identify its challenges.

Electronic databases were searched to identify studies conducted on endoscopists' opinions on the use of AI in endoscopy. A qualitative synthesis of included studies was conducted by categorizing into 5 domains: knowledge, perception, and attitude toward AI; impacts of AI on endoscopic performance; impact of AI on endoscopists; impact of AI on patients; and barriers toward the implementation of AI.

Ten studies were included, comprising 1587 endoscopists across Europe (32.6%), North America (42.8%), and Asia (24.6%). For domain 1, most endoscopists (69%-100%) had a basic knowledge of AI and most (79.5%-87.5%) expressed interest and optimism. For domain 2, most endoscopists (62.5%-97%) believed AI would positively impact endoscopic performance and quality. Domain 3 saw mixed perceptions, with 6.2% to 62.8% of endoscopists suggesting AI would lead to operator dependence, 21% to 81.3% believing AI would prolong procedural time, and most (71%-100%) disagreeing that AI would replace them. For domain 4, most endoscopists (81.3%) believed AI would improve patient care. For domain 5, most (75.2%-91%) identified costs as barriers to AI implementation. Opinions on ethics and regulation varied (12.5%-100% and 35%-88%, respectively), with most advocating for clear guidelines and regulations.

There is a positive attitude toward the use of AI in endoscopy. Concerns regarding the impact on clinical practice, costs, and medicolegal considerations remain. Establishing robust regulatory frameworks is crucial to the integration of AI into endoscopy.