Landscape of artificial intelligence in laparoscopic cholecystectomy: a narrative review

Laparoscopic cholecystectomy (LC) is the gold standard for symptomatic gallstone disease, yet bile duct injury from visual misperception remains a feared complication. Artificial intelligence (AI) and computer vision offer promising solutions to mitigate these cognitive errors by augmenting intraoperative visualization and enforcing the critical view of safety (CVS). We conducted a narrative review of the literature in the PubMed database up to August 2026. This review categorizes current AI applications into five domains: anatomical landmark detection, automated CVS assessment, surgical phase recognition, the impact of anatomical complexity, and real-world implementation. Current deep learning models demonstrate exceptional specificity (>95%) in identifying landmarks and "Go/No-Go" dissection zones. Automated platforms successfully provide real-time CVS verification and safety documentation. While phase recognition accuracy often degrades during severe inflammation, recent generative AI models are bridging this gap by accurately identifying scar tissue even during intraoperative bleeding. Clinically, AI-based coaching platforms significantly improve CVS achievement rates among surgical trainees and correct anatomical misjudgments made by experienced surgeons. Additionally, large language models are emerging as educational and decision-support tools, though rigorous procedure-specific validation remains necessary. Furthermore, real-world deployment via low-latency cloud architecture has proven technically feasible. AI has evolved into a clinically validated adjunct in LC that fundamentally enhances procedural safety. Future research must address data annotation bottlenecks, model robustness in severe cholecystitis, and integration with robotic platforms to fully realize a safe, data-driven surgical ecosystem.