Background: Artificial intelligence (AI) and robotic technologies are reshaping modern surgery, evolving from mechanical aids into cognitive collaborators. Their impact on outcomes, workflow and surgical autonomy remains debated. This review synthesises the highest-level evidence from the past decade examining AI-assisted surgery in humans.

Methods: A comprehensive PubMed and Embase search (January 2015–October 2025) identified peer-reviewed human studies using combinations of “artificial intelligence”, “machine learning”, “deep learning”, “computer vision” and “robotic surgery”. Eligible publications included systematic reviews, meta-analyses and large observational cohorts. Thirty-eight studies were narratively appraised and grouped into seven domains: foundational frameworks, comparative outcomes, predictive analytics, intra-operative computer vision, specialty exemplars, training and ergonomics and ethics and safety.

Results: Across high-quality meta-analyses and multicentre cohorts, AI-assisted or robotic surgery consistently reduced blood loss (20-35%), shortened hospital stay (by 1-2 days), and lowered conversion rates (up to 40%) without compromising oncological margins or increasing complications. Machine-learning models outperformed conventional risk scores for morbidity, mortality and postoperative delirium prediction. Computer-vision and augmented-reality systems enhanced anatomical recognition, navigation and instrument precision. Specialty-specific studies demonstrated reproducible advantages in hepatobiliary, colorectal, urological, thoracic and paediatric surgery. Training platforms incorporating AI shortened learning curves and reduced ergonomic strain, while ethical analyses emphasised the necessity of transparency, data security and sustained human oversight.

Conclusions: Evidence from 38 studies shows that AI augments rather than replaces surgical expertise. When ethically implemented, it enhances precision, safety and efficiency across disciplines. The future of operative care rests on a partnership between human judgement and algorithmic insight.