Fanxuan Chen, Haoman Chen, Ting Yu, Ruoyun Wang, Yi Wang, Xian Zhang, Jiachen Li, Kaishuo Liu, Darong Hai, Xueying Bao, Zefei Mo, Dongren Yang, Zhao Wang, Yuxin Lin, Qinghua Xia, Gen Yang, Jianwei Shuai
Artificial intelligence (AI) is catalyzing a paradigm shift in medical robotics, transforming medical robots from teleoperated tools into intelligent partners across clinical domains. This evolution is pivotal in addressing global challenges like aging populations, driven by core AI pillars-including computer vision (CV), deep reinforcement learning, and large language models (LLMs)-that support perception, decision-making, and naturalistic communication, enabling varying degrees of autonomy and adaptive care. However, the literature still lacks a holistic analysis that integrates these advances and tackles the translational challenges hindering clinical adoption. This review bridges this gap by systematically charting the evolution of AI-driven robotics across intelligent surgery, adaptive rehabilitation, and multimodal healthcare delivery. We dissect the core technologies powering this revolution, from digital twins for surgical simulation to LLMs for enhanced human-robot interaction, and critically analyze the associated technical, ethical, and regulatory hurdles. By synthesizing current progress and outlining future frontiers, including embodied AI, nanorobotics, and the concept of the AI-augmented surgeon, this review provides a comprehensive roadmap for accelerating the translation of intelligent medical robotics into routine clinical practice.