随着 AI4Science (利用人工智能驱动科学发现)作为科学研究的新范式兴起, 人工智能正深刻推动植物生命系统研究向数据驱动与智能化转型。根系作为植物获取水分和养分、感知环境胁迫并调节地上-地下相互作用的关键器官, 对作物产量和生态系统功能起着决定性作用。然而, 由于根系生长在土壤的不可见性, 表型复杂且难以直接观测, 对其结构与功能认识长期滞后于地上部分。人工智能的发展为破解这一“地下黑箱”提供了新工具和新路径。如何以人工智能为核心驱动力, 整合多源根系与环境数据、前沿算法及根系科学知识体系, 进而揭示根系结构与功能的关键关系, 已逐渐成为当前根系研究的前沿方向。该文系统综述了 AI4Root 即 AI 在根系研究中的主要应用与发展趋势, 涵盖数据获取、结构建模、机制推理及管理决策等环节。基于 AI 的图像分析可实现根系结构的自动识别与量化; 建模与数据融合揭示根系-土壤多尺度、多过程的相互作用。AI 在根系表型测定、构型模拟、分泌物及微生物互作解析等方面显示出巨大潜力, 为深入理解根系功能提供前所未有的手段。然而, AI 应用于根系研究仍面临田间高质量数据获取困难、不同实验尺度数据融合不足及根系-土壤-环境交互复杂性建模不足等挑战。未来, 依托多源数据深度整合、跨学科算法优化及数字农业平台协同应用, AI 有望推动根系科学进入智能化研究新阶段, 为作物改良与智慧农业发展提供创新路径。

With AI4Science (AI-driven scientific discovery) emerging as a new paradigm for scientific discovery, artificial intelligence (AI) is reshaping plant science toward data-driven and intelligent frameworks. Roots, as the key organs responsible for water and nutrient acquisition, environmental sensing, and above-belowground interactions, are decisive for crop productivity and ecosystem functioning. However, due to their hidden growth in soil, structural complexity, and limited observability, our understanding of root structure and function has long lagged behind that of aboveground parts. Recent advances in artificial intelligence have provided new tools and pathways for decoding the “underground black box.” Leveraging AI as a core driving force to integrate multi-source root and environmental data, cutting-edge algorithms, and the broader knowledge base of root science has increasingly become a frontier direction in belowground research. In this context, we outline AI4Root, which summarizes the major applications and emerging trends of AI in root studies, encompassing data acquisition, structural modeling, mechanistic inference, and management decision support. AI-based image analysis enables automated recognition and quantification of root architectures, while modeling and data-fusion approaches reveal the multi-scale and multi-process interactions between roots and soil. AI also shows great potential in root phenotyping, architectural simulation, and deciphering root exudate-microbiome interactions, offering unprecedented tools to advance our understanding of root function. However, the application of AI in root research still faces several challenges, including the difficulty of acquiring high-quality field data, insufficient integration across experimental scales, and limited capacity to model the complexity of root-soil-environment interactions. Looking ahead, the deep integration of multi-source data, cross-disciplinary algorithmic advancements, and the coordinated development of digital agriculture platforms are expected to propel root science into a new era of intelligent research, offering innovative pathways for crop improvement and the advancement of smart agriculture.