[论文解读] Towards Agentic Intelligence for Materials Science

Huan Zhang, Yizhan Li+19 more|arXiv (Cornell University)|Jan 29, 2026

Machine Learning in Materials Science被引用 0

一句话总结

本综述倡导以管线为中心的端到端视角来看待AI4MS，提出具备代理能力的LLMs在完整材料发现循环中进行规划、行动与学习，以加速发现并确保安全。

ABSTRACT

The convergence of artificial intelligence and materials science presents a transformative opportunity, but achieving true acceleration in discovery requires moving beyond task-isolated, fine-tuned models toward agentic systems that plan, act, and learn across the full discovery loop. This survey advances a unique pipeline-centric view that spans from corpus curation and pretraining, through domain adaptation and instruction tuning, to goal-conditioned agents interfacing with simulation and experimental platforms. Unlike prior reviews, we treat the entire process as an end-to-end system to be optimized for tangible discovery outcomes rather than proxy benchmarks. This perspective allows us to trace how upstream design choices-such as data curation and training objectives-can be aligned with downstream experimental success through effective credit assignment. To bridge communities and establish a shared frame of reference, we first present an integrated lens that aligns terminology, evaluation, and workflow stages across AI and materials science. We then analyze the field through two focused lenses: From the AI perspective, the survey details LLM strengths in pattern recognition, predictive analytics, and natural language processing for literature mining, materials characterization, and property prediction; from the materials science perspective, it highlights applications in materials design, process optimization, and the acceleration of computational workflows via integration with external tools (e.g., DFT, robotic labs). Finally, we contrast passive, reactive approaches with agentic design, cataloging current contributions while motivating systems that pursue long-horizon goals with autonomy, memory, and tool use. This survey charts a practical roadmap towards autonomous, safety-aware LLM agents aimed at discovering novel and useful materials.

研究动机与目标

• 将材料科学从任务特定模型转向端到端、目标驱动的发现管线的动因与意义。
• 通过信用分配使上游的预训练、领域适应和工具使用与下游的发现结果对齐。
• 提出一种以管线为中心的框架，将预训练与适应视为可调整的组件，由实际世界的发现奖励所塑造。
• 在AI与材料科学社区之间建立共享的术语与评估框架。
• 勾勒在材料发现领域实现自主、具备安全意识的LLM代理的实际路线图。

提出的方法

• 提出一个管线为中心的视角，将语料库整理、预训练、领域适应、指令微调与代理与仿真与实验的交互联系起来。
• 分析AI进展（预测、生成、推理和多模态能力），并将其映射到材料科学的需求。
• 从AI角度描述材料科学中的被动任务（预测、挖掘、生成、优化），以识别自治发现的差距。
• 讨论当代具代理的系统及其从被动模型转向使用工具和长期奖励的目标导向代理的转变。
• 提出端到端信用分配与将发现结果反向传播到上游组件的机制（如数据相关性的影响函数）。
• 概述将自治实验室、工具使用代理与安全框架整合为一个连贯发现管线的路线图。

Figure 1: An overview of the key sections and an illustrative end-to-end pipeline encompasses key elements like general pre-training tasks & data, foundation language models, domain-specific tasks & data, materials-oriented model adaptation, goal-driven Large Language Model (LLM) agents, and iterati

实验结果

研究问题

• RQ1 Q1 自动性演变：在材料科学背景下，通用机器学习模型如何从被动处理器演变为具代理能力的系统？
• RQ2 Q2 尽管在任务基准上表现出色，现有的材料科学AI系统为何不足以实现自治发现？
• RQ3 Q3 要 bridging end-to-end 的差距，使AI能力与自治、以发现为驱动的工作流在材料科学中对齐，需要哪些要素？
• RQ4 Q4 如何通过以管线为中心的评估与信用分配，使预训练和适应与实验发现结果相一致？

主要发现

• LLMs 在模式识别、文献挖掘和性质预测方面具有优势，若通过具代理设计可扩展到端到端的发现。
• 以管线为中心的视角揭示了上游设计选择（数据整理、训练目标）如何通过信用分配影响下游的实验成功。
• 具代理系统能够使用工具、具备记忆与长期决策能力，超越静态的、特定任务模型。
• 与外部工具的集成（如密度泛函理论DFT、机器人实验室）可以加速计算工作流与材料设计中的实验规划。
• 以发现奖励引导的反馈驱动的预训练循环，可以自适应语料与目标以提升现实世界的结果。
• 本综述提出了走向自治、具安全意识的LLM代理用于新材料发现的实际路线图。

Figure 2: Technology tree of AI4Material science research. With the emergence of LLMs and agents, research on materials science initially focused on domain specific task, primarily concentrating on seperate reactive tasks. Subsequent research has delved deeper, gradually integrating more with agenti