Towards Agentic Intelligence for Materials Science
Huan Zhang, Yizhan Li, Wenhao Huang, Ziyu Hou, Yu Song, Xuye Liu, Farshid Effaty, Jinya Jiang, Sifan Wu, Qianggang Ding, Izumi Takahara, Leonard R. MacGillivray, Teruyasu Mizoguchi, Tianshu Yu, Lizi Liao, Yuyu Luo, Yu Rong, Jia Li, Ying Diao, Heng Ji, Bang Liu
2026-02-10
Summary
This paper explores how artificial intelligence, specifically large language models, can be used to dramatically speed up the process of discovering new materials. It argues that current AI approaches are too focused on small tasks and need to be developed into systems that can independently plan and execute the entire materials discovery process, from initial idea to actual experimentation.
What's the problem?
Discovering new materials is traditionally a slow and expensive process. While AI has shown promise in helping, most AI tools are designed for specific steps, like predicting a material's properties. They don't handle the whole process – figuring out what to research, designing experiments, analyzing results, and then using those results to plan the next steps. This means the full potential of AI isn't being realized, and progress is limited by how well each individual AI tool performs.
What's the solution?
The paper proposes a complete 'pipeline' approach, thinking of the entire discovery process as one system. This pipeline includes collecting and preparing data, training AI models, adapting those models to materials science, and then using them to control simulations and even physical experiments. It emphasizes connecting decisions made early in the process, like how data is collected, to the success of experiments later on. The paper also highlights the shift from AI that just reacts to information to AI 'agents' that can proactively pursue long-term goals, remember past actions, and use different tools to achieve them.
Why it matters?
This research is important because it outlines a path towards creating AI systems that can autonomously discover new and useful materials. This could revolutionize fields like energy, medicine, and manufacturing by allowing us to find materials with specific properties much faster and more efficiently than ever before, potentially leading to breakthroughs we haven't even imagined yet.
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.