材料科学作为一门关键学科,在推动社会进步和科技创新中发挥着不可替代的作用.随着人工智能(artificial intelligence,AI)和大数据技术的飞速发展,材料科学的研究范式正在经历一场深刻的变革.本综述探讨了 AI与大数据结合,如何重塑材料科学的研究范式(AI for materials science),加速计算材料学的发展,并推动实验过程的革新.首先,概述了大数据背景下材料数据库的基础设施建设.这些数据库作为科研工作的基石,为材料数据的存储、管理和分析提供了强大的支持.接着,讨论了 AI技术在材料发现周期各阶段的应用,并介绍了 AI与实验室自动化和机器人技术融合形成的自我驱动实验室(self-driving laboratories,SDLs)的兴起.SDLs实现了材料发现的完整闭环流程,推动了科学研究向自主科学发现模式的重要转变.此外,大型语言模型(large language models,LLMs)的发展更是为自然语言理解带来了革命性变化,推动科学LLMs的兴起,拓宽了从文本理解到科学探索的能力.本文进一步综述了材料科学中大型语言模型的最新进展,强调了它们在加速材料发现过程中的关键作用.最后,探讨了 AI在材料研究中的巨大潜力,并审视了构建材料研究智能生态系统所面临的挑战.通过对相关研究进展的梳理,本文旨在为科研人员提供信息参考,并揭示AI在材料研究中的重要性.
Integration of artificial intelligence and big data in materials science:New paradigms and scientific discoveries
Materials science plays an irreplaceable role in driving societal progress and technological innovation.With the rapid development of artificial intelligence(AI)and big data technologies,the research paradigm in materials science is undergoing profound transformations.This review discusses how the integration of AI and big data is reshaping the research paradigm in materials science(AI for materials science),accelerating the advancement of computational materials science,and innovating the experimental process.It begins by outlining the infrastructure development of material databases in the context of big data,which serve as the cornerstone of scientific work by providing robust support for the storage,management,and analysis of material data.These databases facilitate efficient data handling,enabling researchers to extract valuable insights from vast amounts of experimental and simulation data.Subsequently,the review explores the application of AI technologies across different stages of the material discovery cycle,including theoretical calculations,experimental design,data collection,and synthesis.AI algorithms,particularly deep learning,have revolutionized these stages by enhancing the ability to process and analyze complex datasets,revealing intricate relationships between material structures and their properties.A significant highlight of this review is the introduction of self-driving laboratories(SDLs).Resulting from the integration of AI with laboratory automation and robotics technology,SDLs have realized a complete closed-loop process for material discovery,promoting a significant shift towards autonomous scientific discovery models.These laboratories can independently design and execute experiments,analyze results,and iteratively refine hypotheses,greatly increasing the efficiency and accuracy of material discovery.Furthermore,the development of large language models(LLMs)has brought about revolutionary changes in natural language understanding,leading to the emergence of scientific LLMs,thus expanding the capabilities from text understanding to scientific exploration.The review provides an overview of the latest advancements in LLMs within materials science,emphasizing their critical role in expediting the material discovery process.These models can parse and understand vast amounts of scientific literature,enabling researchers to stay abreast of the latest developments and identify novel research directions.The review concludes by evaluating the challenges involved in building an intelligent ecosystem for material research.These challenges include the need for high-quality,standardized data,the integration of diverse AI tools,and the development of robust methodologies for cross-disciplinary collaboration.Despite these challenges,the substantial potential of AI in materials science is evident.AI technologies promise to transform material research,enabling the discovery of new materials with unprecedented speed and precision.In summary,this review aims to inform researchers about the significance of AI in materials science,highlighting the transformative impact of AI and big data on the research paradigm.It underscores the importance of developing intelligent systems and methodologies to harness the full potential of AI,thereby advancing the field of materials science and contributing to technological innovation and societal progress.
artificial intelligencematerial discoveryself-driving laboratorylarge language model
万方数据
维普
