来源：北京科协 你是否曾仰望星空，思考宇宙的边界？ 你是否好奇，构成这个世界的最小单元，到底是什么？ 当人工智能闯入科学的世界，一切认知，正在被重新书写。 带上你的好奇心，让我们一起走进这场正在发生的科技革命。 科学，不再只是试管、望远镜与公式推演； AI，正成为我们理解宇宙、探索未知的"新器官"。 从宏观层面—— 太阳耀斑能预测吗？怎么提高准确率？ 到微观世界—— 如何在海量数据中找到极少数信号粒子出现的事例？ 从"人工操控"—— 人类做不到的精雕细琢，人工智能模型能完成吗？ 到"智能自驱"—— AI设计的火箭发动机真的能用吗？ 2025年7月，北京发布全国首个专注于"AI for Science"的专项地方政策，推动人工智能与科研的深度融 合。这不是技术的简单叠加，而是一场科研范式的根本变革。 （来源：北京科普发展与研究中心） 北京正成为科学智能的"策源地" 让我们一起见证 科学智能的黄金时代 "科技创新调研行"——六集"人工智能+"主题纪录片《AI向新力》，由北京市科学技术协会出品，北京 科普发展与研究中心承制，通过纪实镜头全景展现首都以人工智能为核心引擎，培育和发展新质生产力 的生动实践。 敬请关注" ...

在实验桌前，一名"智能科学家"正抬起结实的机械臂抓取试管，依次走过液体进样台、磁力搅拌台、烘干 工作台，进行样品称重、搅拌、离心、烘干…… 精准智能化学全国重点实验室的19个分布式实验室中，分布着110台这样的"智能科学家"。它们可以通过自 主实验，精准完成试剂配置、样品合成、性能表征等一系列操作，将实验数据实时同步至智能操作系统。 除了能24小时"做"实验，这些"智能科学家"还可以自己"看"论文、"想"方案甚至"学"经验。 化学实验如何告别传统"试错法"，破解新物质制作周期长、成本高的难题？ 在中国科学技术大学精准智能化学全国重点实验室里，大模型成为科研"超级助手"，AI机器人成为实验 室"主力军"。可以自主设计实验方案、24小时不间断做实验的智能科研基础设施——"智能科学家"，正成为 人工智能引领科研范式变革的代表之一。 当看到"人工智能+"赋能科研创新的新路径后，科研团队为"小来"制作了机械臂，又接入了多个生成式大模 型。优化了"大脑"和"手臂"的第二代"机器化学家"，被命名为"小临"。 "十五五"规划建议专章部署"加快高水平科技自立自强，引领发展新质生产力"，提出要提升国家创新体系整 体效能。在中国科 ...

精准智能化学全国重点实验室的19个分布式实验室中，分布着110台这样的"智能科学家"。它们可以通 过自主实验，精准完成试剂配置、样品合成、性能表征等一系列操作，将实验数据实时同步至智能操作 系统。 在实验桌前，一名"智能科学家"正抬起结实的机械臂抓取试管，依次走过液体进样台、磁力搅拌台、烘 干工作台，进行样品称重、搅拌、离心、烘干…… 新华社合肥12月10日电 化学实验如何告别传统"试错法"，破解新物质制作周期长、成本高的难题？ 在中国科学技术大学精准智能化学全国重点实验室里，大模型成为科研"超级助手"，AI机器人成为实验 室"主力军"。可以自主设计实验方案、24小时不间断做实验的智能科研基础设施——"智能科学家"，正 成为人工智能引领科研范式变革的代表之一。 团队花了三年时间，像"蚂蚁搬家"一样从教科书、论文和专利中整理出百万条化学数据，同时，还把本 校相关领域专家的研究经验"灌"进计算机，让"智能大脑"去理解更为复杂的化学知识。 2024年，"智能科学家"通过机器阅读和机器实验，创制相变隔热阻燃材料，应用于新能源电池、消防服 中可耐高温，且背面温度处于热安全范围。该成果已在2025年通过量产型测试和产业验 ...

Core Insights - The report titled "2025 Research Frontiers" indicates that China ranks first in the research frontiers heat index in six fields, showcasing its strong research capabilities and active exploration in these areas [1][6]. Group 1: Research Fields - The "2025 Research Frontiers" report identifies 110 hot frontiers and 18 emerging frontiers across 11 highly integrated disciplines, including agricultural science, botany and zoology, ecological and environmental sciences, earth sciences, chemistry and materials science [3]. - The research frontiers heat index reveals the research activity levels of major countries and regions in these 11 disciplines, with China leading in agricultural science, botany and zoology, ecological and environmental sciences, chemistry and materials science, physics, information science, as well as economics, psychology, and other social sciences [6]. Group 2: Research Quality and Trends - The president of the Chinese Academy of Sciences Technology Strategy Consulting Institute, Pan Jiaofeng, stated that the quality of basic research in China has steadily improved, emphasizing the importance of artificial intelligence in transforming research paradigms and its application across various fields [8].

0:00 《中共中央关于制定国民经济和社会发展第十五个五年规划的建议》提出，以人工智能引领科研范式变 革。 "人工智能赋能科学研究"旨在运用人工智能学习科学原理、构建科学模型以解决实际问题。近年来，在 移动互联网、大数据、超级计算等新理论新技术的驱动下，人工智能加速发展，并且正在以前所未有的 深度和广度重塑科学研究的基本逻辑与方法体系。 11月2日拍摄的宿州市算力产业园淮海人工智能计算中心服务器。 11月6日，观众在2025年世界互联网大会"互联网之光"博览会现场体验与宇树科技G1机器人"格斗"。 《建议》提出，全面实施"人工智能+"行动，以人工智能引领科研范式变革，加强人工智能同产业发 展、文化建设、民生保障、社会治理相结合，抢占人工智能产业应用制高点，全方位赋能千行百业。 11月12日，在葡萄牙里斯本，中国企业宇树科技的人形机器人和与会者在2025年网络峰会现场互动。 "十五五"时期，通过进一步拥抱场景创新，不仅能让技术和产业、研发和市场有机结合，更能推动科技 创新和产业创新融合发展。 编导：杲均丰 包装：胡心懿 新华社音视频部制作 "十四五"期间，我国在人工智能领域取得了令人瞩目的成果：我国人工智能综合 ...

Core Viewpoint - The article discusses the emergence of AI for Science as a transformative direction in scientific research, highlighting the establishment of companies like Xaira Therapeutics and the initiatives by OpenAI and DeepMind in this field. It emphasizes the potential of AI to accelerate scientific discoveries and the journey of Chinese entrepreneurs Zhang Linfeng and Sun Weijie in founding DeepMind Technology, which focuses on applying AI to scientific research and industrial applications [3][4][5]. Company Background - DeepMind Technology was founded in 2018 by Zhang Linfeng and Sun Weijie, with initial funding of 12 million RMB from a disruptive technology innovation competition, rather than venture capital [4][5]. - Zhang Linfeng developed the Deep Potential Molecular Dynamics (DeePMD) method during his PhD at Princeton, which later won the prestigious Gordon Bell Award [4][5]. Technological Innovation - DeePMD integrates AI to optimize the long-standing issue of solving first-principles calculations, expanding the range of quantum mechanical calculations from hundreds of atoms to billions, thus enabling the discovery of new materials and drugs [5][6]. - The method allows for significant computational efficiency, achieving over six orders of magnitude acceleration, enabling complex simulations that were previously only feasible on supercomputers to be run on standard laptops [21][24]. Vision and Goals - The founders aim to create an open-source system that spans scientific research to industrial development, aspiring to contribute to a shared human destiny [9][30]. - The company has set a goal to become a leading technology firm originating from China, with a vision to influence global scientific research [8][30]. Product Development - DeepMind Technology has launched several platforms, including the Hermite drug design platform and various pre-trained scientific models, serving notable clients such as CATL, BYD, and others [8][30]. - The company’s first product, Hermite, was developed in response to the existing market needs in drug discovery, differentiating itself by incorporating machine learning methods [30][31]. Market Positioning - The founders identified a significant opportunity in the pharmaceutical and materials sectors, where understanding atomic interactions can lead to breakthroughs in drug development and material science [31][32]. - The company aims to build a comprehensive platform that can serve multiple research directions and stages, rather than focusing solely on vertical applications [50][51]. Educational Initiatives - DeepMind Technology emphasizes the importance of cultivating a new generation of interdisciplinary talent, integrating knowledge from physics, chemistry, and engineering to address complex scientific challenges [27][34]. - The company has developed a unique educational framework to train young talents, fostering a community that encourages collaborative learning and innovation [36][37]. Future Directions - The article suggests that the next phase for AI in science will involve the development of AI scientists, capable of autonomously conducting research and integrating various scientific tools [42][44]. - The integration of pre-trained models and multi-agent systems is expected to enhance research efficiency and redefine the roles of researchers in the scientific process [47][49].

Core Insights - The development of artificial intelligence (AI) is significantly supported by advanced computing power and innovative technologies, as highlighted in the recent policy recommendations from the Chinese government [1] - China's computing power centers have reached a total scale of 10.85 million standard racks, with intelligent computing power at 788 billion billion operations per second, positioning the country among the global leaders in AI infrastructure [1] Group 1: AI in Research and Innovation - AI is driving a transformation in research paradigms, enabling faster and more accurate scientific discoveries, particularly in fields like computational biology [2][3] - The collaboration between universities and computing companies is accelerating the application of intelligent computing in research, enhancing the efficiency of model training and inference [2][3] Group 2: Diverse Applications of Intelligent Computing - Companies like Yili are leveraging AI and cloud computing to create smart health management systems for livestock, improving operational efficiency and product quality [4][5] - The integration of AI in manufacturing processes, such as in the production of high-speed trains, has significantly reduced simulation times and improved design accuracy [5] Group 3: Industry Collaboration and Innovation - The establishment of the "Supernode Computing Cluster Innovation Alliance" aims to enhance collaboration among companies in chip development, system design, and AI applications [8] - Innovations in computing architecture, such as the development of ultra-node servers, are addressing the challenges of high energy consumption and system scalability in AI applications [7][8]

Core Insights - The development of intelligent computing power, based on the latest AI theories and advanced computing architectures, is significantly supporting the advancement of artificial intelligence [1] - China's computing center has reached a total scale of 10.85 million standard racks, with intelligent computing power at 788 billion billion times per second, and storage capacity exceeding 1,680 exabytes, positioning the country among the global leaders in AI model development [1] Group 1: AI in Research and Innovation - Intelligent computing is driving a paradigm shift in scientific research, accelerating the generation of original innovative results [2] - Researchers at Westlake University are utilizing intelligent computing technology to efficiently analyze sequencing data related to non-coding RNA, significantly speeding up research processes [2] - The collaboration between universities and computing companies is enhancing the application of intelligent computing in scientific research [3] Group 2: Diverse Applications and Industry Integration - The State Council's policy emphasizes the need for a coordinated approach to intelligent computing supply, making it accessible, economical, and environmentally friendly [4] - Companies like Yili are leveraging AI computing to create health profiles for dairy cows, improving monitoring and management efficiency in production [4] - The development of over 800 intelligent agents at Yili has optimized supply chain scenarios, reducing risks related to raw material shortages and excess [5] Group 3: Technological Advancements and Collaborations - Baidu's AI foundation, built on the Wenxin model and PaddlePaddle deep learning platform, is enhancing efficiency across various industries [6] - Inspur Information has introduced AI computing systems that significantly reduce the cost and time of model training and inference [7] - The establishment of the "Super Node Computing Cluster Innovation Alliance" aims to advance standards and applications in intelligent computing [8]

Core Points - The event "Implementation of the Future Pact University Dialogue" was successfully held in Beijing, marking the 80th anniversary of the United Nations, focusing on the role of universities in promoting the Future Pact and accelerating the 2030 Sustainable Development Agenda [1][2] - The dialogue attracted representatives from universities worldwide, discussing how educational cooperation, research innovation, and policy advocacy can drive sustainable development goals [1][2] - A significant outcome of the dialogue was the release of the "University Implementation of the Future Pact Action Plan," which aims to establish mechanisms for systematic implementation of the Future Pact in talent cultivation, research collaboration, and policy advocacy [3] Group 1 - The dialogue emphasized the importance of multilateralism in a chaotic and uncertain era, with universities playing a crucial role in translating words into action [2] - The Chinese Ministry of Education expressed support for deepening cooperation between universities and international partners, highlighting the proactive role of Beijing Foreign Studies University in the Future Pact [1][2] - The dialogue included discussions on artificial intelligence governance, interdisciplinary collaboration, and changes in research paradigms, addressing challenges faced by future universities [2] Group 2 - The "Future University Alliance" and "Future Learning Excellence Center" are proposed as part of the action plan to enhance collaboration among universities [3] - The event also initiated the preparation process for the "Sustainable Development Goals Series Dialogue," showcasing the youth's creativity and concern for sustainable development through performances [3] - The dialogue marked a significant step for global higher education in multilateral cooperation and sustainable development, aiming to create a more inclusive, equitable, and resilient future [3]

Core Insights - The article discusses the transformative impact of artificial intelligence (AI) on scientific research and innovation, highlighting the launch of the "Three-Body Computing Constellation" by Zhijiang Laboratory as a significant step in this paradigm shift [5][9]. Group 1: Impact of AI on Scientific Research - AI is revolutionizing scientific research methodologies, making traditional approaches increasingly unsustainable due to time and cost constraints [6][8]. - The integration of AI into research is not merely a tool revolution but a scientific revolution that transcends disciplinary boundaries, enabling the discovery of complex information previously unseen by humans [8][11]. Group 2: The "Three-Body Computing Constellation" - The "Three-Body Computing Constellation" consists of a thousand-satellite scale space computing infrastructure aimed at real-time data processing to overcome traditional satellite data processing inefficiencies [9][10]. - This constellation will facilitate inter-satellite communication, significantly enhancing operational efficiency and reducing costs, while also improving response times for emergency situations [10]. Group 3: Future of Scientists in the AI Era - AI is viewed as a tool to extend human creativity rather than replace scientists, emphasizing the importance of creativity in research [11][12]. - The ability to utilize AI in research will become a critical skill, with those who adapt likely to outpace those who do not [12][13].