A correct understanding and accurate assessment of the health risks associated with noise are essential for formulating effective noise policies and for learning to live within noisy environments. This review systematically summarizes the pathways through which noise affects human health, identifies noise−related health impacts among specific population groups and within particular spatial settings, and explores the potential of soundscape approaches to enhance occupant well−being. The findings show that noise can cause physiological effects such as hearing loss, tinnitus, cardiovascular diseases, and sleep disturbances, and can also lead to psychological and cognitive consequences including annoyance and cognitive impairments. Certain population groups exhibit heightened vulnerability to noise exposure. Children, older adults, and individuals with high noise sensitivity are key susceptible groups. Compared with other spatial contexts, urban public spaces and residential environments require particular attention due to their pronounced noise−related health risks. In addition, positive soundscapes can facilitate psychophysiological restoration, alleviate cognitive fatigue, and improve the well−being of socially disadvantaged groups. Soundscape interventions are increasingly becoming an important approach, beyond traditional noise control, for improving the acoustic quality of urban environments.

China's science and technology development has entered a critical stage of tackling key challenges and therefore needs to strengthen original innovation. Basic research is the source of original scientific and technological innovation, yet at present China's basic research faces problems such as oversized research teams, homogenized research directions, a lack of core capabilities, and consequently insufficient capacity for original innovation. At the national level, this manifests as team research that is neither specialized nor distinctive and often duplicative, along with an overall pattern that runs counter to diversity. The resulting "resource−driven" research paradigm leads to inefficient use of research resources. As a result, the research ecosystem is degraded, hindering the cultivation and growth of original innovation. This paper argues that, in the realm of basic research, a "small but excellent" paradigm—characterized by small teams that each play to their unique strengths and researchers who are highly specialized and each bring distinctive expertise—has high value for original innovation. Under resource constraints, it can also preserve the overall diversity of national research and the security of the national science and technology chain. Delivering the core value of basic research through rigorous, specialized exploration and original breakthroughs requires each team to focus on a distinctive direction and cultivate it deeply, while team members build deep expertise and tackle hard problems through years of sustained effort. This approach helps foster diversity in the research ecosystem and enables efficient resource allocation, thereby ensuring the resilience of China's scientific and technological development in international competition.

From a historical perspective, this paper reinterprets the magnificent epic of human civilization evolution driven by continuous breakthroughs in cognitive revolutions, and identifies three cognitive revolutions experienced by humanity: The first cognitive revolution, centered on the invention of writing and education, enabled humans to break through the limitations of biological instincts, establish an ecological system for cultural inheritance, and realize the intergenerational accumulation and dissemination of knowledge. The second cognitive revolution originated from the explosive development of science and technology in the past 500 years, promoting the birth of industrial civilization and reshaping humanity's ability to transform nature. The current rapid advancement of artificial intelligence marks the arrival of the third cognitive revolution, whose core feature is the symbiosis, complementarity, and co−creation of human intelligence and machine intelligence. This paper proposes the philosophical distinction between "thinking soft constructs" and "material hard constructs," and establishes a common cornerstone governing both human and machine cognition through the four−element theory consisting of matter, energy, structure, and time. It points out that science and technology, as an accelerator of human social development, has led to the subversive nature of machine brute−force thinking. By analyzing the symbiosis and co−creation of human intelligence and artificial intelligence, this paper re−examines the relationships between individuals, between humans and society, and between humans and nature, providing a fundamental coordinate for the intelligent era. With forward−looking humanistic care, it explores the fate of humanity in the era of intelligent creation, aiming to embrace an unprecedented new civilizational paradigm.

Developing efficient material recovery strategies is crucial for achieving a sustainable transformation of the plastic economy. However, the complex composition and high heterogeneity of real−life plastic waste pose significant challenges to the process. Professor Ma from Peking University, along with collaborators, proposed an orthogonal transformation strategy. They integrated solid−state NMR analysis with catalytic conversion methods. This approach successfully established a highly adaptable route. The route can convert unknown real−life plastic waste into a series of valuable chemicals. The strategy still requires further optimization and validation in its economic feasibility, environmental benefits, industrial applicability and so on. Nevertheless, it offers a complementary new perspective for addressing this complex real−world problem. This strategy holds promise for achieving high−value utilization of carbon and hydrogen resources in real−life plastic waste. It also provides theoretical support for advancing a circular economy in the plastics sector.

With the confirmation of data elements as the latest production factor, accelerating data sharing and value mining on the basis of ensuring data security has become an industry consensus. Privacy computing is the core technology for achieving "data available but invisible". Privacy computing platforms have flourished and gradually forming a phenomenon of "platform islands" with privacy computing platforms as the core. How to solve platform silos and further promote the integration and value of data elements has become an important research direction. This paper summarizes the typical functions of privacy computing platforms at home and abroad, the current status of the interconnection ecology of privacy computing platforms in China, and industry practice cases. The research scope of the paper is clarified as node interconnection and algorithm interconnection. It analyzes the different modes of node interconnection as well as algorithm, and finally summarizes the algorithm, data, security, and performance difficulties of privacy computing cross platform interconnection, and elaborats on the trend and prospects. Although privacy computing platforms face various challenges in interconnection, cross−platform interconnection remains a significant trend for future development. In terms of standards and regulations, both the financial industry and the telecommunications industry are exploring their own internal standards for interconnecting privacy computing platforms. From a technical perspective, the dual adaptation functions under the two−way adaptation model will gradually be replaced by "adapter"−style conversion products. Meanwhile, large−scale, open protocol−based platform interconnection is increasingly becoming the preferred choice for more technical providers.

Artificial intelligence (AI) is driving a paradigm shift in scientific research—from functioning primarily as an "accelerator" to emerging as a genuine "discoverer." Using the "Machine Chemist" platform at the University of Science and Technology of China as an illustrative example, this work provides a systematic analysis of the potential and challenges of AI in chemical knowledge discovery. Through machine learning, knowledge graphs, and automated experimental systems, AI can achieve a true transition from data to knowledge in areas such as molecular design, spectroscopic analysis, catalyst screening, and materials development. However, for AI to become an autonomous discoverer of chemical knowledge, three critical bottlenecks must be addressed: The scarcity of high−quality data, the limitations of human cognitive frameworks, and the low efficiency of experimental validation. This study further examines how chemical foundation models, multimodal data integration, and industrial−scale intelligent laboratories can drive systematic transformation of future scientific research paradigms by enabling data−driven decision optimization, accelerating interdisciplinary research, and restructuring automated experimental workflows.

This paper focuses on the developmental progression of robots evolving from "automation−autonomy" towards "selfhood" in technological evolution. Robots have undergone industrial automation and collaborative autonomy and are now advancing into the new generation of embodied intelligent robots characterized by a stage of "selfhood". Only with certain cognitive abilities and a sense of intelligent boundaries can robots serve humans more effectively. The connotations of automated, autonomous, and selfhood robot systems are discussed, key technologies of robot selfhood are analyzed, and the characteristics of the three different stages are compared. The journey toward robot "selfhood" still faces challenges including technical bottlenecks, pressures from ethical and social restructuring, and lagging legal frameworks. These challenges point to the evolutionary directions for the next decade. In terms of technological breakthroughs, neuro−symbolic fusion architectures are emerging; deepening application scenarios carry greater humanistic significance; the construction of embodied intelligence ecosystems is accelerating; governance mechanisms are evolving with promise; and the ultimate significance of robot selfhood lies in expanding the boundaries of intelligence. "Selfhood" is not only a technical breakthrough in robotics but also has profound implications for the future development relationship between robots and humans.

As a class of artificial materials possessing extraordinary properties that not found in natural matters, metamaterials offer a technological pathway for reconstruction of matters to get "artificial matters". These highly functional and designable artificial matters provide more ideal physical carriers for the implementation of various types of artificial intelligence and open up new possibilities for the realization of corporeal artificial lives. Based on exploring the relationship of matter, intelligence, and life systems, this commentary reviews the applications of metamaterials in various artificial intelligence systems, especially the research progress in high-performance optical computing and embodied intelligence, and analyzes the feasibility of constructing artificial lives with metamaterial strategies.

China has achieved remarkable accomplishments in infrastructure construction, with its scale and speed ranking among the world's foremost. However, behind this rapid development, engineering design software, the "nerve center" of digital engineering, has long been dominated by foreign products, creating severe "bottleneck" risks due to external dependencies. The inclusion of the "Issue of Indigenous Engineering Design Software in the Infrastructure Sector" among the Top Ten Engineering and Technological Challenges for 2024 by the China Association for Science and Technology (CAST) underscores the extreme urgency of addressing this "Achilles' heel". This paper first reviews how foreign software established market dominance in China through first−mover advantages during the technological evolution from 2D drafting to 3D modeling. It then provides an in−depth analysis of the multidimensional risks arising therefrom, including threats to national information security, constraints on key core technologies, and stifled industrial innovation vitality. Building upon this, the paper systematically evaluates the profound "growth dilemmas" currently faced by domestic software in achieving autonomy in key core technologies, fostering a collaborative innovation ecosystem, and adapting to the strategic demands of "new quality productive forces". In response to this situation, this paper focuses on the core question of "How can indigenous engineering design software forge the 'Chinese cornerstone' of digital infrastructure?", rejecting superficial, symptomatic remedies and instead endeavoring to explore systemic breakthrough strategies for developing indigenous and controllable engineering design software. It emphasizes that this is not merely a technological challenge but a critical strategic imperative concerning the consolidation of national digital sovereignty, the cultivation of new industrial growth drivers, and the shaping of future national core competitiveness. The aim is to stimulate profound reflection within the industry on how to construct an autonomous, secure, and efficient "Chinese cornerstone" for digital engineering, and to foster consensus that drives proactive practical actions.

电池储能是新能源安全运行的“超级稳定器”，测试仪器则是贯穿电池储能乃至新能源系统研发、生产、制造、应用等全生命周期的关键利器。以电池测试技术与仪器为典型案例，系统回顾中国电池测试仪器从产品进口到技术自主创新的演进之路。创建高压电池包“快速精准激励-智能建模估计”一体化测试方法体系，填补电池测试领域理论方法空白；推动高压大功率测试仪器国产化，核心技术指标国际领先（充放电转换时间 < 3 ms、能量变换效率95%），抢占了国际电池测试领域的科技制高点。剖析了测试技术促进我国新能源装备革新升级，新能源装备控制的实践反哺测试技术创新，二者相互促进，协同发展的关系，总结了中国科研人员不惧封锁，勇挑重担，坚守学术自信、道路自信；紧密围绕国家重大战略与市场需求，攻坚克难，真正做到产学研结合，快速推进创新技术走向产业一线，自主研制测试仪器的实践经验和体会。

分析了激光无线能量传输（laser wireless power transmission，LWPT）技术的基本原理、发展背景及未来应用。该技术基于受激辐射原理，将电能转化为激光能量，经过大气或真空介质传输后，最终由激光电池将光能高效转换为电能，具备高方向性、高能量密度、远距离传输等优势。在低空产业领域，LWPT可为无人机、飞行汽车等提供持续能源补给，显著提升续航能力与作业效率；在商业航天领域，其可为卫星、空间站及深空探测器构建灵活的能量传输网络，支撑长期太空任务。展望未来，LWPT技术与激光通信、卫星遥感定位等技术结合，可实现能源与信息的协同传输，构建交通–信息–能源三网融合的智能化蓝图，推动交通系统向高效化、自动化迈进。这一目标的实现，需依托多行业深度合作与持续技术创新，并同步推进跨领域协同研发及标准规范制定。通过推动该技术的发展，有望催生能源传输领域的革命性变革，为现代社会的发展提供坚实的支撑。

探讨了人工智能的学科体系，从人工智能的定义出发，提出人工智能的学科体系是由人工智能的科学基础、人工智能技术方法和人工智能应用3部分组成，并认为人工智能的科学基础是符号主义、连接主义和行为主义，人工智能的主要技术是基于知识的人工智能技术、基于数据的人工智能技术、人工智能的算法和人工智能的算力，而人工智能的技术要素为知识、数据、算法和算力，人工智能的应用领域涉及经济、科技、民生、社会和其他领域。

挖掘了新时代高质量发展背景下推动科研机构加速数字化转型的深层次动因。分析了科研机构的独特运营模式和业务特点，以此为基础，构建了推进科研机构数字化转型快速落地的工作思路，即“全在线”工作体系，以及为实现“全在线”工作体系落地的V模型理论。梳理了科研机构在数字化转型进程中可能遭遇的业务整合困难、价值效益不显性、指导和引领缺乏、科研进度制约的各类挑战，并针对此提出助力科研机构推进转型工作，全面提升创新能力的建议：明确转型规划，均衡能力建设；制定评价体系，开展价值宣传；建立合作机制，强化指导交流；制定协同机制，保障科研进展。

概述了DeepSeek在通用人工智能领域的最新进展，重点讨论了其在大语言模型、推理技术方面的创新。DeepSeek-V3引入了新的模型架构和算法设计，基于相对有限的智能硬件，对模型训练方法进行了全面和深入的优化，显著提升了模型训练效率。在推理技术方面，DeepSeek-R1创新性地结合了强化学习（RL）与监督微调（SFT），提升了推理深度和逻辑推理能力。结合DeepSeek的创新工作，讨论了通用人工智能发展趋势，重点涉及3个问题：开源开放生态对发展通用人工智能的作用；依赖于模型规模扩展的“Neural Scaling Law”是否还能发挥作用；如何基于DeepSeek这类基座模型，以“通专结合”的方式实现行业大模型的落地等。

科技发展和科技促进发展的重大决策更加复杂和不确定，对科技战略咨询的综合集成提出了更高要求。通过对系统方法论、智库双螺旋法、哲理-数理-物理-事理-人理、融合科学等理论研究的梳理，在把握科技战略咨询规律，掌握科学方法和实践路径的基础上，归纳了科技战略咨询综合集成的3种基本路径：智库科学与工程导论为科技战略咨询的综合集成奠定了基础、应用数字技术为科技战略咨询的综合集成提供了手段、技术识别预测系统为科技战略的综合集成提供了实践案例。更好综合集成科技战略各要素，提升科技智库的战略研究与战略咨询能力，以及政策建议的科学性和可行性，以高质量科技战略研究与科技战略咨询支撑高水平科技战略决策。

在量子信息技术与信息通信技术融合背景下，面向提升信息通信安全性和便携性的迫切需求，采用了理论分析和实际应用案例结合的方式，探讨了量子保密通信与5G融合的探索实践，展示了量子保密通信在5G网络中的融合策略与应用成效。量子保密通信正携手5G在承载网络、网络切片等基础设施层面探索融合，同时催生出量子加密通话、量子加密对讲等融合创新应用。展望了量子保密通信与5G融合的未来发展：随着量子保密通信和5G在基础设施、应用场景等方面不断深度融合，将共同构筑高速泛在、自主可控的数字基础设施，为数字经济的蓬勃发展提供坚实而强大的安全保障，有力推动产业数字化转型升级，助力数字经济持续发展。

阐述了因DeepSeek横空出世引发的关于AI发展路径的思考。首先解释了为什么DeepSeek会引起全球性的科技震撼，接着讨论了“规模法则”（Scaling Law）是否已遇到天花板、发展“通用人工智能”（AGI）应选择什么道路、发展人工智能应该追求高算力还是高算效（高能效）、“开源”为什么有这么大的威力等问题。最后对中国在人工智能领域的实力提升和如何实现人工智能自立自强提出建议。

人工智能的飞速发展给生物学研究带来了深远影响，其中，AlphaFold2在蛋白质结构预测领域引发了革命性的突破。评估了AlphaFold2对G蛋白偶联受体（GPCR），结构预测的可靠性，凭借其对蛋白质三维结构的高度准确预测。然而，在实际应用中，AlphaFold2的预测结果并非在所有场景下都足够精确。以GPCR为例，这类重要的药物靶点参与了广泛的生理过程，其结构研究对理解功能机制和药物开发具有重要意义。结果表明，尽管AlphaFold2能够准确捕捉GPCR整体骨架的主要特征，但其预测模型在胞外域与跨膜域的组装、配体结合口袋的形状，以及信号传导界面的构象等方面，与实验解析的高分辨率结构存在显著差异。这些差异限制了AlphaFold2模型在GPCR功能研究和基于结构的药物设计中的应用能力。因此，尽管AlphaFold2为结构预测提供了强大的工具，但其在特定场景下的局限性表明，AI结构预测作为一种辅助工具，尚不能完全取代实验结构生物学，需要联合使用以辅助药理学研究和药物设计。

地球系统科学作为一门新兴的交叉学科，近40年来在地学界、生态学界乃至环境学界引起了广泛关注。指出了地球系统科学与传统地球科学的关系，强调地球系统科学不能代替现有的地球科学各分支学科的研究，地学的各个分支学科将继续其自身的发展，但应该考虑圈层间的相互作用；分析了人类活动主导世界前的地球系统运行规律、现代人类在地球系统中的作用，探讨了全球变暖的地球系统科学命题。