2025年12月，数字科学促进与研究协会（ADSPR）正式发布了《数字科学发展年度白皮书（2025）》（以下简称"白皮书"），这是协会成立七年以来首次系统性地对全球数字科学研究态势进行全景式梳理。白皮书由数智研究院学术团队历时六个月编撰，涵盖人工智能、数据科学、网络安全、区块链与分布式系统、云计算与边缘计算、数字伦理与治理六大核心方向。
白皮书指出，2025年全球数字科学研究呈现出几个显著趋势。在人工智能方向，大语言模型从"规模竞赛"转向"效率优化"和"安全对齐"，模型可解释性和AI安全成为学术研究的核心议题。多模态融合技术快速成熟，视觉-语言模型在医疗诊断、工业检测和自动驾驶等领域实现了从实验室到产业应用的关键跨越。
数据科学方向，合成数据技术的突破为小样本学习和隐私保护场景提供了新的解决方案。同时，数据治理从技术问题上升为战略议题，全球多个地区出台了更为严格的数据合规法规，推动了隐私计算和联邦学习技术的加速落地。
网络安全方向的变化尤为剧烈。随着AI技术被广泛应用于攻防两端，传统的安全防御体系面临重构。后量子密码标准的推进、零信任架构的普及，以及供应链安全事件的频发，共同塑造了2025年网络安全的新格局。
白皮书还特别设立了"数字伦理与治理"专题章节，反映了ADSPR在2024年启动"AI治理与伦理"专题研究计划以来对该方向的持续关注。章节梳理了全球主要AI治理框架的演进路径，分析了算法公平性、AI决策透明度和数字权利保护等热点议题的最新进展。
协会秘书长张瑞晨在发布会上表示，白皮书的编撰动员了DSRI评审专家团队和专家顾问委员会的大量专业力量，力求在数据翔实的基础上提供有深度的分析视角。未来白皮书将作为ADSPR的年度品牌出版物持续发布，为数字科学从业者提供可靠的行业参考。
白皮书全文将向ADSPR资深会员和正式会员优先开放下载，预备会员可通过协会官网获取白皮书摘要版本。
In December 2025, the Association for Digital Science Promotion and Research (ADSPR) officially released the Annual White Paper on Digital Science Development (2025) (hereinafter referred to as the "White Paper"). This is the first time since the Association was founded seven years ago that it has systematically and panoramically reviewed the global state of digital science research. The White Paper was prepared over six months by the academic team of the Digital Science Research Institute, covering six core directions: artificial intelligence, data science, cybersecurity, blockchain and distributed systems, cloud computing and edge computing, and digital ethics and governance.
The White Paper points out that global digital science research in 2025 shows several notable trends. In artificial intelligence, large language models are shifting from a "scale race" toward "efficiency optimization" and "safety alignment", while model interpretability and AI safety have become core topics in academic research. Multimodal fusion technologies are maturing rapidly, and vision-language models have made a critical leap from laboratory research to industrial application in fields such as medical diagnosis, industrial inspection, and autonomous driving.
In data science, breakthroughs in synthetic data technology have provided new solutions for small-sample learning and privacy-preserving scenarios. At the same time, data governance has risen from a technical issue to a strategic one. Multiple regions around the world have introduced stricter data compliance regulations, accelerating the implementation of privacy computing and federated learning technologies.
The changes in cybersecurity are especially dramatic. As AI technologies are widely applied on both the attack and defense sides, traditional security defense systems are facing reconstruction. The advancement of post-quantum cryptography standards, the popularization of zero-trust architecture, and frequent supply chain security incidents have jointly shaped the new cybersecurity landscape in 2025.
The White Paper also includes a dedicated chapter on "Digital Ethics and Governance", reflecting ADSPR's continued attention to this direction since launching the special research program on "AI Governance and Ethics" in 2024. The chapter reviews the evolutionary paths of major global AI governance frameworks and analyzes the latest progress on hot issues such as algorithmic fairness, AI decision-making transparency, and digital rights protection.
At the release event, Secretary-General Zhang Ruichen said that the preparation of the White Paper mobilized substantial professional expertise from the DSRI review expert team and the Expert Advisory Committee, aiming to provide in-depth analytical perspectives on the basis of solid data. In the future, the White Paper will continue to be published as ADSPR's annual flagship publication, providing reliable industry references for digital science practitioners.
The full text of the White Paper will be made available for priority download to ADSPR Senior Members and Full Members, while Associate Members may obtain the summary version through the Association's official website.