Group 1: Technological Innovation and Achievements - In 2025, China made significant advancements in scientific research, particularly in quantum technology and lunar studies, achieving original breakthroughs [1] - The global innovation index ranked China 10th, reinforcing the foundation for high-level technological self-reliance [1] - Major breakthroughs in lunar science were achieved with the Chang'e 6 mission, revealing new lunar oxidation reaction mechanisms [2][3] Group 2: Quantum Computing Developments - Quantum computing is identified as a key future technology, with China focusing on achieving full autonomy in its development [4] - The third-generation superconducting quantum computer "Benyuan Wukong" has been launched, participating in numerous application collaborations across various sectors [4][6] - The development of quantum computing is likened to building a rocket, emphasizing the importance of both hardware and software integration [4] Group 3: Brain-Computer Interface Technology - Brain-computer interface technology is evolving to connect biological intelligence with machine intelligence, with significant clinical advancements reported in 2025 [7][9] - The technology is moving from unidirectional reading to bidirectional interaction, enhancing its potential applications [7] Group 4: Digital Intelligence and Industrial Transformation - Digital intelligence technologies, including cloud computing and AI, are seen as essential tools for enhancing innovation capabilities across industries [10][11] - The integration of digital technologies into industrial processes is expected to transform manufacturing paradigms and improve efficiency [10] Group 5: Hydrogen Energy and Renewable Resources - Hydrogen energy is recognized as a crucial component of the renewable energy revolution, complementing electricity and promoting low-carbon transitions [12][13] - China has made progress in mastering hydrogen fuel cell technologies and establishing related industrial chains [12][13] Group 6: Research Talent Development - The cultivation of research talent is emphasized as critical for supporting high-level technological self-reliance, with a focus on aligning educational programs with national strategic needs [14][15] - Young researchers are increasingly contributing to scientific advancements, with a significant proportion of key research personnel being under 45 years old [15] Group 7: Support for Basic Research - Basic research in China is receiving increased attention and support, although challenges remain in funding and stability [16][17][18] - The government is implementing reforms to enhance the support for basic research, including funding mechanisms and cultural improvements [16][17][18] Group 8: Science Popularization and Innovation - The relationship between scientific innovation and public science education is highlighted, with efforts to engage the public in scientific knowledge [19][20] - Enhancing public understanding of science is seen as foundational for fostering future research talent and innovation [19][20]

本次大会为期两天，由GAIR研究院与雷峰网联合主办，高文院士任指导委员会主席，杨强院士与朱晓蕊教 授任大会主席。 作为观测AI技术演进与生态变迁的重要窗口，GAIR大会自2016年创办以来以来，始终与全球AI发展的脉 搏同频共振，见证了技术浪潮从实验室涌向产业深海。 2025年，是大模型从"技术破壁"迈向"价值深 耕"的关键节点，值此之际GAIR如期而至，携手智者触摸AI最前沿脉动，洞见产业深层逻辑 。 大会上，之江实验室科学模型总体组技术总师，天壤智能CEO薛贵荣博士亲临现场，为参会者带来了一场 精彩纷呈的演讲分享。 " 大语言模型受限于「语言的边界」，无法理解高维度、跨模态的 科学数据。 " 作者丨胡清文 编辑丨徐晓飞 12月12日， 第八届GAIR全球人工智能与机器人大会 在深圳正式启幕。 薛贵荣博士指出， 以大语言模型为代表的AI技术虽已在多个学科研究中展现出潜力，但其本质上仍受限 于"语言的边界"，难以真正理解高维度、多类型的科学数据，更无法独立完成可验证的科学发现。 基于此，薛贵荣博士系统分析了大语言模型与科学基础模型之间的本质差异，并详细阐述了之江实验室所 研发出的 021科学基础模型在突破语言 ...

目前，021模型已服务地球科学、天文学、生命科学、材料科学等多个领域，成为打破学科边界、激发 创新思维的"科研伙伴"。(完) 来源：中国新闻网 编辑：董文博 之江实验室021科学基础模型首次亮相 突破语言局限 中新社杭州12月18日电(鲍梦妮)浙江之江实验室18日在杭州举行021科学基础模型创新合作大会，首次 全面展示021科学基础模型及系列领域科学模型研发进展。 据悉，021模型构筑形成跨学科知识、跨领域推理、跨语言理解(覆盖204种语言)三大基石，具备出色的 科学推理能力，能够深入分析、推导、验证多类科学问题。 "语言所表达的维度，远远低于科学所需表达的维度。"之江实验室科学模型总体部技术总师薛贵荣认 为，科学数据涵盖时间、空间、能量等多个维度，是对复杂物理系统演变规律的高维表征。为解决科学 问题，科学界需要突破语言空间的局限，研发科学基础模型，构建"科学空间+语言空间"于一体的更高 维空间，建立跨学科数据之间的深层连接，变革科学研究范式。 对此，之江实验室研发团队探索将科学数据和文本语料编码到统一的高维空间，让模型能够识别、处理 科学数据，认识并解决复杂科学问题。经过近万次实验，团队形成了模型训练框架， ...