李国杰,赵伟.提高基础研究能力和品味的思考[J].中国科学院院刊,2026,41(6):1259-1265.

提高基础研究能力和品味的思考

Viewpoints on elevating capability and appreciation of basic research

李国杰^1*
中国科学院计算技术研究所 北京 100190
LI Guojie^1*
Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China
赵伟²
深圳理工大学 深圳 518107
ZHAO Wei²
Shenzhen University of Advanced Technology, Shenzhen 518107, China

         纯基础研究;定向基础研究;人工智能驱动的基础研究;稳定支持强度;范式变革

        pure basic research;mission-oriented basic research;AI-driven basic research;stable support strength;paradigm shift

        一个国家基础研究的整体能力和品味，不仅与科研人员个人的努力有关，还取决于国家的科技体制机制和政策。文章从全面深入理解基础研究的内涵、提高基础研究的投入比重、增强基础研究稳定支持力度和人工智能（AI）驱动的基础研究范式变革等角度，提出思考和建议。包括：到2030年，纯基础研究与定向基础研究投入之比应从3.5∶6.5调整到4.5∶5.5，总体保持自由探索与国家目标并重；全国稳定支持的基础研究经费占比达到35%—45%；人工智能驱动的科学研究（AI4S）很可能先于通用人工智能（AGI）进入“不可逆阶段”，真正危险的不是“AI变聪明”，而是“人类失去最终判断权”等。这些建议可供科技决策部门参考。

        The overall capability and appreciation of a country’s basic research are not solely determined by the efforts of individual researchers, but also hinge on the country’s scientific and technological institutional mechanisms and policies. This article presents several viewpoints and suggestions from the perspectives of comprehensively and deeply understanding the connotation of basic research, increasing the proportion of investment in basic research, enhancing the stable support for basic research, and the paradigm shift of AI-driven basic research. These suggestions include: by 2030, adjusting the ratio of pure basic research to oriented basic research from 3.5∶6.5 to 4.5∶5.5, maintaining a balance between free exploration and national goals; achieving a stable support for basic research funding that accounts for 35%—45% of the total; AI4S is likely to enter the “irreversible stage” before AGI, and the real danger is not “AI becoming smart”, but “humanity losing the right to withdraw”. These suggestions may serve as references for science and technology strategic planning agencies.