Core Insights - The importance of curiosity over competitiveness is emphasized, suggesting that exploration and play are more valuable than winning, as they contribute to social value and the underlying drive of individuals and culture [1][18][6] Group 1: Funding and Research Support - Collaboration between university researchers and corporate R&D often focuses on short to medium-term practical contributions, while foundations should support non-consensus, free exploration in fundamental fields for long-term innovation culture [2][19] - Current funding for exploratory research is still in its early stages, but the awareness of its importance can significantly influence future development in this area [20] - The concept of a "reward function" in AI training is likened to funding incentives for researchers, indicating that the direction of funding can have lasting impacts on individuals and the research environment [20] Group 2: Historical Context and Examples - The reference to the 1939 article "The Utility of Useless Knowledge" by Flexner highlights the value of basic scientific exploration that may not have immediate applications but can lead to significant future contributions [21][22] - The story of the 1956 Dartmouth Summer Research Project on Artificial Intelligence illustrates how curiosity-driven research received funding despite its complexity and lack of immediate clarity, ultimately leading to groundbreaking advancements in AI [27][29][31] Group 3: Encouraging Curiosity - Curiosity should be encouraged in education, particularly in primary and secondary schooling, to foster a spirit of exploration that can adapt to the changes brought by AI [25] - A cultural shift is needed to support curiosity-driven exploration, allowing for diverse behaviors and ideas without strict judgment based on success or failure [25][26] - The potential for a global community to support curiosity-driven research is growing, with individuals increasingly willing to engage in meaningful work that contributes to society [26]

尽管这部分的支持总额，还不是那么大，还处于一个开始的阶段，但这个阶段的觉悟却深刻影响着这条 道路日后蓬勃发展的样貌和气象。借用AI大模型训练领域的说法，训练时用的奖励函数，会决定着最 终训练出的这个智能体的立场、倾向性。对人，对学者，经费支持的奖励函数，更是如此。激励的导 向，对人、风气和驱动力都会有持久的影响，而一旦这种影响形成，再去扭转它，要费比之前大得多的 力气。所以，现在说，挺合适。 我跟同行们分享了科学探索奖和新基石研究员项目的积极探索和基石铺垫，已经促成了一些鼓舞人心的 变化；当然，因还都是刚刚开始，一定得持续做扎实，才能把已经看到的对于人、风气和驱动力的影响 夯实、扩延——变革最怕（过早）宣布成功。 我跟大家引述了1939年美国普林斯顿高等研究院首任院长弗莱克斯纳那篇名为"无用知识的有用性"的文 章，与同道们共勉。钱颖一教授前些年在清华经管学院做院长时，总在跟学生下午茶或开学讲话时提起 这篇文章，意在鼓励学习经济管理"致用之学"的学生，更要珍视内在价值。"无用知识的有用性"这文章 最开始的本意，是礼赞很多自然科学的探索、基础领域的突破，当初并不因"用"而起，但几十年甚至百 年后对产业界产生意想不 ...