第一财经

Can AI pose original questions for scientists? Does a “moment of inspiration” still belong to humans?

原文:AI能替科学家提出原创问题吗?“灵光一现”还属于人类

Summary of Key Points

The 2026 Science Exploration Awards were announced, featuring the first winners born after 1995, with the highest number of female award recipients (10) in the history of the awards. The review process explicitly prohibited the use of AI, emphasizing that AI cannot yet replace the core aspect of scientific research: the ability to make original judgments, which relies on scientists' long-term accumulated “intuitive understanding” and independent thinking. The awards are tilted towards younger scientists, with funding that can be freely allocated without the immediate demand for tangible results, aiming to protect the passion and patience of young researchers.

1. No matter how advanced AI is, it can’t replace scientists in choosing research questions

Many believe that AI can assist in scientific research, but determining which questions are worth studying remains beyond its capabilities. Deng Li, the vice president of Westlake University, explains that scientists choose their research directions based on more than just logical reasoning; it often involves an “intuitive understanding” — for example, suddenly seeing a breakthrough after years of pondering an unsolved problem. This kind of insight is something AI lacks, as it does not possess the same long-term disciplinary expertise, imagination, and courage as humans. Deng Li even doubts that AI will ever reach that level. As a result, all judges signed a commitment statement this year, stating that they would not use AI to review materials or write evaluation reports, to prevent AI from diluting the most essential aspect of scientific research: originality.

2. How to determine if research is truly original? Find the most knowledgeable experts to critique it

Originality cannot be measured by simple indicators such as the number of papers or citation rates. Deng Li explains that true originality should impress top international peers in the same field, either by introducing new knowledge, changing existing perspectives on a problem, or expanding the boundaries of the discipline (for instance, using new methods to discover something previously unseen). Therefore, the proportion of international judges this year increased significantly (over 40%) to ensure that the evaluations are conducted by the most knowledgeable experts in the field. After all, while outsiders can watch the excitement, only insiders can discern the true value of the research.

3. Young researchers are in high demand: First-time winners from the post-1995 generation, and a record number of female recipients

This year, the average age of the 50 award recipients was 40.5 years, with 10 being “young scientists” (under 35 for males and under 38 for females), including the first winners born after 1995. Ten female recipients accounted for 20% of the total, the highest number ever. Pan Jianwei, the vice president of the University of Science and Technology of China, stated that young researchers are given priority under equal conditions, as they have a longer research career ahead of them. Moreover, the level of young scientists in China is indeed improving; for example, this year, two Chinese mathematicians won the Fields Medal (the highest award in mathematics), and the quality of their papers has been improving year by year.

4. 3 million yuan to be spent freely, with no pressure for immediate results

Each recipient receives 3 million yuan over five years, with complete freedom to spend the money as they see fit. The foundation does not monitor how the funds are used and does not rush for results. As a foundation official put it, “There’s no need to rush for spectacular achievements, just like waiting for a hen to lay an egg.” Why this approach? Because private funds are limited compared to government funding and are better suited as a supplement, supporting independent explorations driven by interest, allowing young scientists to focus on long-term research without the pressure of immediate outcomes.

Conclusion

This news highlights a crucial point: the soul of scientific research lies in humans. While AI can assist with experiments and writing papers, the critical decisions of choosing research questions and assessing originality still rely on human judgment. The Science Exploration Awards demonstrate how funding and patience can protect the “originality” and “curiosity” of young scientists.