Top scientists respond to the teacher-student relationship in the AI era: What should be done when students think AI is stronger than professors?

"Is AI squeezing the development space of young people?" "What else surprises us?" "What should we do when students think AI is stronger than professors?" On October 30, at the "Sharp Mind Roundtable Forum" of the Young Scientists Conference of the 2026 World Laureates Forum, outstanding young scientists from around the world threw their genuine bewilderment at the predecessors who have reached the summit of science. Young people represent the future of science, and this conference was also named "Youth Defines Questions, Science Reshapes the Future." Young scientists took the stage one after another, describing their respective cutting-edge explorations in fields such as quantum, semiconductors, nanomaterials, and neuroscience, calmly answering follow-up questions from top scientists, and expressing deeper thoughts about the future in their hearts—in an era when the scientific research paradigm is being restructured, where will science go in the next ten or twenty years? How will scientists establish themselves and find their place? Is AI squeezing the growth of young scientists? The first question from the young scientists was: In scientific research, does using AI promote or hinder research? Will it affect young people's ability to carry out work independently and squeeze their development space? Seiji Fujita, winner of the 2018 Wolf Prize in Chemistry, said that there are already about 100,000 kinds of metal-organic framework materials, "We do not need another 100,000; we need to find ways to make them more stable. I cannot solve this problem; I need AI's help." "I have used AI for almost 4 years and am a super supporter of AI," said Michael Levitt, winner of the 2013 Nobel Prize in Chemistry. He said he does not believe anyone, so he will not be frightened by AI hallucinations, "Calculating the same thing with two or three methods and seeing whether the results are consistent—this is easy." He believes that using AI is essentially improving the user's own intelligence. Michael Levitt at the SharpMind roundtable forum In the view of Laurent Lafforgue, winner of the 2002 Fields Medal, the name "artificial intelligence" is somewhat misleading. "These systems operate like black boxes, merely 'weaving' text. What truly makes them seem impressive is powerful computing power and the data and corpora written by humans." This determines that AI cannot exist without humans, while scientists are always exploring territory beyond existing knowledge. As for some young scientists' worry that AI will "make them unemployed," Fujita said that around the 1990s, when computers entered research on a large scale, everyone was discussing the same question at that time. But now it seems that computers are only a tool. "I believe that for at least ten years, AI will also only be a tool." What else can surprise you? "Recently, what in your research field has surprised you?" Behind this question posed by the young scientists, the subtext was "What else can surprise you?" Unexpectedly, many top scientists cast their votes for large language models. "Even a month ago, I might still have said this was something 100 years away," Levitt said. When large models emerged, he found it hard to imagine that merely being trained to predict missing words in sentences could give rise to such impressive intelligence—as large models continue to iterate and upgrade, he is amazed by this almost every day, "We may be approaching an understanding of what intelligence actually is." Although Lafforgue believes AI cannot exist without humans, as a mathematician, he is also curious why machine learning, which merely turns "linear combinations" into "composite combinations," can make efficiency jump astonishingly, "Even today, we still do not know why it is so effective." Of course, many scientists were also delighted by discoveries in their own fields. For example, Liqun Luo, professor of biology and neurobiology at Stanford University, was delighted that optogenetics won this year's Nobel Prize in Physiology or Medicine—using light to illuminate a channel and let ions enter and exit neurons; this minimalist idea completely reshaped a research field. Perhaps this is the mindset young scientists need to maintain: challenge the unknown with endless curiosity and continuously bring surprises, big and small, to their fields and the world. What should be done when students think AI is stronger than professors? The most piercing question of the whole session was: In the AI era, some students think AI is stronger than professors and no longer need professors; some professors, especially those engaged in theoretical research, think AI is easier to train than students and no longer need students. "May I ask how you get along with students?" "I think we need an education revolution," Levitt said. In a sense, supervising an AI agent is more tiring than mentoring a human student, "because it works fast but understands limitedly." Someone at the scene mentioned that a student made a digital twin of an advisor, fed all the advisor's papers into it, and when there was a question, first asked the computer and then asked the real person. "This shows that the teacher-student relationship is indeed reversing," admitted Jürgen Caro, winner of the 2020 Eni Award. Students are better at using AI than he is, and he is being "forced" by students to make his classes perfect—students check facts online while listening, and if he says something wrong, they stand up and criticize him on the spot. So what value is left for professors? "It can only be critical thinking." Ma Jianpeng, a professor at Fudan University, reminded: If students merely use AI to expand a two-page draft paper into ten or twenty pages, it may look beautiful, but it is actually padding with water, "If students think that in this way, relying on AI, they can surpass their advisors and surpass themselves, that is definitely not a good thing."