Summary of Key Points
This article highlights two contrasting phenomena: Harvard graduates' resistance to AI, and Fudan University students' approach of designing questions that cause AI to make mistakes. It illustrates two distinct mindsets in the age of AI—one of anxiety and resistance, and the other of proactive control. The central argument is that AI is revolutionizing the fundamental logic of education and the rules of competition. The core competence of the future will no longer be simply "answering questions" but rather "creating questions, making judgments, and managing AI." The true competition will be between "humans + AI" versus those who rely on traditional work methods. Education must shift from nurturing "question answerers" to training "AI adjudicators."
Attitudes: Anxiety and Resistance vs Proactive Control
Harvard students express hostility towards AI, fearing that it will take their jobs, alter education, and ruin their life prospects—essentially treating AI as an enemy, out of fear of being replaced. In contrast, Fudan students use AI as a tool or even a challenge, seeking to identify its weaknesses. The difference between these two attitudes lies in their focus: the former focuses on what AI can do, while the latter considers what AI cannot do and how to utilize it. It's like the difference between someone who fears cars will replace carriage drivers and someone who studies how to drive or build better vehicles; the outcomes are vastly different due to this difference in mindset.
The Change in Educational Logic: From Testing Knowledge to Understanding AI
For the past 200 years, education has revolved around teachers creating questions and students answering them, emphasizing memorization and standard answers, as the industrial era required obedient followers. However, now AI can perform mathematical calculations, write code, and translate faster and more efficiently than humans. Testing who remembers the most or calculates the fastest is meaningless when compared to AI's capabilities.
The purpose of Fudan's experiment was not to test AI but to ensure that students understand AI better so they could create questions that stump it—identifying misconceptions in knowledge points, logical boundaries, and even AI's limitations (such as its attention span and inability to handle unanswered questions). This shifts the educational goal from merely mastering knowledge to understanding both the knowledge and how AI functions.
The Future's Top Students: Those Who Can Design Rules
The data from Fudan is noteworthy: 50 out of 51 students were able to make AI answer a question incorrectly, but only 4 were able to make AI score zero on the entire test. Even the most advanced AI system, Claude, could not defeat all of them. This shows that occasionally finding AI's flaws is easy, but systematically defeating it requires genuine expertise.
How did these high-achieving students do it? Some used multiple AI systems to generate questions together, others challenged AI with extensive data, and still others created questions without clear correct answers to test its ability to handle such situations. What they were competing for was not just solving problems but designing them, utilizing AI, and assessing the results. The future's top students will be those who can establish rules to make AI work for them, rather than being confined by predefined rules.
The Nature of Competition: "Humans + AI" vs Those Who Work Alone
Many mistakenly see AI as a competitor, but this is a misconception. The highest-scoring student at Fudan did not work alone; he used multiple AI systems to collaboratively create questions, which he then reviewed and optimized. The key is for humans to direct AI's actions.
This will be the case in all industries: doctors won't be replaced by AI, but those who use AI can outperform those who don't; lawyers won't be replaced by AI in legal analysis, but those who utilize it can work more efficiently; bosses won't be replaced by AI in strategic decision-making, but those who use it can advance faster. The real competition will be between those who know how to use AI and those who do not.
Educational Goals: From "Executors" to "AI Adjudicators"
Fudan professor Xiao Yanghua emphasizes the importance of becoming an "AI adjudicator" rather than just an executor:
- In the past, the focus was on finding standard answers; now, it's about judging whether AI is correct or not.
- In the past, memory was key; now, it's about defining goals, establishing rules, and making value judgments.
- In the past, tasks were executed; now, decisions are made.
After all, the true leaders of society have never been those who simply answer questions. Who compiles stock indices? Who sets accounting standards? Who creates exam questions? These abilities to define problems and establish rules are still beyond AI's grasp and represent the core values of humanity in the future.
Conclusion
The anxiety of Harvard students is genuine, but the approach taken by Fudan students is more instructive. AI is like the steam engine during the Industrial Revolution—while you can fear it, it's more important to learn how to use it. The most valuable people of the future will not be those with the most knowledge but those with the strongest judgment, those who can control AI, and those who can create questions and establish rules. The mission of education is to prepare individuals to be "AI adjudicators" capable of working alongside AI, rather than being eliminated by it. This is the profound lesson we can learn from Fudan's innovative testing approach.