Summary of the Core Content
This article discusses a major upheaval in the mathematics community that has not been seen in a century: For centuries, mathematicians have worked at a leisurely pace, with a single difficult problem often taking decades or even centuries to solve, with many scholars spending their entire careers on just one proof. This year, however, AI has suddenly become incredibly powerful, solving unsolved mathematical problems that have been around for decades. Even untrained individuals, with the help of AI, can produce work that meets the standards for publication in academic journals. One of the world's top mathematicians, Terence Tao, is alarmed, calling this the most severe crisis in the mathematics community in over a century and stating that there are only a few months left for the industry to adjust. If they do not quickly change the old rules and redefine the value of mathematicians, the discipline, which has been passed down for thousands of years, could fall into the hands of tech companies like OpenAI. Not only would the work of mathematicians be fundamentally altered, but the entire human mathematical knowledge system could become incomprehensible. Tao is leading global mathematicians in pushing for change to preserve the core values of mathematics before the situation gets out of control.
---
Detailed Explanation of the Content
1. Why Does the Usually Slow-Moving Mathematics Community Only Have a Few Months to Respond?
It may be hard for outsiders to understand: Mathematicians have always been known for their patience. Even during the major crisis that shook the foundations of mathematics at the beginning of the 20th century, it took decades to rebuild the system, and no one panicked. So why are they now counting down in months?
The impact of AI is truly transformative: In the past, producing a mathematical proof was a rare and difficult task, and it was common for a qualified paper to take years to write and for peer reviews to take half a year. Now, it's like a group of trucks 100 times faster than horses suddenly rushing onto an old road, crushing the entire old system:
- In the past, it was impressive to produce a few significant new proofs a year; now, AI can generate several such proofs in a week, leaving mathematicians overwhelmed by the workload and blocking the old review process.
- Solving difficult problems used to be the exclusive domain of mathematicians, but now, even amateurs can produce work that meets academic standards with the help of AI, turning what used to be a specialized skill into something anyone can do.
All the flaws and weaknesses of the old system have been exposed by AI, leaving no time for gradual adaptation. If changes are not made quickly, the entire industry will fall into chaos.
2. Don't Get It Wrong: AI Solving Math Problems Does Not Mean It Can Replace Mathematicians
Many people think that AI solving problems is a good thing, as it would free mathematicians from tedious work. However, Tao sees this as a crisis because there is a major misunderstanding about the role of mathematicians: The ultimate goal of mathematics is not just to find answers but to understand the logic behind them and to transform that logic into something that everyone can understand, reuse, and pass on to the next generation.
AI can take you to the destination, but it doesn't show you the path there. The proofs it generates are often based on common knowledge that humans can understand at a glance, while it skips over the most critical and innovative steps. Moreover, AI's problem-solving process is a black box; we don't know how it arrives at the correct answer after countless attempts. As a result, we end up with a conclusion that is correct but with no understanding of how it was reached, making it impossible to apply the method to other problems or to teach it to others.
If mathematicians' job were merely about producing proofs, AI would have replaced them long ago. What makes them irreplaceable is their ability to explain proofs clearly, connect new findings with existing knowledge, determine which research directions are worth pursuing, and pass on their judgment and intuition to the next generation—tasks that AI cannot perform at all.
3. An Even More Threatening Trend: Tech Companies Will Define What Constitutes Good Mathematics
The real danger is not the loss of jobs for mathematicians but the gradual shift of power in the mathematics community:
The primary goal of tech companies using AI is to optimize their models; their main selling point is how many long-standing problems they have solved, without caring about the logical continuity of mathematics. Top mathematicians like those who have won the Fields Medal have already started working for AI companies. If nothing is done, the evaluation standards in mathematics will be determined by these companies: Those who solve the most problems will be considered the best, and those who produce answers quickly will be regarded as outstanding scholars, regardless of whether their proofs are understandable or useful for future generations.
It's like a family-owned restaurant that has been run by descendants for thousands of years, following traditional methods. If it's acquired by a delivery platform, the platform will focus on speed and volume, ignoring the quality of the food and the ability to pass on the culinary skills. In the end, the traditional recipes will be replaced by processed food, and the thousands of years of mathematical knowledge will become an incomprehensible pile of AI-generated results.
4. Mathematicians Are Taking Action to Save the Situation
The changes being pushed by Tao and his colleagues are not about resisting AI but about leveraging its capabilities:
- First, they are asserting their autonomy by signing the Leiden Declaration, stating that the output of mathematics is more about human understanding, clarity, and judgment, and not about what tech companies define as valuable.
- They are testing the true capabilities of AI by using previously unpublished problems that cannot be found online, to avoid the illusion of AI innovation based on trained datasets.
- They are giving AI a proper role, using it to solve smaller, less challenging problems to collect data and drive progress in the discipline, similar to how particle accelerators in physics generate massive amounts of data.
- They are changing the academic rules, for example, by creating a new video-based academic journal that evaluates the ability to explain findings clearly to peers, emphasizing the importance of readability.
5. A Warning for Students: Overreliance on AI Can Lead to Irreversible Decline in Skills
Tao warns students that relying too much on AI for learning mathematics can lead to a decline in their abilities. There are already cases of students with great potential who, after using AI to complete assignments for months, lose their ability to perform basic logical reasoning. Mathematical training is about developing the intuition to understand and derive conclusions, similar to learning to ride a bicycle. If you always rely on a support wheel (AI), you can never master balance on your own. Tao sets a practical rule: You can use AI, but only if you first solve the problem without its help and can explain the entire process clearly to a mentor, confirming that you truly understand it before using AI to streamline the process. Otherwise, you're just wasting your brainpower.