On September 9, 2026, I was astonished to see the announcement that OpenAI had solved the smoothness and existence problem for the Navier-Stokes equations, one of the seven Millennium Problems. Unlike the LEAN formalization of Fermat's Last Theorem or further advances on constants in the twin prime conjecture, this was a genuine proof of a very difficult problem, and it included a completed LEAN formalization. That day was momentous.

I actually began thinking about whether AI could solve human mathematical problems early this year, and I even considered doing research in that direction. In my naive understanding, because of the intrinsic limits of Gödel's incompleteness theorems and Turing computability theory, we cannot design a machine that solves everything; the obstacle arises from self-referential propositions. At first I had a simple intuition: AI cannot possess human "inspiration"; it can only derive new, obvious knowledge from what already exists, and it cannot quickly find "shortcuts" by "thinking." In other words, AI only has statistical inference and exhaustive search. Therefore I thought the development of AI4Math would proceed by AI aggregating useful knowledge from massive data to obtain new obvious facts, after which humans would apply inspiration to that new knowledge, ultimately producing new systems and methods. Iterating this process, humans and machines would collaborate to solve hard problems.

But the Navier-Stokes problem, as one of the seven Millennium Problems, was attacked by AI over 88 hours, with tens of thousands of agents, and the proof was completed independently. That shattered my filter that AI can only "integrate + brute force," or at least made me realize I had seriously underestimated the capability boundary of that approach. All of this far exceeded my expectations for the pace of AI4Math development. I was left dizzy and stunned, as if struck by a two-dimensional foil attack.

Speaking of the two-dimensional foil attack, at that moment I really had a similar feeling. A problem that had consumed decades of mathematicians' concentrated effort was proven so easily by AI; human pride in "inspiration" crumbled under the statistical averaging of massive thought and brute-force methods. That belief broke in an instant, and I suddenly understood how despairing the statement "the escape velocity is the speed of light" can be. When your brain is developing at the pace of the information age—learning extremely slowly and being easily distracted—competing with the rapid development of AI is like comparing a chemical-fueled spaceship to light speed.

Against that backdrop, public opinion was swept by various despairing statements: mathematicians will all be replaced by AI, programmers will have no jobs left. I disagree. Mathematics is the mother of science, and programming is synonymous with logic. The reason mathematical problems were so easily cracked and software engineering so easily taken up by AI is not that other fields are harder to replace, but that these areas represent the hardest, most representative problems and the most profitable directions. If programming and mathematics are replaced by AI, then other industries will either be replaced even more heavily, be unprofitable, or, as is more commonly the case, become trivial problems solvable by time and computational power. When the nest collapses, there are no intact eggs; future changes will not be limited to these two disciplines, nor will they be as simple as merely "not finding a job."

I think we should not view the future with a perspective limited to the present. Human society will undergo dramatic changes in the coming years.

First, because AI currently shows no apparent upper bound in development speed and momentum, I believe industries' problems will be solved one by one in order of the value they contain. The upstream constraint of the hardest problems is actually cognitive. When we do not know the correct theory, materials, structures, mechanisms, or search directions, we must spend decades and vast resources on trial and error. If AI greatly reduces the cost of "acquiring correct knowledge," downstream engineering difficulty will change by orders of magnitude, and resources will naturally flow there. Gradually, from the Millennium Problems troubling humanity, to intractable diseases, cancer, perhaps even questions of immortality, and to deeper insights in physics and cosmology, these challenges will be overcome one by one. Given current capital inflows and the unlimited stacking of compute, as long as human society remains stable, resources will flow to directions that represent the future. People can use these discoveries to pose new conjectures, thereby rapidly driving explosive scientific progress.

Second, in politics and the social sciences we need to redefine the production relations of the new era, the role of government, the form of the state, and so on. The arrival of new productive forces means old production relations face major challenges. Because unemployment will inevitably rise sharply, radical social welfare may replace a large number of jobs in the short term, leading to a further widening of wealth inequality and the gradual monopolization of tokens and other resources. Perhaps after a breakthrough in controllable nuclear fusion, material resources will be greatly abundant and humans will have material equality; more differences will then come from varying development of individuals' spiritual lives, which could be a step toward communism. This process will be driven more by historical development and the basic laws of societal progress than by any single individual or regime. We need more research to address the problems that may arise during this transition.

Finally, and most importantly, religion, art, and music must help explain how humans should find their place in the AI era. If humanity truly moves toward communism—an era of universal fairness—it might, relative to the present, suffer from a lack of competition and scarcity of curiosity and exploratory drive, and slowly become stagnant. I pessimistically think humans might end up in a state of waiting for the "AI overlords" to pay wages, a kind of "sustaining God" status, and the human spiritual world will become increasingly impoverished.

Therefore I think that in the next few years to decades, the natural sciences and engineering will experience explosive growth. Individuals must aim to be top in their fields, rather than "lying flat and letting the AI take care of everything." After that, I believe the humanities, including social sciences, economics, and the arts, will face ever higher demands. Otherwise, humanity may be buried by the very prosperity it created.

For me and for many individuals, first, take care of yourself; do not cling to trivial issues at work or in life that harm your health. Maintaining stable emotions and a higher perspective, looking after your body—sleep, diet, moderate exercise—comes before having a good job. Scrambling for short-term gains is not worth much in the face of rapid AI development, and if poor health or bad emotions prevent you from enjoying the dividends of the AI era, that is the real loss.

Second, strive to be among the best in your field; pursue a broader vision instead of more immediate rewards. For the reasons above, catching the fast train in the first stage and making changes within your industry will become increasingly feasible. If you spend all your time scheming for promotions, raises, or exploiting others, you will only waste time and energy.

Third, cultivate the ability to discern right from wrong; do not be led astray and keep an independent mind. AI cannot replace humans because many times you need a real person who is often a better judge, more trustworthy, and who has a heart like yours and can understand your human feelings. Thus preserving what makes us human is the most important quality in the AI era. This is not merely a trick to "outrun AI" or to declare human sovereignty; it is my most basic answer to the question of how humans should find their place in the AI era. In the AI age, upholding human dignity becomes ever more important.