AI enables us to acquire more knowledge, but we seem to be less happy than we used to be.
A subtle change may be taking place among people who are accustomed to continuous learning: we are gaining knowledge more easily than ever, yet we do not necessarily feel as happy as we used to.
When you read a book and stumble upon an exquisite viewpoint, you will get excited for a long time. When you think about a complex problem, go through repeated deliberation and finally grasp the truth behind it, you will have a strong sense of achievement. Even playing chess, playing a musical instrument or learning a new skill, seeing your own progress in continuous practice is also a delightful thing.
The advent of AI has changed the way we obtain these experiences. For a problem that used to require consulting a large amount of materials and thinking for hours, AI can give a structurally complete and quite insightful answer in just a few minutes. People will still marvel at the brilliance of the answer, but this kind of joy seems different from the excitement of suddenly figuring things out after racking one's brains.
What is more subtle is that AI can continuously provide new knowledge, new viewpoints and new frameworks. We are exposed to wonderful content every day, and we can indeed understand the principles behind it, but before long, most of it may have been forgotten. The more knowledge we come into contact with, the more things that truly feel like our own do not necessarily grow synchronously.
Thus, an interesting paradox arises: we clearly have the most powerful learning tools in history, but why is learning possibly less interesting than before?
A noteworthy explanation is: What AI has changed is not only the efficiency of learning, but also the reward mechanism of learning.
I. A large part of the joy of learning comes from "I am getting stronger"
Why can learning make people happy? Of course there is the attraction of knowledge itself, but besides that, learning also provides a very direct psychological reward.
A knowledge point that you could not master before is now learned; a problem that could not be solved is finally solved through hard work; a skill gradually becomes proficient after long hours of practice. All these experiences will bring a clear feeling that you are becoming more capable.
The growth of ability often means a certain scarcity. Mastering knowledge that others do not have, having cognition that others do not possess, and being able to solve problems that others cannot solve, not only gives people a sense of achievement, but also may bring competitive advantages and social recognition. Many people are willing to spend a lot of time reading, studying and practicing, not just because the process is interesting, but also because this feeling of constantly getting stronger is fascinating.
Therefore, there is actually a very stable reward chain behind learning: Hard work in learning brings the growth of ability, the growth of ability forms a scarce advantage, and the scarce advantage further brings a sense of achievement and rewards.
But AI is loosening this chain.
Knowledge can be queried at any time, complex tasks can be completed with the assistance of AI, and even insights that used to require long hours of thinking to form can now be generated quickly and in large quantities. Of course, learning can still make a person stronger, but this growth may not bring obvious competitive advantages as before, because other people can also use the same tools to quickly obtain similar results.
The knowledge that a person spends years mastering may have become a capability that everyone can call on; the brilliant viewpoint formed by one's own painstaking thinking may be given by AI in several versions in just a few minutes.
This does not mean that personal ability is no longer important, but there is no longer always a direct connection between the growth of ability and personal advantage as there used to be.
II. From Producer to Consumer: Another Change in the Learning Process
Compared with the change in competitive advantage, AI is also changing a more hidden thing: the role we play in the learning process.
Imagine two such scenarios. One person spends several hours thinking repeatedly about a math problem and finally finds the solution; the other person asks AI the same question, gets the correct answer in a few seconds, and fully understands the logic behind it.
Both people finally know the answer, but they may not get the same sense of satisfaction.
Because for the former, the joy comes not only from the answer itself, but also from the process in which the answer is generated. Those moments of confusion, trial, failure and sudden enlightenment allow something that did not originally exist in one's own cognition to be constructed through hard work.
This is more like the joy of a producer.
It has a subtle difference from the joy of a consumer. Making a beautiful table by hand and buying an equally beautiful table in a store can both make people happy, but the former also contains an additional sense of satisfaction: I made this myself.
AI makes it easier and easier for us to get the joy of consumers instead of being producers.
Faced with a complex problem, you can let AI give an explanation; if you think it is not wonderful enough, you can ask it to change to another angle; you can also generate five different viewpoints at one time, and then select the most convincing answer from them. Of course, people still need to understand, compare and judge, but in many cases, the main work has changed from finding answers in person to selecting and understanding answers.
Thus, we may be exposed to more than a dozen very brilliant insights in one day, each of which makes perfect sense, and we can't help but exclaim "This idea is really amazing". However, a few days later, these viewpoints become blurred. They once briefly entered our cognition, but may not leave a deep imprint like those ideas that we thought out by ourselves.
AI has made getting good answers easier than ever before, but it does not necessarily make people get the same degree of creative satisfaction. This change may be particularly strong for those who have long taken the pursuit of creating knowledge as their goal.
In 2026, the mathematics community is experiencing such an impact. The capability of AI on research-level mathematical problems is improving rapidly. The results released by OpenAI show that its internal model can already give new results to mathematical problems that have existed for decades, and more mathematical researches completed with the participation of AI have emerged one after another.
Mathematical research is a work that relies extremely on long-term accumulation. A mathematician may spend more than ten years or even decades studying a certain problem, trying different methods repeatedly, and continuing to explore in days that seem to make no progress. At the moment of final breakthrough, what is gained is not only an academic achievement, but also the satisfaction that long-term thinking finally pays off.
But AI is changing this experience.
In September 2026, Steven Strogatz, a professor of mathematics at Cornell University, even burst into tears emotionally when talking about the mathematical breakthroughs brought by AI. His collaborator Alex Townsend expressed another kind of loss: he had invested 15 years in mathematical research and was at the peak of his professional ability, only to find that AI had begun to touch or even surpass the abilities he was best at.
This feeling is not difficult to understand. For mathematicians, solving a difficult problem is not just about getting a correct answer, but more about taking a step towards the boundary of human knowledge with their own abilities.
AI can help humans solve problems faster, which is of course a kind of progress. But for individual researchers, discovering an answer in person and understanding and verifying it after seeing the answer given by AI are, after all, two different experiences.
In the field of Go, this transformation from producer to consumer appeared even earlier. After AlphaGo defeated Ke Jie in 2017, AI has gradually become an important tool for professional Go players to train. It allows players to quickly master high-level moves, but it also changes the way people learn Go.
Ke Jie once mentioned that AI has made it harder to widen the gap between players. Everyone is learning the moves recognized by AI, personal playing styles are becoming less and less distinct, and Go has lost part of its past beauty.
Behind this feeling is the subtle change of the player's role. Players originally needed to go through a large number of games, researches and trials and errors to gradually form their own understanding of Go and create moves with personal styles. But with the advent of AI, players can directly obtain better moves, and then understand, select and learn them. Players are still making progress, but the focus of learning is shifting from creating their own answers to mastering the answers that AI has already given.
The stories of mathematics and Go reveal the same change: AI makes high-level knowledge and capabilities easier to obtain, but it also makes learners increasingly play the role of consumers rather than producers. Producers enjoy the joy of exploration, construction and creation, and consumers are also happy, but their persistence and sense of acquisition are different from those of producers.
III. When the reward mechanism of learning changes, why do we still need to learn?
This change is entering the work and life of ordinary people in different forms. Writing can let AI generate the first draft, learning a complex concept can let AI explain it directly, and encountering difficulties can let AI give solutions. Even if you want to get ideological inspiration, you don't have to wait to accidentally read a good book or go through a long period of thinking. As long as you ask a question, you may get a number of wonderful viewpoints.
These capabilities have significantly lowered the threshold of learning, giving more people the opportunity to access high-quality knowledge. For some people, AI may even stimulate stronger curiosity and desire for exploration. But on the other hand, the easier it is to obtain knowledge, it does not necessarily mean that learning itself is more and more satisfying; the higher the efficiency of obtaining cognitive results, it does not mean that a person can always feel the corresponding growth of ability.
This may be the source of the crisis of meaning of learning in the AI era: Two important rewards that support learning are changing: one is the joy of establishing a scarce advantage through progress, and the other is the joy of constructing and creating knowledge in person.
When AI makes knowledge and capabilities easier and easier to obtain, people need to re-find the connection between hard work in learning and personal value; when AI begins to participate in or even undertake a large number of cognitive productions, people also need to re-understand what those experiences of exploration, trial and error and creation in the learning process mean after all. This does not mean that learning is losing its meaning, let alone that AI will definitely weaken people's enthusiasm for learning. But it does make a problem that has not needed to be explained repeatedly for a long time start to resurface.
For the education industry, in the past, more efforts needed to be made to solve the problem of how to make people learn faster and better. Whether it is improving teaching methods, optimizing learning paths, or using technology to improve efficiency, they are all built on a relatively stable premise: people are willing to invest time to acquire knowledge and capabilities, because these accumulations will eventually bring returns.
When AI continuously reduces the cost of acquiring knowledge and capabilities, are learners still willing to spend a lot of time personally going through those long, difficult and even frustrating learning processes?
When answers are getting easier to get, when capabilities can increasingly be realized with the help of tools, when learning no longer naturally means having something that others do not have, why on earth do we still need to learn in person?
Perhaps, education in the AI era not only needs to answer how to make learning more efficient, but also needs to face a more fundamental problem again: When the results of learning are getting easier and easier to obtain, what on earth makes the learning process itself still worthwhile?
This article is from the WeChat Official Account "Duozhiwang" (ID: duozhiwang), written by Tcoh, and published with authorization from 36Kr.