What Should We Learn in the AI Era?｜A Conversation with 于红: Three Skills That Won’t Become Obsolete

Summary

• At the peak of the AI boom, 于红 walked away from an eight-figure carry and a 15-year investing career spanning China Renaissance, GGV and Meituan Longzhu to build a children’s SEL education business, guided by the logic that “investing is about the next 10 years, while education is a model for the next 20 or 30.” The trigger was a 2023 conversation in which she was asked about her life mission and “cried uncontrollably”; she concluded that AI would radically reshape education, while going all in on AI with her team only strengthened that calling.
• Her answer to what children should learn in the AI era is a framework rather than a conclusion: rank subjects by the Lindy effect—Chinese language > English > math > coding—because “coding probably won’t need to be learned.” Chinese is not just knowledge but the ability to think and understand the world, even the foundation for interacting with AI; English still merits enough effort to reach a practical “60 or 80,” while AI may soon address much of the marginal value above that level. But this must be paired with a principle of fit: a child with an IQ of 150 who wants to build rockets should still learn the fundamentals of coding.
• Her decision to build in SEL was a classic white-space-market call: homework platforms such as Zuoyebang, Yuanfudao and Sparks are already throwing thousands or tens of thousands of people at Chinese, math and English, while the lack of social-emotional ability is “already an extremely serious problem in China today”—important, and largely unaddressed. UNESCO’s statistics on non-academic competencies put SEL at two-thirds of the total; overseas systems have taught it for nearly 30 years. The biggest obstacle is Chinese parents’ unfamiliarity with the term, so the strategy is to start with highly informed parents and expand from there: “five or 10 years is just the beginning.”
• Her end-state view of education rests on a disruptive premise: the knowledge monopoly, education monopoly and “monopoly on ranking people” represented by universities could be eroded in the AI era—Tencent has already begun recruiting at the high-school level, while overseas companies are hiring high-school students directly. The objective should therefore be the child’s life at 30, not admission to an Ivy League school; the head of Moonshot School has said success could even mean “a student received an Ivy League offer and turned it down.”
• On “good schools,” she brings contrarian data: Hengshui High School’s Tsinghua and Peking admissions fell 70% this year, which she attributes to its no-poaching rule; overseas research she cited suggests that for students with similar rankings, attending a “good school” may produce a negative long-term effect, although she admitted she had not examined the specific reasons closely. Shangdi Primary School’s strong placement rates, she says, are partly explained by high-IQ software engineers sending their children there. She also dismissed the theory that people are determined by the average level of the 5 or 6 people around them: “This theory is genuinely ridiculous.” Her happiness equation is 50% genetics, 10% circumstances and 40% personal agency.
• Her investing playbook is heavy on intellectual honesty and post-mortems, because the private-market cycle is so long—one of Hangzhou’s “six little dragons,” she says, may have been funded in 2014 and only listed in 2026, leaving ample room for self-deception. Reviewing an English-learning company she passed on before it rose to a $3B valuation or more and eventually collapsed, she acknowledged that both “too expensive” and “market too small” were wrong: it had more than 1M users, and families were willing to spend RMB10K-20K a year. That lesson led to her investment in Sparks Math; she says taste must be learned by shadowing world-class partners, because following the wrong person for 8 years can leave you with a portfolio of bad investments.
• Wang Xing’s investing style has 3 pillars: extreme long-termism (“5 or 10 years is long term? 30 years is long term”), slow thinking—founder meetings start at 2 hours—and investment discipline, with the fund seeking to be the largest outside shareholder in every deal and investing as much as possible once it has commitment. 于红 cited Zhipu, Unitree and several listed AI companies as examples. Her happiest investment moments came from seeing her theses validated—Zuoyebang and Full Truck Alliance both reached $10B valuations when China had only 14 internet companies at that level, by her count—while her biggest pain was FOMO. Her antidote was focused, sequential research, which led to investments in Zuoyebang, Full Truck Alliance and Musical.ly, now TikTok.
• She sees a positive mindset as the most durable capability, with openness essential for both founders and learners. Smart people’s biggest problem is often a lack of openness: they focus on being “the person who is most right” rather than doing “the thing that is most right.” Her linguistic fix is to stop saying “I have an idea” and say “I have a hypothesis.” A positive mindset can be trained through daily, seconds-long exercises in savoring joy, recalling joy and expressing gratitude, which reshape neural circuits; as AI makes knowledge readily available and shortens the time needed to acquire it, children can—and should—start early on the question of what kind of person they want to become.

Deep dive

1. Fifteen years in top-tier investing, then an abrupt turn into children’s education

• 于红 is 41, a Zhejiang University computer-science graduate, ENTJ and Aries from China’s Northeast. After graduating in 2010, she spent 15 years in investing: China Renaissance, GGV and then managing partner at Meituan Longzhu. Her current product, 兔咚咚, is “China’s first online SEL education product”; the team has 12 people.
• SEL stands for Social Emotional Learning: how to understand yourself, interact socially with others and adapt to society. Citing UNESCO statistics on non-academic competencies, she says SEL accounts for two-thirds of the total and has been taught in multiple overseas markets for nearly 30 years. It is “a highly scientific academic system with observable growth and outcomes.”
• Koji notes that this episode was a “two-year appointment”: two years earlier, 于红 had told him she was at OpenAI, prompting him to suggest a discussion about AI applications. By the time they sat down to record, she had left both VC and the AI industry altogether.

2. The moment she left: “crying uncontrollably” and a life mission

• This was not an impulse decision. At university, she had already concluded that “education is the most important thing in the world, without exception.” When Duan Yongping visited Zhejiang University for talks in 2007 and 2008, she asked a question from the audience and was recorded. When she joined Meituan Longzhu, she told the CFO upfront: “Investing is a medium-term career plan for me; my long-term life mission is education.”
• The turning point came in 2023, when an elder she had just met asked about her life mission. After answering “education,” she says, “I started crying uncontrollably”—asking through her tears, “I don’t feel anything inside. Why am I crying nonstop?” The elder said perhaps the time had come. When she told her husband at home, he responded gravely: “It’s time. You should also work for yourself.”
• AI was a catalyst, not the reason she left in the opposite direction. Her team had stopped looking at SaaS and new energy and gone all in on AI. “The deeper I interacted with AI, the stronger the calling became,” she says, as she saw the era bringing seismic change to education.

3. The Lindy effect and the principle of fit: what should children learn in the AI era?

• The core is a way of thinking, not a fixed answer: “The longer something has existed in the past, the longer it may continue to exist in the future.” Applied to Chinese, English, math and coding, Chinese is the oldest and coding the newest, so from today’s perspective, “coding probably won’t need to be learned,” followed by math, English and Chinese. She stresses that this is a framework, not an absolute answer.
• Chinese is not merely knowledge but a capability: the ability to think and understand the world. “Even writing prompts and figuring out how to interact with AI are highly important.”
• Modern schooling has existed for only about 200 years. Before that, humans spent much more time learning how to behave and socialize. Those skills, along with attitudes toward life, will become more important in the AI era.
• But the Lindy framework must be overlaid with the principle of fit—structure and each child’s position within it. A child with an IQ above 150 who wants to research rockets can still spend substantial time on math and the fundamentals of coding. “This is not an absolute formula”; the key is allocating time proportionally, not making a binary judgment about whether a subject should be learned.

4. English only needs to reach 60; AI will reshape how we learn

• Her view on English is about degree, not direction. The amount of painful time people born in the 1980s spent learning English “will definitely no longer be necessary in the new era.” A relatively modest investment should get a child to a practical “60 or 80”; beyond that, the return on each additional hour declines, and AI will soon help solve and improve much of it. English remains valuable as a tool, but the investment should be proportionate.
• New tools will inevitably change the method. She built her child a customized character-learning program in which each new character earns points that can be exchanged for doll clothes and hairstyles; each session lasts only 5 minutes. The principle is personalized learning based on observing the child.
• She identifies a supply gap: English has mature graded-reading systems, while Chinese graded-reading material and curricula are “incomplete or even absent.” AI may eventually give children “painless graded reading,” creating substantial future demand and room for experimentation. “Making learning genuinely enjoyable for children is extremely important—perhaps the most important thing.”

5. Why SEL: giants crowd the core subjects while no one tackles social-emotional ability

• The topic was selected with an investor’s logic. Chinese, math and physics already have “a huge number of smart minds working on them”: Zuoyebang, Yuanfudao and Sparks each have thousands or even tens of thousands of people studying how to teach traditional subjects better every day. “They don’t need a team like ours.”
• By contrast, the lack of SEL “is already an extremely serious social problem in China today.” It is important, and no one is addressing it. That is the problem the team actually wants to solve. 于红 had invested in online education before and therefore had a firsthand view of the supply landscape.

6. The full SEL system: self, others and the world

• 于红 corrects the instinctive definition: “emotional stability” accounts for only about 20% of emotional capability. The full system has 3 parts: the self—self-awareness and emotional capability; others—social skills and conflict resolution; and the world—responsible decision-making, a positive mindset and learning how to learn. 兔咚咚 organizes these into 6 dimensions and 12 capabilities. “This really is the foundational capability most needed in the AI era.”
• She also reverses several Chinese cultural assumptions. Keeping emotions off one’s face and having a deep reserve do not equal strong emotional capability. A “highly sensitive child” simply experiences more emotional fluctuation and sensation; the more sensitive the child, the more important emotional development becomes.

7. Emotions are signals: acceptance comes before regulation

• Her foundational view is that emotions are a protective mechanism evolved over hundreds of thousands of years. They should be treated “as signals, not as something negative to be eliminated.” Otherwise, sensitive children may ask, “Why do I have so many negative emotions? Is there something wrong with me?” and turn the emotion into further self-attack.
• The sequence cannot be reversed: accept first, then understand, and only then regulate—5 steps in total. “If someone remains unable to accept themselves, regulation is ultimately castles in the air.” That is why many courses on emotional regulation leave students with plenty of knowledge but no ability to use it.
• The long-term result is counterintuitive: accepting emotions leads to greater emotional stability. A self-accepting person is more internally grounded, less prone to anxiety and less likely to generate emotional turmoil through social comparison.

8. What the 4 basic emotions mean—and adults who fear insects

• Every emotion carries information. Happiness is not merely a pleasant feeling; it tells you that you want to move toward something and repeat it. A child’s talent should not be identified by what they are good at—“many smart people can do almost anything well”—but by what they do with genuine enjoyment. Fear signals danger ahead; anger signals a violated boundary or unfairness; sadness signals that something or someone important has been affected.
• The story that carries the point is about an adult friend who had feared insects since childhood. Adults told him, “What is there to be afraid of?” He could neither stop being afraid nor accept the fear, so shame accumulated and followed him into adulthood. After studying psychology and learning acceptance, he learned to coexist with the fear. “Nothing about the situation changed, and the emotion didn’t change either, but with an accepting attitude,” he could proudly say, “Even though I’m afraid of insects, I can still go camping with everyone.”
• With children, if a child is afraid of thunder, tell them that “being afraid is natural and normal.” Repeated guidance helps the child accept the emotion. “Accepting your own emotions is the most important step in accepting yourself.” Parent classes also include a review: accept first, stay with the child through the emotion, then examine why it happened and what it means for the child.

9. Outcomes can be observed: “little hands on the belly”

• Parents can track 3 observable indicators. The child is not supposed to stop getting angry—“Never think, ‘My child doesn’t lose their temper anymore.’ That may mean something is wrong”—but the duration of outbursts should shorten, their frequency should fall, and the child’s ability to identify and understand emotions should improve.
• The most entertaining case involved a course teaching children to pause and breathe, using the gesture “little hands on the belly.” After one child’s emotional-recognition skills improved, he saw his father about to lose his temper, put his hands on his father’s belly and said, “Little hands on the belly.” His mother reported that the father’s angry episodes at home became much shorter.

10. SEL is learned like piano: 3 kinds of knowledge and a first lesson refined for a year

• Drawing on Bloom’s taxonomy, she divides learning into 3 categories: knowledge, skill and attitude—behavioral tendencies and values. The piano and sports analogies are designed to make one point clear to Chinese parents: “After you learn it, you still have to practice and apply it in real situations.” SEL is not knowledge mastered by listening once.
• A consultant friend supplied the best evidence: “Why, every time I’ve visited you this year, are you still working on the first lesson?” The team spent an entire year refining the opening lessons, using user testing to determine how children could absorb and master 3 different things at once: knowledge, skills and attitudes.

11. 3 course formats: stories that land, scenario practice and problem-solving

• The first format looks like a cartoon, but “the cartoon is not the most important thing. It draws children into learning; what really reaches them is the story.” Humans have transmitted values through stories for thousands of years.
• The second is scenario-based practice, an AR format with 4 to 5 scenarios a day. The third gives animated characters a problem and asks children to apply what they have learned; the experience of solving the problem through their own application reinforces both skill and attitude.
• Deep breathing illustrates the full sequence. On day 7, a story about cutting in line at kindergarten introduces Teacher Dashan’s rule: pause first, then take a deep breath. More scenarios are designed the next day. On day 3, a colleague spends 2 days building a town plunged into chaos by a power outage; the child uses deep breathing to help everyone solve the problem, with a concrete “breathing kit”—smell a flower, blow out a candle. “Concrete memories are extremely important for children.”

12. Parents are the biggest obstacle; the one-line definition is teaching children how to achieve happiness

• The friction comes from parents, not children. Chinese parents are unfamiliar with both the 3 letters SEL and the 6 Chinese characters for Social Emotional Learning. “These capabilities were seriously lacking in our own growth over the past 20 or 30 years,” so parents do not understand what children are learning and conclude that time would be better spent on Chinese, math and English. She acknowledges this is “the biggest challenge we face today.” The strategy is to find highly informed parents first and then influence more families. “I keep saying that 5 or 10 years is just the beginning for this team.”
• Asked to define SEL in one sentence, she answered candidly: “This is very difficult; our team thought about it for a long time.” The current version is that SEL “teaches children the capability to learn how to achieve happiness.” The 8-character objective is “an abundant inner life, positive action.” Children know what they want, may find meaning and a mission in life relatively young, build stable relationships and know how to put themselves in a state of enjoyment rather than constant internal depletion. Koji notes the alignment with Moonshot School’s values; 于红 says she chose the school for her child before the AI era.

13. Building SEL means raising herself again: resolving more than a decade of “dishonesty”

• Asked whether doing SEL also means re-raising herself, she answers: “Absolutely. I’m learning every day.” The idea that emotions are signals resolved a professional problem that had troubled her for years. She could not accept dishonesty from anyone; anger triggered an intense reaction and others felt targeted. She refused to regulate it because “if I accepted other people’s anger, I would be betraying my values.”
• The solution was to interpret the signal. Anger could mean a boundary had been triggered—which was clearly not the case, because “this has nothing to do with me”—or that something was unfair. “Was I seeing dishonesty as unfair because I insist on honesty and integrity?” That realization freed her. But the sense of unfairness “was still located in the small self”; the relevant question was whether the other person’s behavior helped or harmed the organization. In investing, people often share only favorable information and withhold the bad. The job is to decide whether the deal is worth investing in, not fixate on what someone failed to disclose.
• She accepted an editor’s summary: “When dealing with others, don’t judge their intent; judge their actions.” The operating rules followed. If the project is good, communicate in good faith while pointing out the problem. If it is bad, speak privately and encourage the person to abandon it or share a different view. That balances values, colleague relationships and organizational responsibility. “This troubled me throughout more than a decade of my career, and one sentence in SEL solved it.”

14. Critical thinking in practice: what you saw, what you thought and updating your view

• 于红 poses a live question: a friend walks past without saying hello—what do you think? Koji answers, “I’d worry that I had offended him.” The breakdown is simple: “he didn’t acknowledge me” is what he saw; “I offended him” is what he thought; the resulting low mood and hour of suspicion are the emotion. These are 3 different layers. “What you think may not be true, but that doesn’t mean it is necessarily untrue”; you need evidence.
• The key step is to convert “what I thought” into “a possibility,” then ask, “Are there any other possibilities?” Perhaps the friend genuinely did not see him or was not wearing contact lenses. The final and most important step is to update the view when new evidence appears.
• Her child’s friend on a smartwatch deleted him. The child immediately concluded, “He’s not my good friend anymore.” Her training was to make him consider other possibilities; when they later met and asked the friend, the first interpretation proved not necessarily correct. Rigidity develops because “he has already invested so much effort in that thought and found so much evidence to prove himself right. By the time you ask him to consider another possibility, it’s too late.” Updating one’s view is itself an attitude that requires practice. “People who insist on their own views often lacked this practice when they were young,” although openness also has a genetic component.

15. To compete or not: the 8 hours outside school are the school-selection metric

• Koji admits he oscillates between “happiness matters” and “maybe we should still compete harder.” 于红’s framework is that both competition and non-competition are matters of degree. Happiness matters, and learning matters, but learning happily matters most. Studying hard does not mean excessive competition, nor does it mean letting children drift.
• Her practical metric: children spend only about 8 hours at school. “The time outside those 8 hours is what parents most need to care about.” If parents at a school are all using the after-school hours for extra tutoring, “that is not a school I believe should be chosen in the AI era.” It places children in an environment where they still have to devote enormous effort to Chinese, math and English and have no time to develop other capabilities.
• Her 3-tier test for a good school starts with whether children like going. “If most children say they don’t like school, I think that is the most important indicator.” Second, children need enough free play after school and the ability to find companions—“free play does not mean playing alone.” Only third come sports, music, art, SEL and social practice.

16. Age 30 is the endpoint: Ivy League schools are not the goal, and AI may erode the university monopoly

• Pressed to choose between an intense school with strong admissions and a relaxed school that may not lead to a top university, she says choosing Moonshot School had already answered the question. She first visited in 2022 or 2023 and paid tuition a year early after hearing its philosophy. The headmaster’s definition of success persuaded her: “It is not about how many Ivy League students we produce. If a student receives an Ivy League offer and turns it down, that is when Moonshot has succeeded”—because the school develops people with strong self-awareness who know what they want.
• Her timeline ends at 30. “Many people treat university as the endpoint, but it isn’t. I hope my child’s life at 30 is the purpose of education.” The exam system does not assess psychological state, meaning in life or stable social relationships. “No one thinks these things are unimportant, but how many people include them when choosing a school?”
• She calls this her “inner radical take”: as the knowledge monopoly, education monopoly and monopoly on ranking people represented by universities gradually erode, pre-university education could help each person become a better version of themselves. The signs include Tencent beginning to recruit in high schools, while overseas companies are more aggressive, with multiple firms recruiting high-school students directly and skipping university.

17. If that future never arrives: a vocational-school stand-up champion and self-actualization

• Koji asks what happens if the monopoly system still exists 15 years from now and the child ends up at a bad school. 于红 says it comes down to “how much effort you are willing to invest in taking responsibility for your choice.” Her own attitude was, “I would rather die than repeat a year, even if it meant going to a junior college.” “I have always wanted to become myself, rather than let external things define me.”
• Her example is this year’s stand-up comedy champion, who attended a vocational secondary school. Stand-up was once a niche art, and many performers were PhDs or returned from overseas study. “Even if that future does not arrive in a world-shaking way, social diversification is certainly the trend.” Many Ivy League graduates and academic high-fliers remain anxious for years; the relationship is “not a linear metric.”
• Her substitution is to cultivate wisdom, character and the ability to understand the world rather than a “success index.” “I don’t think this should be called success. It should be called self-actualization in one’s own life.” People with higher happiness levels are often better able to self-actualize.

18. The doors opened by elite schools versus the choices behind them; happiness is 50/10/40

• She acknowledges that an elite-school background opens doors, but “what matters more is the choices behind it.” At graduation she rejected the universally highest-paying path of public-markets investing and chose venture capital. In 2021, when the market was hottest and she had 4 companies about to go public, she left GGV for the entrepreneurial Meituan Longzhu. “It is precisely because I had so many choices that I became who I am.” The counterexamples are Jack Ma and Pony Ma, neither a graduate of the very top schools; and the workers who enter ByteDance or Tencent today—are their life and mental states necessarily better? “All of that is a question mark.”
• The science of happiness assigns roughly 50% to genetics, only 10% to circumstances and 40% to personal agency: how I interpret my circumstances, my mindset and what I actively choose to do. That makes her skeptical of determinism based on one’s family of origin; she says family background accounts for only that 10%.
• Koji suggests the 40% is also shaped by one’s family. 于红 pushes back: adults can always “raise themselves again.” At the company’s weekly Tuesday sessions on life experiences, one SEL teacher who had been deeply hurt by her family of origin “raised herself all over again” and became an excellent teacher. An art-design teacher with a junior-college degree considers himself lucky “like an alien prince.” Neuroplasticity supports the view: the brain remains malleable throughout life, although it is easier to change in childhood.

19. The empirical dismantling of elite-school mythology: Hengshui’s 70% drop and the “good-school negative effect”

• Her own background sets the stage. She did not attend a key high school and was the first student from her school to enter Zhejiang University. She largely taught herself English; her teacher would say, “What’s the answer to this question? 于红, explain it,” forcing her to prepare to the level of the teacher. Both her primary and middle schools have since closed. “What my schools gave me was actually not that important.”
• The key reverse-causality data point: the number of Hengshui High School students admitted to Tsinghua and Peking “fell 70% this year,” which she attributes to the school’s ban on poaching top students. “When evaluating a school’s educational quality, you cannot separate it from how many of the best students it recruited.” Some international high schools use the same tactic, bringing in top students in the second year, posting results and awarding scholarships in the third, leading parents to believe the school can “transform” their children.
• An even harsher finding: in overseas, US and other-country research she has read, of 2 children with roughly the same top-20% ranking, the one who attends a “good school” may end up at a lower percentile after graduation. “The effect of a good school is negative,” although she admits she has not examined the specific reasons closely. Shangdi Primary School’s strong placement rate is partly because large numbers of high-IQ software engineers in Xierqi send their children there. The marketing chain—“If you can’t get into a good kindergarten, you can’t get into a good primary school…”—is wrong. “Life is absolutely not planned in such a linear way.” On Moonshot School, she offers questions rather than advice: “I do education, but I’m sorry, I’m only clear about SEL education.”

20. A direct clash over peer effects: “This theory is genuinely ridiculous”

• Koji puts forward the idea that people are determined by the average level of the 5 or 6 people closest to them, and that the value of elite schools lies in their excellent peers. 于红 is unsparing: “I thought about it after I got home, and I felt this theory was genuinely ridiculous.” It exploits the human instinct to compare: comparison creates anxiety, anxiety creates drive, and the result is greater anxiety.
• Her counterexample is herself. Her university motto was “Follow your own path and let others talk.” Ten years later, she had discarded the phrase—“because I had already done it.” Self-actualization is about the self, not one’s rank relative to others.

21. The missing ability to choose: young people who have never made a hard choice

• The most important skill for an investor is the ability to choose, so she always asks young candidates: “What is the hardest choice you have made in your life?” The answer surprises her. A substantial share have never made an important choice; they simply followed the post-graduation hierarchy—investment banking, consulting, investing and now AI. They are products of an exam system that turns “99 out of 100” into a single-dimensional comparison.
• The most painful example was a Tsinghua PhD asked why he pursued a doctorate. He answered, “Back then, only people with good grades could do a PhD. I couldn’t waste my score.” “I was sitting behind the table, and my heart felt extremely cold.”

22. Giving up an eight-figure carry: begin with the end in mind

• When she left GGV, a partner called to ask, “Have you calculated how much you’ll lose by leaving?” She had not. The partner later calculated it for her: “Definitely more than RMB10M”—an eight-figure sum. She refuses to package the decision as proof of a resolute persona. “It was still a complex decision.” The core method came from The 7 Habits of Highly Effective People: “begin with the end in mind.” It was a revelation.
• Her reasoning ran as follows. She had enormous enthusiasm for managing teams; only months after joining China Renaissance, she was already running training and was nicknamed the “model queen.” Watching others grow gave her “an extremely strong sense of joy,” so she wanted an entrepreneurial investment firm. GGV was already too established: all 6 global partners were in the Forbes top 100, and the strongest partner had apparently ranked in the top 5 or even top 3 in prior years, although she could not remember exactly. “There wasn’t that much room for me to create value.” Money matters—“it absolutely isn’t unimportant”—and her family background could be described as poor. But if you begin with the end in mind, you have to make difficult choices. “Only difficult choices force you to interrogate what matters most and move toward self-actualization.” Of 100 opportunities, “99 are temptations, and even 10% are disasters.”

23. The hardest choice is a partner; major decisions should be made in a good emotional state

• Asked whether giving up the carry was her hardest decision, she thought carefully and reversed the premise: “None of my career choices was the hardest choice.” There are hundreds of thousands or millions of jobs. “Choosing a life partner—that is truly difficult.” She and her husband fought and adjusted for years before reaching their current state as a “perfect couple.” After every major fight, deciding whether to give up, separate or divorce—that was harder.
• Her decision framework includes beginning with the end in mind and “never making a major choice when you are in a bad emotional state.” In a good mood, the prefrontal cortex is more active and open; in a bad mood, you simply react and cannot make a rational, comprehensive choice. When a close friend said she wanted a divorce, 于红 first asked how things were outside the marriage. The friend had just moved and changed jobs; her overall emotional state was already poor, so her view of the marriage was naturally distorted. 于红 says she has also made every job change when she was in a good state.
• Decisions also require multiple perspectives. On relationship problems, she vents to female friends and then seeks views from male friends to supplement her perspective. The foundation of beginning with the end in mind is self-awareness—“the most important part of self-awareness is where my future is”—and that future cannot be imagined into existence. Broad, extensive practice is essential.

24. Exploring by crossing options off: 2 years of job hunting and 3 screening criteria

• Her philosophy of choice is elimination. When filling out university preferences, she said, “I rejected every major with a clearly defined career path”—law, medicine, the military and teaching—because she wanted a path with more possible futures. “I need to know what I don’t want to choose. That is important. I spend the remaining time exploring.”
• She spent 2 years looking for a job after China Renaissance. One excellent early-stage firm asked how much time she wanted to spend on deal sourcing versus research. She answered 50/50; the firm said early-stage investing was mainly sourcing. Her response was not disappointment but, “Great. I’m not suited to early stage. I can cross another option off my life.” At GGV, partner Jenny asked what kind of firm she wanted. She answered, “A top-tier firm investing in Series B and C.” Jenny said, “That sounds a little hypocritical.” 于红 explained that she had spoken to early- and late-stage firms and eliminated them one by one.
• Her 3 criteria for leaving GGV were a world-class partner—“I’ve worked with world-class partners, so I know what that means”—the ability to raise capital consistently and an entrepreneurial culture. “After filtering through these 3 criteria, there may have been only 2 or 3 firms in China.” The same applies to teaching children to choose: parents must guide them away from irreversible dangers involving health or life, but children have to experience smaller pitfalls themselves, get back up and review what happened.

25. The most important thing in investing: intellectual honesty and post-mortems

• Her answer is that investing is about “learning and growing from every mistake.” In private markets, that requires intellectual honesty. Public-markets performance is visible; private-market cycles are so long that “people easily deceive themselves.” One of Hangzhou’s six little dragons may have been funded in 2014 and listed only in 2026. Reflecting on the investment outcome takes too long to support learning.
• Her post-mortem involved a famous English-learning company she met at Series A but passed on. Its valuation later climbed to around $3B or more before the company eventually collapsed. A dishonest investor would say, “See? I was right,” and learn nothing. Her honest review was that both her judgments—that the product was too expensive and the market too small—were wrong. More than 1M Chinese families were willing to spend RMB10K-20K a year on English lessons. That lesson led her to invest in Sparks Math, where families might spend several thousand yuan to more than RMB10K a year.
• The discipline extends to a quarterly cadence. “Every 3 months, you should honestly review the predictions you made and choices you made the previous month—what was right, what was wrong and which models can be iterated and improved.”

26. Learning by shadowing and developing taste: private markets require the right mentor

• Private markets take an 8-year cycle to validate an investment judgment. “It is not a process where an individual can learn and improve quickly,” so finding the right mentor is essential. At GGV, she was fortunate to work one-on-one with 3 partners. Her method for training subordinates is shadowing: whenever she joined a boss or senior colleague for a discussion or project review, she wrote down every question the person asked that she had not considered. “The answer is not the most important thing. Highly intelligent people reach broadly similar answers to the same question; the question itself is more important.” When you know all the questions your boss will ask—and can ask better ones—you have mastered the first stage.
• The next stage is judgment. “It is somewhat like aesthetic taste.” To evaluate a founder, one person may say, “The product is copied and has an overseas prototype—great,” while another says, “He hasn’t demonstrated innovative thinking.” A novice cannot tell which is right; Meituan itself was originally a copy. Whose judgment shapes your taste “is extremely important.”
• Her conclusion is blunt: “Follow the wrong person and you learn the wrong taste. After 7 or 8 years, you discover that none of your investments are very good.” An 8-year cycle makes the cost of learning the wrong lesson extremely high. “That is why I don’t think private markets are a particularly good place to train people.” It is also why her criteria for an entrepreneurial firm included world-class partners: taste cannot be built or tested in the short term.

27. Wang Xing’s investing style: 30 years is long term, slow thinking and the largest outside stake

• The first pillar is extreme long-termism. When people say “5 to 10 years is long term,” Wang Xing replies, “5 or 10 years is long term? 30 years is long term.” The second is slow thinking: meetings with founders “usually start at 2 hours,” versus the half-hour snap judgments common among investors. He performs extremely deep analysis, thinks about fundamentals and long-term trends, and uses definitions and models extensively.
• The third is investment discipline: be the largest outside shareholder in every deal. “Largest” does not mean being first or reflect ego; it means investing as much as possible so the return from a successful investment is maximized. 于红 cites Zhipu, Unitree and several listed AI companies, saying Meituan generally required the maximum allocation. Koji summarizes the logic: if you invest at all, you need extreme commitment, and once committed, invest as much as possible. 于红 says this is both a style and a strategy—and it works.

28. The happiest part of investing: validating the thesis, not the return

• Before China’s tutoring crackdown, Zuoyebang had already reached a $10B valuation, and Full Truck Alliance also reached $10B after listing. At the time, she calculated that only 14 Chinese internet companies had truly reached $10B. When people asked whether she was surprised, she said no. “Their development matched my original judgment. That was what made me happiest. Whether they listed or generated how much return, honestly, I didn’t really care.”
• Her financial motivation is unusually low. When her husband occasionally makes money trading stocks and calls to share the good news, she asks, “How much money did you have originally?” The same exchange happens 5 or 6 times a year. “He says it’s incredibly satisfying to tell you.” She is highly sensitive to numbers, but “these things don’t move me.”
• The common thread between investing and education is the time horizon. “Investing is about the next 10 years; education is a model for the next 20 or 30. Today I’m studying how to influence a child’s model for the next 20 or 30 years—and I find that incredibly enjoyable.”

29. The pain of investing is FOMO; her answer is focus through “crossing things off”

• FOMO is the least enjoyable part: “It feels as if, to avoid missing something, you have to cover as much as possible.” Her natural disposition is the opposite. She spent 2 years studying education and invested in Zuoyebang, 1 year on logistics and invested in Full Truck Alliance, then studied overseas markets and invested in Musical.ly, now TikTok. Colleagues were surprised she could invest across industries. “Actually, it wasn’t like that. I studied them one by one with extreme focus.”
• She did not look at crypto, P2P internet finance or O2O—though she acknowledges teams had different divisions, and “saying I wasn’t interested in any of them would be fake.” She passed on cross-border e-commerce based on a thesis that users were simply buying different products and could ultimately be captured by broader platforms such as Tmall, Taobao and JD.com; she did not follow the wave even when Yangmatou gained scale. She studied B2B platforms for 6 months to less than 1 year before concluding the sector was not suitable for investment.
• Private markets are always changing, and the market still requires investors to catch those changes. Education offered relief from FOMO: humans have not changed much over the past thousands of years, while SEL is the refined product of a century—or even millennia—of accumulated religious and philosophical thought. “There isn’t a new theory or model suddenly appearing and telling you to study it tomorrow.” She no longer has to fear missing out.

30. What founders lack most is openness: “I have a hypothesis”

• Her founder model has 3 elements: being visionary—seeing 3 steps ahead, understanding the whole picture and forecasting the future, closely linked to self-awareness and critical thinking; leadership—which is fundamentally social capability plus conflict resolution; and openness. Schools teach almost none of this. “More than 90% of the capabilities of an entrepreneur seem not to be taught in school.”
• Openness is the biggest weakness. “Most founders look extremely smart, but many smart people are not open. They are used to being the person who is most right. They focus on whether they can become the person who is most right, rather than whether they can do the thing that is most right. Those are very different.” One reason she admires Wang Xing is his ability to hear multiple sides and remain highly open. Learning has 2 dimensions: IQ, or learning by oneself, and openness, or learning from others.
• When managing an investment team, she changed the language: “Don’t say, ‘I have an idea.’ Say, ‘I have a hypothesis.’” A hypothesis invites discussion and is inherently open; it keeps the focus on the matter rather than the person and avoids the feeling that “I must have made another mistake.” Many people resist updating their views because they fear being wrong. 兔咚咚’s own story contains the same lesson: “We are simply making the matter clear. There is no right or wrong here.”

31. The capability that never goes out of date: a positive mindset, trained in seconds

• Asked to give children one capability that will never become obsolete, she answers: a positive mindset—the final module in the SEL system. “Once a person has a positive mindset, they can already solve many problems.” A positive mindset is itself a discipline, centered on one’s perspective.
• The exercise is as specific as eating an orange: “When I eat half an orange, can I feel that I’m happy?” Savor the happiness in the present, deliberately recall happiness from the past and practice sustained gratitude. “Do this a few times a day, perhaps for only a few or 10-odd seconds each time, and the brain’s neural circuits will change.”
• Deliberate practice is necessary because humans store far more negative memories than positive ones, a product of evolution. “When you go hunting, is failing to find fruit more serious, or being eaten more serious?” The brain naturally stores negative experiences more deeply. “That is why we need to practice happiness and positive experiences more.”

32. Closing: AI has not changed the answer, but it has changed the allocation of time

• Koji’s closing observation is that AI has not changed much: an abundant inner life, positive action and a positive mindset “seem to have little to do with AI.” 于红’s correction is about time. Before AI, people had to accumulate extensive knowledge before acquiring skills, and the time required to learn knowledge kept increasing. That is why competition intensified. AI makes knowledge readily available and interactive, shortening the time spent learning it.
• The time released flows toward “the human part.” A child building a rocket no longer needs to study every piece of source material from scratch; efficiency in getting things done rises. “I can start exploring what kind of person I want to become from an early age—and learning this early has a stronger impact and shaping effect.” Self-awareness and a positive mindset therefore become more important in childhood, which is the real impact of the AI era.