Answers have become cheaper; what's truly scarce is the courage to commit to a bet
Shaonan believes that AI has made answers readily available, but what people really face is not information scarcity, but whether they dare to choose one possibility, commit to a bet, and then give their time to it in the face of infinite possibilities.
The video won't play here. Listen to the audio instead:
The argument · timestamps estimated from transcript position
The opposite of anxiety is not calm, it's commitment
At twenty, you have countless possible lives ahead. If you keep holding onto those possibilities without committing, and drag on until forty still choosing, still waiting—what exactly are you waiting for? Shaonan's judgment is this: anxiety should point toward a concrete problem or direction. First figure out what you actually want to do, what problem you want to solve, then discuss whether and what you should learn. No one can answer your core question for you—not even AI.
— Shao NanAI is a tool that pulls you back toward the average
Large language models answer questions by predicting the probability of the next word—essentially an average calculated from vast amounts of text, the greatest common denominator of human thought. Much of the anxiety, Shaonan argues, comes from measuring yourself against an external yardstick that grows more anxiety-inducing the more you measure, because that yardstick was never meant for you. What you need to do is strive to escape these averages, find your own outlier point, and not expect AI to hand it to you.
— Shao NanAI flavor is the gap between taste and capability
Why do most people hate ‘AI flavor’—that telltale AI-generated quality? Our taste in things now far exceeds our actual ability to produce them, while AI can only produce average-level output: no point of view, no lived experience, either rehashing things everyone knows or speculating without detail. A simple test: read aloud what AI wrote. Language and text have their limits; accurately describing a song is far more complex than just listening to it.
What's worth recording is not knowledge, it's yourself
General knowledge no longer needs to be memorized by rote. Knowing about knowledge doesn't matter; knowing how to apply it does. What's worth recording is your thinking process: why a piece of knowledge matters, what views it corrected, what insight it gave you, how it connects to you. Even just excerpting a line—if you first tell it to yourself in your own words, you've truly thought it through. That act of investing attention is the value itself.
— Shao NanReading's value is in attention, not the book itself
If AI can cover a book in five minutes, what's the point of reading? Shaonan's answer: it's not that reading is inherently better; it's that human attention works this way—the results are proportional to the attention you invest. Reading requires a sustained commitment of attention over time. The examples and ideas in a book that aren't directly aimed at your goal are often what spark new thinking. Spending the same time asking AI repeated questions about a topic could yield equally rich returns—what matters isn't the tool, but how much attention you bring and what goal you bring it to.
Each week, have AI recommend only three pieces of information
Shaonan offers a specific filtering method to push back against algorithmic spoon-feeding. The baseline: no recommendations from the past six months. First, news: using Hamming's ‘You and Your Research’ lens—are you working on the most important problem in your field?—have AI recommend one news item based on that. Second, a paper, and require AI to explain why it's worth reading. Third, high-entropy content: recommend something tangentially related (but not directly) to your usual discussion direction, perhaps art or design, to push back against fragmented information consumption.
The point of learning well: you still get to choose
Humans have infinite choices but cannot choose how they are born. Yet at every moment, even when not actively choosing, they're still making a choice, developing toward some possible future. Shaonan's point: people only get anxious when they have choices; when they don't, they just work. You don't get anxious when a grizzly bear is chasing you. Existential crisis and crisis of existence don't happen at the same time. Right now, what concerns him most is his children's growth, how to be an agent that cares for people rather than efficiency, and understanding himself.
— Shao NanAsk why you'd bet, not what you should learn in the AI age
The conversation closes on a more fundamental question: rather than asking what you should learn in the age of AI, the better question is what you actually want to commit to, what you're willing to invest time and attention in over the long term. As tools can do more and more, if you don't know what you want to use them for, each new possibility becomes a burden of ‘should learn this’—and in the end it's just exhausting. Possibilities may be infinite, but the time you can invest is always finite.
— Shao NanIn their own words · checked verbatim
The universe is desperately trying to pull you back toward the average, and AI is its best tool for doing it.
宇宙拼命想把你拉回平均值，AI 是它最好用的工具
Shao Nan19:31
If you can finish a book in five minutes, what's the point of reading?
如果只花五分钟就能看完一本书，读书这件事还有意义吗？
Rather than what's happening in the world, I care more about when the sweet osmanthus flowers will bloom in Hangzhou.🌼
比起世界正在发生什么，我还是更关心杭州的桂花哪天会开🌼
Shao Nan1:11:15
Glossary
- OPC
- One-person company: using AI tools to accomplish, as a single person, what previously required a full team
- OpenClaw
- An open-source autonomous AI agent that rose to prominence in early 2026, capable of executing tasks on behalf of users
- Amor fati
- A Latin phrase meaning the attitude that all events in life, including suffering, are inevitable and should be accepted and embraced with love
- Zhengzhou gang (郑州帮)
- A grassroots traffic-operation group that mass-produces anonymous accounts and soft-advertising notes through low-cost assembly-line methods and distributes them at scale
- Lindy Effect
- Things (books, ideas) that will not naturally die out increase their expected remaining lifespan the longer they have already existed
- Shikinen Sengu (式年迁宫)
- The Ise Grand Shrine's tradition of reconstructing the main sanctuary on adjacent land every 20 years, then demolishing the old one—a practice that has continued for over thirteen hundred years
How to listen
Founders, investors, and knowledge workers who have long relied on AI for research but increasingly don't know what to learn or what to believe.
54:40-58:37: Discussion of historical anecdotes about Hangzhou's ancient pagodas, Mangshan, and Qian Hongchu—tangential to the AI theme; skip if you want to go straight to the main argument.