Original video in Chinese.
Key Takeaway
- AI has moved from a tool to autonomous agents (such as Claude Code and AI browsers), replacing repetitive white-collar work and affecting many industries (such as finance); the fate of ordinary people is grim, and passive waiting means being eliminated.
- Use the most advanced foreign AI tools regardless of cost (beyond the scope of normal cognition); Vibe Coding deepens understanding of AI’s boundaries, symbiotic collaboration, and customized productivity tools/workflows.
- ALL IN AI investment (compute such as Nvidia, nuclear power stocks), steadily gaining returns or buffering the impact; calling for action, with the community sharing practice.
With tools, there are tool users. Because tools need people to operate them, whether it’s a steam engine or a computer.
And in every previous wave of technology, although old tools were phased out, new tools would always be born, and new workers would be needed. For example, horse-drawn carriages were replaced by cars, but that created driver jobs.
So after the old workers were laid off, as long as they had not given up on themselves, after a period of training, they could still return to work.
This model has been running for more than a hundred years. Look, even our education system is designed to train workers. Otherwise, why wouldn’t schools teach things like how to start a company, or the difference between personal tax and corporate tax?
But this model, this logic, does not work in the AI era. Because AI has completely taken over tools. It is no longer new tools replacing old tools, but tools replacing the people who use them.
This is a slow-boiling-frog process. There are roughly two stages.
In the first stage, AI’s intelligence and automation are not advanced enough yet, so it still needs a user, that is, a person using AI. The productivity of these people, to be honest, is like one person doing the work of ten.
This is what I have been emphasizing for the past year: it is not AI that replaces you, but the people using AI.
And now, we are at the beginning of the second stage, at a turning point—AI is starting to take the field itself. Its intelligence and automation are sufficient, and it has already become capable of replacing white-collar work.
In fact, you can understand this by looking at AI programming.
In the early days, it was mainly code completion. The popular tool then was GitHub Copilot. It could predict the next line of code based on context, but it still could not handle complex architecture design or end-to-end tasks.
Later, Cursor went a step further and added more fully automated features. It did not just complete code; it also supported multi-line editing, intelligent rewriting, agent mode, command execution, and more.
By today, the main character has become Claude Code. It is not just a tool or a little assistant anymore; it has become the user’s programming partner, capable of autonomously completing all kinds of delivery tasks.
And from Copilot to Cursor to Claude Code, as I’ve used this whole progression, my biggest feeling—and what I have been stressing all along—is that Claude Code will absolutely not stop at programming. Its positioning is as a more foundational Agent engine.
Then there are AI browsers. Once you’ve used them, you get this feeling: intelligent, automated operations for webpages are already pretty mature.
This creates pressure from both above and below.
From above, one scenario after another is being taken over, such as programming and browsing the web.
From below, the most foundational Agent engines, represented by Claude Code, are continually maturing. Once they cross the tipping point, they can be embedded into all kinds of Agent products and complete more general-purpose tasks.
This is the evolution of AI intelligence and automation. And this evolution will not only affect programmers.
Take the financial industry as an example. When the market experiences unusual fluctuations, such as when Trump once again throws everything into chaos, you need to quickly understand the causes and effects, as well as the series of impacts it will have, and then produce a report, whether for internal use or for clients. In the past, all of that took a lot of manpower and energy. But now, with Perplexity and Gemini, you can quickly get the general picture.
So does the financial industry still need that many entry-level positions? And going one step further, if there are no entry-level positions, how do newcomers grow? And one step further still, is that once-glamorous class going to have to move?
Situations like this are definitely going to happen, and they will definitely happen in more and more industries. The first to be hit will certainly be execution-type, clearly definable, highly repetitive work—isn’t that what most white-collar workers do every day?
You might say that not doing repetitive work is fine; people are being freed by AI and can go do creative work.
Come on, do you really misunderstand creative work that badly? If you can write code, can you develop software? If you can write articles, can you become an IP? If creative work were really that easy, our system wouldn’t choose to train at scale a workforce that doesn’t need much brainpower.
So I am very pessimistic about the fate of ordinary people under this wave of AI. One wave of technological change, one industrial revolution, can take decades or even more than a hundred years. But for ordinary people, those decades are their entire life.
If you still have some fighting spirit, please don’t keep waiting passively. Passive waiting means passive elimination. I’ve set three survival rules for myself, for everyone’s reference:
First, use the most advanced AI tools at any cost.
Today, the most advanced AI tools are overseas. Domestic products are at least a year behind. This has nothing to do with patriotism; it’s just a fact. So let me ask you one question: as an individual, are you willing to fall behind together with domestic vendors?
There is one thing I believe very strongly: if you have never used the most advanced productivity tools, you cannot seize opportunities. Because the reason they are the most advanced is that they go beyond your cognitive range.
For example, Bill Gates had access to computers at an extremely rare level during his childhood at the time, which gave him experience far beyond his peers and ultimately had a profound impact on the course of his life.
AI is the same today. My own experience over the past two years tells me that being able to play freely inside the most advanced AI is truly incredibly important. Believe me, that is the dividing line between you and ordinary people.
Second, you must do Vibe Coding.
Vibe Coding is the best way to deepen your understanding of AI. You will clearly know where the boundaries of current AI capabilities really are, and what role you should play. The future will definitely be AI and humans standing on equal footing, in a symbiotic relationship. So deepening your understanding of AI through Vibe Coding and finding ways to collaborate with AI is extremely important. I’ve already shared the “architect” concept in the community.
Vibe Coding is the best way to improve personal productivity. Even if you don’t build a publicly released product, you can still customize a lightweight webpage or piece of software for your own needs. Even if you don’t make a webpage or software, you can still have AI help you build workflows. I previously shared a method for having Claude automatically create n8n workflows; everyone can look for it.
Third, invest in AI.
My investments are now ALL IN AI. Two main lines:
One is compute. For example, Nvidia—I will definitely keep holding it, and even add more whenever I have money. There are other compute stocks too, such as NBIS.
The other is nuclear power. Without energy, AI cannot develop. There are quite a few nuclear power-related companies in the U.S. stock market that are very good targets.
This is the safest plan. If AI keeps developing all the way, then my returns will be very substantial. It is also a way to prepare for the worst. If AI development really exceeds our expectations and brings huge social shock, then at least I will have enough savings.
These three rules are things I have always believed in and always practiced. The friends in the community know that all of my thoughts and progress along the way are shared. I also hope everyone watching this video will hurry up and take action. I really am not trying to spread alarm.
OK, that’s it for this episode. If you want to understand AI, want to become a super individual, and want to find like-minded people, come join our newtype community. See you next episode!