My Answer to the AI Era

2025-12-02

When my daily pay rose from ___ RMB to ___ USD, I started thinking about what had caused that increase.

The conclusion I arrived at was:

  1. Improvement in technical ability

Then I kept thinking about these questions:

  1. What kind of technical ability actually counts as strong ability?
  2. What kind of technical ability is scarce and durable?
  3. How do you improve technical ability?

So for many years, the main thread running through my life has been: how to improve technical ability. The clearest example is that if a job does not help me grow, I change jobs.

So when it comes to the question of how to improve technical ability, my answer used to be:

  1. Focus on underlying principles and implementation rather than application-layer frameworks and interfaces.
  2. Develop the ability to find and debug critical flows in complex codebases.
  3. Actually participate in some genuinely cutting-edge and worthwhile projects.
  4. Reproduce some relatively low-level and hardcore projects on your own.
  5. Focus on technical ideas rather than specific implementations.
  6. Understand the essence of computation represented by Yin Wang’s courses.

Fact One

But immediately after that, I discovered an unsettling fact:

  1. There is no necessary connection between technical ability and either salary level or job stability.

I came to this conclusion for several reasons:

  1. I do not believe that my technical ability is inadequate, yet I have encountered an extremely large number of interviews that ended with not a fit.
  2. I do not believe that my technical ability is inadequate, yet I have personally seen fairly average people become leaders and interviewers.
  3. I do not believe that my technical ability is inadequate, yet my work has been very unstable.
  4. Some people are simply born into privilege.

Fact Two

At the same time, I discovered another unsettling fact: AI is changing the rules of the game.

  1. The gap between an ordinary person and a highly knowledgeable expert is now just one Gemini away.
  2. Programmers no longer need to hand-write code.

The changes brought by AI are enormous. They directly change the answer to the question of how to improve technical ability:

  1. The ability to write code is no longer important at all, and the ability to master multiple programming languages is no longer important at all.
  2. Experience within a domain no longer needs to be accumulated. Ask AI once, and it gives standard answers for everything.

Looking back, these were some of the mistakes I made before:

  1. EchoEVM: half a year ago, building something like this seemed meaningful. But today, AI can easily build a complete project like it. So EchoEVM no longer has meaning.
  2. EthBFT: two months ago, building something like this might still have exposed the limits of AI. But now AI is gradually breaking through those old limits. So EthBFT no longer has meaning either.

What EchoEVM and EthBFT have in common is that they are relatively low-level, technically oriented implementations that tried to prove technical ability through hardcore projects. But in the age of AI, that kind of hardcore coding ability is exactly what AI is best at, and the first thing it will replace.


New Question

Taken together, these two unsettling facts force a new question:

  1. How do I improve my competitiveness, scarcity, hard power, earning potential, and ability to avoid becoming obsolete?

New Answer

So for this new question, my answer is:

  1. Improve the ability to identify problems.

Explanation of the Answer

You might say: is that not obvious? The ability to identify problems has always been important.

No, this is different. Before AI, you could survive without judgment and without the ability to identify problems. Even if you only followed a product manager’s instructions to implement features, as long as you did the job of a programmer, you could still get by.

But AI replaces exactly that kind of person who only follows instructions and executes.

Identifying problems covers both technical and non-technical domains. In technical work, it means spotting whether code has issues, whether a feature design has vulnerabilities, and whether business edge cases lack constraints. In non-technical work, it means identifying real user needs and spotting major gaps in the market.

So why did I not put improving judgment into the answer? Because predicting whether the stock market will go up or down tomorrow also counts as judgment, and that kind of ability is neither verifiable nor something you can improve through deliberate effort.

Further Questions

That is still not the end of it. Given this new answer, there are two more questions:

  1. How do you improve your ability to identify problems?
  2. Unlike coding ability, how do you quantify the ability to identify problems and compare it with other people?

As for how to improve your ability to identify problems:

  1. You can only see what is lacking in what you have now after you have seen something better. So in everything, you need to benchmark upward.

As for how to quantify the ability to identify problems:

  1. On the technical side, turn the problems you identify into technical documents, design documents, and architecture documents.
  2. On the non-technical side, write down the problems you identify in places like blogs and journals, because writing can preserve the process of thinking.