The more powerful AI is, the more valuable these eight capabilities become.

The other day, an IT leader from a certain company complained to Lao Yang.

He said, "Our boss keeps talking nonstop about how AI outperforms humans these days. He used to listen to our opinions, but now he directly brings AI-generated proposals to us and asks why we cannot deliver them."

I asked him, "Do you think your boss is right?"

He stayed silent for a while and said, "AI is indeed faster, more comprehensive, and cheaper than us. But there is one thing it can never beat us at: it has no idea why that project of our company last year failed, why the sales team and the finance team quarreled at the meeting, and what the customer really meant behind the line 'I'll think it over'."

AI's advantage lies in "knowing how to do things"; but what it does not know is "which task to prioritize, whether the approach is correct, and how to get others to work together on it".

What makes things trickier is that this change arrives quietly and abruptly. In the past, if you were not competent enough, your leader would point it out to your face; now AI can do your work fast and perfectly, and no one notices that you are no longer capable. By the time you realize the problem, it is already too late.

Then what capabilities do we need to improve in the AI era?

AI can give you answers, but it cannot raise your unique questions

First, critical thinking capability

AI is best at generating answers for you. But the quality of the answers depends on whether you ask the right questions.

If you ask it "how to improve customer satisfaction", it will give you a bunch of generic suggestions: optimize service processes, strengthen employee training, and establish a feedback mechanism. All these points are correct but useless, because it has no idea why your customers are dissatisfied, which link your service gets stuck at, and why your employees are unmotivated.

AI cannot dig out these points, only you can.

But Lao Yang would like to remind you: when encountering problems, do not rush to ask AI, but ask yourself three times first "what is the core of this matter". Gradually form the habit of finding flaws in the "standard answers" given by AI. If you believe what your boss says that "AI is better than you" and outsource your thinking to large language models, you will lose for real.

Second, question-raising capability

Critical thinking capability is for asking yourself, while question-raising capability is for asking AI.

AI is an amplifier that magnifies your ability to raise good questions. As we all know, for people who know how to ask good questions, AI saves 80% of their time; for people who do not, AI only generates piles of useless garbage.

When asking AI to write a proposal, some people only input three words "write a proposal", and after AI outputs a bunch of nonsense, they complain that AI is useless. Others treat AI like a human collaborator: they clarify their identity, the background of the task, the goal to achieve, and the required format; if they are not satisfied, they follow up with questions like "why", "can it be better", and "what are the potential risks".

Third, judgment capability.

AI is a generation machine. It can give you ten proposals, twenty viewpoints, and thirty risk points in one second. But it will never tell you which one to choose.

It uses fluent language, rigorous logic and a confident tone, so you are easily convinced by it. But essentially it is just a machine that assembles words according to probability. Its confidence does not mean its output is correct.

Therefore, judgment is the most scarce resource in the AI era. A master is the person who can pick out the "truly correct" option from a pile of seemingly correct ones.

So you need to learn to figure out "what counts as correct" before asking AI. When facing its confident answers, ask yourself reflexively: "If it is wrong, what is the worst possible outcome?"

AI works faster than you, but it does not care what to do and who the output is for

Fourth, knowledge accumulation capability

AI has access to all public knowledge of humanity, but it knows nothing about your company's business processes, the unspoken rules of your clients, and the pitfalls you have stepped into.

These "private exclusive knowledge" cannot be generated by AI, and can only be built through long-term practice. In the future, your competitiveness does not lie in "how much you know", but in how much unique and professional data and materials you have in hand that can be fed to AI.

Do not regard scrolling through short videos every day as learning, which only leaves a shallow impression in your mind, makes you feel amused for a moment, and then leaves nothing behind.

If you have worked for ten years, but find all your experience cannot be applied when you switch to a new company, that means what you have accumulated is only "experiences" rather than "valuable assets".

Fifth, product building capability

AI greatly improves the efficiency of "creating things". A business staff can build a data dashboard with workbuddy in less than an hour.

If the IT department still stays at the stage of "receiving demands, writing codes, and delivering outputs", it will be replaced at any time. You need to shift your focus from "developing functions" to "building products": identify pain points, design solutions, verify the value of the product, and make iterative adjustments.

To build a really good product, you need to go to the front line to observe. Watch how business staff operate your system, see at which step they get stuck and at which point they complain, and find out which functions have not been opened by anyone for three months after launch. You can never figure these out just by sitting at your desk.

Sixth, execution capability

The execution capability we are talking about here does not focus on how many extra hours you work overtime, but on whether you can verify your assumptions quickly and adjust your direction in time.

The market changes extremely fast and the opportunity window is very short. The old model of "three-year planning and one-year construction" can no longer keep up with the pace. What we need now is to move fast in small steps: develop the minimum usable version first, run through a core scenario, get data feedback and user feedback, and adjust as soon as possible.

Allow yourself to make mistakes. The cost of trial and error is low, while the cost of stagnation is the highest.

Do not confuse execution with blind obedience. If you only do exactly what your boss tells you to do, finish the task and hand it over without caring about the actual effect, that is not execution, but perfunctory work.

No matter how perfect the proposal is, it is all in vain if the leader does not approve it

Seventh, expression capability

In the AI era, the technical threshold has been lowered, but the communication threshold has instead risen.

The business department can generate codes in plain language. If you cannot translate "technical constraints" into "business risks", you will be regarded as a stumbling block. Your boss cannot understand what data governance is, and he will only think you are making excuses.

The accuracy of your expression directly determines whether an AI project will succeed or fail.

Use more metaphors and tell more stories, instead of talking too much about technical principles. Do not talk to your boss about the parameters of large models, he will only think you are showing off. But do not go to the other extreme by saying "it cannot be done" as soon as you open your mouth, you need to be able to explain "how it can be done successfully".

Eighth, connection capability

AI can write emails, generate reports and analyze data for you. But it cannot handle cross-department coordination, upward management, and customer relationships for you.

The value in the AI era never comes from "human plus AI", but from the full chain of "human plus AI plus human". You need to connect the capability of AI with business demands, and then connect it with the collaboration of your colleagues.

Therefore, people in the IT department should not always stay in the office, go to the business department more often, and listen to what problems they are really troubled by. Take the initiative to help others solve small problems in daily work, and trust is accumulated little by little in this way. You need to understand the anxiety of the business department, the pressure of your boss, and the difficulties of your colleagues. AI cannot give human warmth, but you can.

Final summary

In the AI era, you need to be a "real human with unique irreplaceable core traits".

Those who thrive better and better in the AI wave are not the people with the strongest technical skills, but the people who clearly know what they have that AI can never take away.

So stop spending all your energy on "learning which AI tool to use". Tools are updated several times a year, but your unique core traits will never change. Spend your time on improving your capabilities of thinking, questioning, accumulating, executing, judging, expressing and connecting. These are the things that AI cannot take away or replace.

This article is from the WeChat Official Account "Xiangjiang Digital Review" (ID: benpaoshuzi), written by Lao Yang, and authorized for release by 36Kr.