Are jobs disappearing? It's just the way of working that has changed.

Under the wave of AI, the way of working, organizational forms, and personal values are undergoing a fundamental reconstruction.

Intelligence has unleashed far more productivity than imagined. The traditional workload standard of "man - days" has lost its meaning, and the traditional job model of exchanging time for pay has become ineffective.

For organizations, the change is no longer about process optimization but a species - level evolution from "managing human resources" to "harnessing AI."

Enterprises must dilute job boundaries, restructure collaboration driven by tasks, make implicit knowledge explicit as reusable AI skills, and upgrade knowledge management to a core strategy. At the same time, organizations need to break free from habitual thinking, abandon the obsession with a definite ROI, streamline non - core capabilities, and embrace the AI - native architecture.

Only by actively evolving and breaking through with new needs can one gain an advantage in cross - species competition and truly unleash the full - scale value of AI.

I hope today's content will be helpful to you.

I. AI Transforms Work: Fragmented Abilities Become an Advantage

1. When "Impossible Missions" Become a Reality

Since the emergence of ChatGPT, I've been one of the first - batch users. Every time a new AI tool comes out, I'll use it right away. By now, except for sending WeChat messages by myself, almost all other things are assisted by AI behind the scenes.

My work mainly consists of three parts: teaching, scientific research, and now there's also development work.

I've never written a single line of code before. When there was a development need in the past, I could only hire an outsourcing team. However, in many cases, our core purpose is not development itself but to solve practical needs.

For example, the teaching requirements for my leadership courses are vastly different from those for physics or computer courses. The school's unified teaching system software can only meet 20% of the common needs, and it can't meet the remaining 80% of personalized needs at all.

Now with AI, every time I start a new course, I'll create an interactive website myself, and all the learning interactions of students will be completed on this website.

For instance, when I teach leadership and communication skills, I can create simulation scenarios directly on the website. Students can directly talk to the large - model and engage in role - playing, just like a virtual sand table.

All the content is customized according to the unique content of my course. This kind of thing was completely impossible in the past. It's not about making your courseware better or your PPT more beautiful. Such ideas lack imagination. What AI can do are things that you didn't even dare to think about and couldn't achieve before.

The same goes for scientific research as for teaching.

A few days ago, we published a paper researching whether AI can have judgment. Specifically, we created a fine - tuned model to teach AI to judge whether a research idea is good or not.

Many people want to try to let AI evaluate their ideas or business plans. However, the effect of ordinary large models is not very good. Then we fine - tuned this model, and the result was quite good. So after we finished, we directly deployed it online and created an open website.

Now my students are all submitting their well - kept research ideas to let AI make evaluations. It can be used not only for scientific research but also for business plans and startup ideas.

This is another thing that was completely impossible in the past. The greatest value of AI is to turn your past dreams and fantasies into reality.

2. The Efficiency of Intelligent Agents Far Exceeds Expectations

Now the ceiling of AI's productivity and efficiency has far exceeded the imagination of many people, especially after skillfully using intelligent agents.

Some time ago, when I was teaching executives, I did a live demonstration. I wrote a book in ten minutes. I only set the core direction, and the theme of the book given by AI was "Organizational Behavior for High School Students." It also said that there was no such content in the market.

The chapters inside, such as how to understand the class collective, how to manage your emotions, how to know yourself, and how to handle relationships with classmates, are actually what students really need, but there is a gap in the market, and AI is particularly good at this.

I let AI act as the general contractor and split it into ten sub - intelligent agents, each responsible for one chapter. It was written and typeset in ten minutes. There were friends from publishing houses in the audience at that time, and they all said that the quality of the book was very high after reading it.

This is a huge impact on many people. AI can complete the workload that used to take you a month or half a year. You can even start three or five projects simultaneously and write three or five books in ten minutes. The core lies in whether you can discover a real need.

This leads to a more thought - provoking question: Things that can be produced in ten minutes theoretically have little value. How can we make them truly valuable?

Even the things you produce may not even be called books but another form. However, their essence is the same, which is to meet the core needs of users to acquire knowledge and spiritual nourishment.

Just like the "Lobster" (referring to a certain AI - related product), it didn't exist before. Its sudden appearance has spawned a new industry.

3. The "Man - Day" Workload Standard Has Become Obsolete

In the past, whether it was for R & D project management or consulting project quotation, the most commonly used unit to evaluate the workload of a project was "man - days," that is, how many people work for how many days. The underlying logic is personnel cost.

This is often seen in primary - school math problems: If Xiaoming needs 3 days to dig a hole and Xiaoli needs 5 days, these 3 days and 5 days are the so - called "man - days."

In the AI era, this standard is obviously outdated. Now the core cost is only the Token cost, and the cost of Token is extremely low. People are liberated, and the core work becomes driving the operation of Token.

The degree of AI adoption in enterprises is no longer measured by the penetration rate because it's too easy to achieve. If you install Doubao (an AI product) for everyone, the penetration rate will be 100%.

The real measurement standard is the enterprise's Token consumption, especially the Token consumption brought by the operation of intelligent agents.

At first, I didn't have this idea. When people use Doubao, DeepSeek, and ChatGPT, it's real - time interaction. You're always working, and there's no such thing as "working a full day."

The first time I had this feeling was when ChatGPT launched its professional version Pro. When you ask it a question, it may reply to you 15 minutes later. At this time, I wondered what I should do in these 15 minutes.

We immediately conducted a study to explore whether the emergence of AI would break our traditional work rhythm. The core question was: What are you doing when AI is working?

Now, people are starting to use Claude Code, Codex, and many heavy - duty AI tools. It can really complete an hour's task independently. At this time, you naturally won't keep swiping short videos but will open new windows and take on new projects.

Suddenly one day, I found that the workload can't be calculated like this.

For example, for a project, it actually took 8 hours from start to finish, but I may only have talked to AI for 10 minutes at the beginning. It gave me several options in the middle, and I made a decision. The rest of the time was just clicking "agree" and confirming the progress. At the same time, I was also advancing another project in parallel.

At this time, you'll find that the unit of "man - days" can't measure the workload at all. This project really took 8 hours from start to finish, but the time I really invested was less than 40 minutes.

So I was thinking at that time whether there should be a new measurement unit?

Now all large models are talking about Attention. In fact, for people, the core is also attention. How much attention time I really invested in Project A, how much in Project B, and how I switched between them.

4. Fragmentation Is a New Way of Working

"Fragmented attention may not be a defect now but a superpower." This is what I've realized from actual work. Without first - hand experience, I wouldn't casually put forward a "counter - intuitive" concept.

Why do people procrastinate? Because the more difficult a task is, the higher the startup cost.

For example, when I want to write a paper, the most difficult part is to open Word and write the first sentence. You have high expectations for this task, but it's hard to meet the expectations at the beginning, so you'll be stuck there. So after finally getting into the state and having your brain activated, you don't want to be interrupted and want to finish it in one go.

But with AI, you'll find that the startup cost of tasks becomes extremely low, and you can instantly enter a state of active thinking without a long - time build - up.

First, AI will work independently and naturally interrupt your rhythm, especially now that it can work for 30 minutes at a time. What will you do in these half - an - hour? Naturally, you'll do another thing, and the work rhythm has completely changed.

Second, any work done by people can be divided into two parts: one is truly original ideas, and the other is implementation.

For example, when writing a paper, the core direction and prototype are already in my mind, but to really implement this idea, refine it, and make it rigorous, 80% - 90% of the work is in the implementation. Most of the so - called deep work in the past was also spent on implementation.

But now, AI can help you complete 90% of the implementation work. On the contrary, the 10% most creative part doesn't require deep work at all.

When you're driving, you can constantly incubate and polish ideas in your mind; even when you're sleeping, you're thinking about them subconsciously. It's a low - power operation state that can produce strong creativity, and AI can help you implement your ideas.

The fragmented state can actually allow you to access more information. Anything you see while walking can be an external stimulus, and with a core goal in your mind, you'll automatically process and feedback this external information to form new ideas.

In the past, deep work was more about deep implementation. Now, AI does the implementation work for you, and fragmented creativity and insights have become the most core value.

II. AI Transforms Organizations: Fading Job Boundaries, Task - Driven

1. Fading Job Boundaries: From Exchanging Time for Pay to Exchanging Knowledge for Output

The term "job" itself is designed by human - centered organizations. Modern organizations have several basic assumptions: the first is bounded rationality, that is, people can't do everything, so division of labor is necessary.

However, after division of labor, a large amount of friction will occur in multi - person collaboration. The core of management is to manage this friction and the uncertainty of people, so jobs were created.

The underlying logic of the job punch - in system is that employees exchange time for pay with the organization. But now it's no longer the case. The core has become exchanging knowledge for output.

If you still have the mindset of punch - in, you'll find that many employees with strong AI capabilities can finish a day's work in ten minutes but won't do more.

In many startups, employees regard work as their own business. After the arrival of AI, they feel that they can do more things, so they become even busier.

Therefore, there are several obvious trends of change now:

First, jobs will definitely be merged. In the future, we'll divide work according to tasks rather than jobs, and the concept of "job" will be gradually diluted.

If AI becomes the main force in work and people are only responsible for coming up with ideas and making decisions, then many standardized and verifiable tasks in jobs can be handed over to AI, leaving only the parts that require high - level collaboration between people and AI and value judgment.

Second, jobs will be broken up and reorganized. One person will be responsible for a wider range of work, and the core work of people has become to arrange the entire work process.

The "cog - like" jobs in traditional large enterprises will disappear in large numbers. It's not that people disappear, but the way people work has changed: People have changed from cogs to system designers.

Just like in the first industrial revolution, everyone was a handicraftsman. Later, there were people who designed factories and assembly lines, and new jobs such as quality inspectors and safety officers emerged. The same principle applies now.

2. Knowledge Management Becomes the Core Strategy

In the past, people's core abilities were internalized in their minds, such as design ability, coding ability, and product - making ability.

Now the most important thing for enterprises to do is to make the implicit knowledge in employees' minds explicit and turn it into individual Skills, making it the public ability of the organization that can be accessed by the whole company.

If knowledge management was a marginal responsibility of enterprises in the past, it has now become the most important core responsibility. Because AI can be infinitely accessed, using it to undertake this knowledge will result in completely different efficiency.

I'd like to share my own practice with you.

I often write annotations on students' papers, whether they are written by students themselves or in cooperation with AI. I'll clearly mark in Markdown where the beginning is poorly written and how to improve it, where there are logical problems and how to adjust them.

In the past, these annotations were only used once. They were useless after students read and revised them. But in the AI era, these annotations have become the most valuable things.