Sam Altman: A four-day workweek is coming faster than you think

The "four-day workweek" predicted by OpenAI may arrive sooner than many people expect.

However, this is by no means a workplace benefit in the traditional sense.

OpenAI CEO Sam Altman mentioned in an interview on April 6 that the work of programmers in 2025 and 2026 will be completely different. When a developer can complete the work of an entire team in the past with the help of AI, the traditional work model will naturally not hold up.

The four-day workweek is just the tip of the iceberg. What is really being subverted is the entire set of social rules that humans have established around work, income, and time allocation.

Section 1 | The Four-Day Workweek: Not a Benefit, but a Result

Do you think the four-day workweek is a benefit given by the enterprise to employees? This understanding may be wrong.

The real reason behind it is that AI has started to do more and more work.

In the field of programming, many developers can clearly feel the experience upgrade brought by the new generation of models. The time required to complete the same task has been greatly compressed. Projects that used to require a small team to collaborate and promote can now be independently delivered by one person with the help of tools. Altman mentioned that some people's productivity has increased by two or three times, and some people even said that one person can now do the work of a team in the past. He even said that the work of programmers at the beginning of 2025 and the beginning of 2026 will be completely different.

The same is true in the field of scientific research. AI is no longer just playing an auxiliary role. It has begun to truly participate in the discovery process. AI can do tasks such as proposing hypotheses, screening paths, and accelerating the formation of experimental ideas. Altman predicts that with the advent of the next-generation model, some people will definitely sigh: It is this tool that helped me complete the most important discovery in my career.

The superposition of these changes results in a rapid increase in output per unit time.

In the past, the relationship between working time and output was relatively stable: investing more time would complete more tasks. However, now this linear relationship is beginning to loosen.

When AI greatly compresses the delivery cycle of tasks, the daily working hours are no longer a rigid indicator. Enterprises still care about the results, but the time required to achieve the results is being recalculated.

The four-day workweek is neither an idealistic institutional design nor a one-sided kindness of the enterprise. It is an inevitable result after the work mode has changed. When output is no longer absolutely dependent on working hours, spending time is no longer the only standard for measuring value.

What has really changed is time itself. In the past, you had to sit for 8 hours, but now you can adjust the working hours according to the actual tasks.

Section 2 | It's Not the Time That Has Changed, but the Allocation

If you only regard the four-day workweek as an adjustment of working hours, you underestimate this change.

What Altman really wants to express is: When AI undertakes more and more actual work, how should the income of the whole society be distributed?

The logic in the past was relatively simple: individuals provide labor, enterprises pay wages, and society maintains its operation through wage taxes. No matter how the positions change, this system of exchanging time for rewards has always been very stable.

However, after AI starts to undertake actual work, this system begins to become unstable.

On the one hand, more and more commercial value no longer directly comes from the time invested by humans, but from models, computing power, and data. The output is exploding, but the required manpower is decreasing. On the other hand, when enterprises can complete more tasks with fewer people, the part of the income that was originally distributed to ordinary people through wages will inevitably flow more to capital and technology owners.

Under this trend, some radical proposals have begun to appear within the industry and the government.

For example, establish a public wealth fund to allow more people to share the long-term dividends brought by AI;

For example, adjust the tax structure, shifting from mainly relying on personal wage taxes to taxing more on corporate profits and capital gains;

It also includes how to redistribute the wealth created by automation.

When "work" is no longer the main way to make money, can personal income still only rely on work?

If not, society needs to find a new distribution method to ensure that most people still have income instead of directly losing their source of livelihood.

The four-day workweek is not just a change in the numbers on the schedule. It is a signal: work is no longer the only source of income, and the system of paying by time must be redesigned.

Section 3 | Change Won't Wait for You to Be Ready

Reconstructing a social system takes a long time, but the development of technology will not give people enough time to prepare. The biggest crisis lies in the fact that the change is accelerating rapidly, while the whole society's preparation is obviously lagging behind.

This lag is first exposed in the security aspect.

Altman warned that as early as next year, the field of network security may face a major impact. The reason is simple: as the capabilities of AI improve, both the attackers and the defenders are upgrading. However, if the update of the protection system cannot keep up with the expansion speed of AI's destructive power, the systemic risk will increase sharply.

The same is true in the biological field. When large models can easily process complex biological information, once someone misuses it, the consequences will far exceed the scope of the laboratory. He believes that in the face of this level of ability, relying solely on the self-restraint of technology companies is far from enough, and the whole society must establish stronger defense capabilities.

However, what is more hidden and more common than security loopholes is the cognitive lag of the whole society.

So far, the vast majority of people still only regard AI as an efficiency tool to save time and optimize KPIs. However, in Altman's view, this is a serious misjudgment of technological evolution: AI is crossing the boundary of assistance and directly taking over the core work that originally belonged only to humans.

When the technology has fundamentally changed, while the public's cognition still remains in the old framework, there will be a huge misalignment: on the surface, everything is still running as usual; but in fact, the rules of social operation have already changed.

This misalignment is often very hidden until it suddenly breaks out at a certain point in time. The disappearance of jobs comes faster than expected, the entire industry accelerates the reorganization in a short period of time, and the stable career path that people rely on for survival suddenly breaks. Technological evolution is smooth and continuous, but the adjustment of human society is often lagging behind and painful. The greater the gap between the two, the more intense the resulting shock will be.

Fortunately, Altman mentioned that in the past few weeks, he has truly felt for the first time that the world has finally started to take this matter seriously. Relevant parties across national borders and industries are becoming more and more willing to sit down and discuss countermeasures.

The good news is that serious discussions have finally begun.

The bad news is that change will never stop waiting for the discussion to produce results.

Conclusion | Three More Urgent Questions

The four-day workweek may come, but before that, everyone has to figure out three questions first:

• When AI can do more and more work, what can you do that it can't?
• How can you cooperate with these tools to maximize your value?
• If work is no longer the only source of income, where will your money come from?

Figuring out these questions is far more urgent than caring about "how many days to work a week".

📮 References:

https://www.youtube.com/watch?v=XUu-i9Wbh-c&t=1352s

https://www.youtube.com/watch?v=B21KxGs8zDI

This article is from the WeChat public account "AI Deep Researcher", author: AI Deep Researcher, published by 36Kr with authorization.