I.

In China, almost every discussion about education today seems to revolve around AI. Yet I believe that many misconceptions have emerged in the process. At their root lies a failure to recognize one fundamental fact:

With AI's enormous reservoir of knowledge, the way human beings acquire knowledge has already undergone a revolutionary transformation.

Put simply, the educational model emerging from AI is fundamentally incompatible with the Prussian education system.

What does this mean?

As we know, the Prussian education system emerged in response to the needs of its time. Its purpose was to educate large numbers of children efficiently, and it therefore adopted a standardized, assembly-line model of instruction.

In terms of its physical organization, classrooms were designed much like factory workshops. Children of the same age were grouped together and subjected to a fixed curriculum over a period of ten or twelve years, during which knowledge was systematically delivered to them.

The curriculum itself was similarly broken down into separate subjects and predefined stages of learning, ultimately leading toward university and adult education.

But given the sheer number of students involved, the system necessarily had to disregard individual differences.

The results are obvious: students came to be divided into so-called "high achievers" and "struggling learners." It also produced a large population of children who simply did not fit into the assembly-line model of schooling.

AI directly challenges this arrangement. Its vast knowledge base makes it possible for every individual to develop an entirely personalized learning plan, and even to build a learning community of their own.

AI requires no fixed educational assembly line. In fact, this is precisely what it does best.

In other words, it does not inherently require fixed academic calendars, grade levels, standardized textbooks organized by difficulty, standardized examinations, or even conventional subject divisions.

This means that AI is fundamentally at odds with the Prussian education system.

Of course, this is not difficult to understand. The Prussian model emerged because the demands of its time made it necessary. But the times have changed.

This does not mean that education is no longer necessary. Rather, it means that the traditional methods of teaching and learning must be comprehensively and fundamentally reformed.

As the times change, AI—the defining technological force of this era—must become an integral part of education. The question is no longer whether we should have education, but how education itself must change. That has become the most important question of all.

II.

Yet, at present, whether in China or elsewhere, the education sector seems to lack the momentum for fundamental transformation.

The reason is not difficult to understand. Education is, first and foremost, a social institution. More importantly, it is a downstream institution: it generally responds to changes in society only after they have taken place. It observes those changes, identifies emerging patterns, and then translates them into practical reforms.

At present, however, changes in the broader social structure are not yet sufficiently visible. But if, one day, society's demand for talent and its criteria for selecting people change substantially, education will have no choice but to change as well.

Does this mean that education has nothing to do in the meantime, other than wait for the future?

Not quite.

Social transformation is already happening in practice, and it is happening at extraordinary speed. These changes are already putting pressure on the existing educational model. They may not yet be fatal, but the impact is unmistakable.

If education fails to respond effectively to these rapidly intensifying pressures, it will soon find itself in an increasingly difficult position.

Consider what might happen if schools continue producing students who are utterly unprepared for the changes brought about by AI. Students will gradually lose their sense of direction and their ability to navigate the world ahead. Eventually, they may begin leaving traditional schools in search of alternatives. Parents, too, will soon recognize that the teaching methods currently employed by schools are no longer adequate.

So, what should we do?

III.

Based on more than six months of practical experimentation, I have come to believe that the transformations ahead will be extensive and profound. Many aspects of education will need to change so fundamentally that our preparations must be guided by a long-term vision.

The first priority is to establish a clear understanding of what I would call the fundamental premise.

What is this premise?

It is that AI will inevitably become an indispensable part of education, and indeed one of its core elements.

Under these circumstances, what we need to build, step by step, is an AI-driven data infrastructure platform for education.

In practical terms, the entire school must become an interconnected network in which AI supports organizational structure, administration, teaching, learning, and student development.

Many schools appear to be working toward something similar. But their approach is clearly misguided.

Many schools and so-called educational experts regard AI as little more than an upgrade to smart classrooms and smart campuses. This is a profoundly mistaken view.

AI is capable of autonomous reasoning emerging from complex model behavior. It is not simply a system driven by big data. Therefore, the fundamental question is whether it can think, reason, and make judgments autonomously.

In other words, the design, deployment, and functionality of AI models must be considered systematically and from first principles. This is not an equipment upgrade. It is closer to rebuilding the entire system.

But how can this be achieved in education?

What we need are genuine FED engineers—professionals who understand both AI and education, and who possess a high level of expertise in both fields.

At present, such people are scarce, and organizations with these capabilities are even rarer. The field is largely populated by people who make grand claims without the expertise to back them up.

Most cannot even distinguish between different AI models, let alone understand how to use them effectively. Their imagination remains stuck at the idea of building a massive surveillance network, perhaps supplemented by occasional use of tools such as WorkBuddy.

Expecting them to integrate an AI-driven data infrastructure into the fundamental logic of education is, quite frankly, unrealistic.

So, what is the primary risk we face today?

It is a risk of misunderstanding.

But this is not even the greatest risk. An even greater challenge is that AI continues to evolve at extraordinary speed, while today's AI providers may not necessarily develop their products in ways that align with the needs of education.

In other words, if a school or educational institution wants to build an AI architecture that will remain useful in the future, it must be able to anticipate the direction of AI development while maintaining the capacity to adapt and continuously adjust its technological choices.

Clearly, without the necessary understanding or a sufficiently long-term perspective, institutions could easily spend enormous sums on software and hardware infrastructure that becomes obsolete within six months, ultimately turning into nothing more than electronic waste.

This is why a dedicated team must work alongside the school, continuously advancing the system as the institution develops. The school itself, meanwhile, must cultivate its own internal team throughout this process.

IV.

The second priority is to update and transform the curriculum.

At a macro level, AI is challenging the underlying logic of virtually every existing curriculum.

Should foundational education still be structured around a twelve-year system? How should we define subject-specific expertise? And how should teachers approach the teaching of their subjects?

These are all fundamental questions that demand serious reconsideration.

I suspect that the coming decades will be a period of comprehensive curriculum reconstruction. What we are witnessing today is merely the beginning.

If we consider education in terms of the humanities and sciences, general education will probably experience the greater disruption over an extended period.

Why? Because the transformation of our times is fundamentally changing how we understand the world as a whole. As a result, the established answers traditionally embedded in general education will increasingly come under pressure. Teachers will have to constantly update and renew their knowledge.

This presents an enormous challenge. Teachers must not only keep pace with the present era, but also understand the thousands of years of development of human civilization. Only then can they adequately explore questions that have no standard answers but continually give rise to new possibilities.

And only on this basis can they communicate their own thinking and insights to their students.

The general education framework with which we are familiar today may take decades to gradually stabilize into a new form. In the meantime, we are living through an era of uncertainty. There can be no assumption that we already possess a body of knowledge that will remain definitive and useful for decades to come.

However, the subjects facing perhaps the most immediate and dramatic disruption are the modern sciences, particularly STEM (+A).

The reason is straightforward: AI is becoming a major driver of scientific and technological progress. One of the most obvious consequences is that, once AI is introduced as a variable into these subjects, the barriers to entry across almost every discipline begin to fall.

Over the past few months, we have started using ESP32 boards to build a wide range of devices, from small rings and drones to robotic dogs. What we have discovered is that these projects are no longer particularly difficult for children.

Previously, the main obstacle was programming. Today, AI can write virtually all the code required for these projects. The central question has therefore shifted: can children use AI effectively?

At the same time, rapid advances in hardware—particularly the continuous evolution of development boards, expansion boards, and open-source firmware—are making many forms of manufacturing and engineering that were once extremely complex much easier.

In this process, learning mathematics, physics, science, and engineering has already moved beyond the age restrictions imposed by traditional educational structures.

Take programming, for example. Much of what was previously taught is no longer appropriate. The entire curriculum needs to be rewritten and redesigned from the ground up.

The same applies to STEM education as a whole.

The point is simple: if education refuses to change, children will inevitably turn to AI. After all, from their perspective, AI is simply far more convenient.

V.

To sum up, we have arrived at a critical moment in history. There is a great deal that can be done for education today, yet very few people are actually willing to do it.

I have thought about why this is the case. The answer is quite clear: every major reform in history has required pioneers, and those pioneers are more likely than not to fail.

Yet failure itself can serve a purpose. It clears the path for better approaches and makes future progress possible. But failure also comes at a cost. Choosing to step into this process and take responsibility for it is therefore never a decision without consequences.

For many people, the conclusion will probably be that it simply is not worth it. This is especially true from an individual perspective.

When an undertaking offers no visible short-term returns, the chances of receiving personal recognition or tangible rewards for pursuing it become extremely small. More often than not, those who take the initiative end up becoming pioneers who bear the costs of paving the way for a future society they themselves may never fully benefit from.

This is precisely the predicament facing education today.