On September 9, the Department of Computer Science and Technology at Tsinghua University released the Redbook Artificial Intelligence General Education. It defines the theoretical foundations and historical mission of AI general education (AIGE) and proposes a curriculum system, together with a plan for putting it into practice, built around “cognitive autonomy” as its core goal. This series uses plainer language to walk through the Redbook's overall structure, key concepts, main claims and central evidence.
As the first article in the series, this piece gives a unified overview of the Redbook's overall line of thought, setting out its core ideas and the logic of its argument.
When we talk about AI general education today, on the surface we are talking about a new course. The deeper question, however, is this: when machines take a large part in the production and life of human society, what kind of people should education produce? When artificial intelligence gradually becomes society's information intermediary and cognitive infrastructure, what will a society rely on to form shared understanding, sustain public reason and secure the long-term development of humanity?
This is the question of our time that AI general education has to answer, and it is the weighty sense of historical mission behind Tsinghua's Redbook Artificial Intelligence General Education.
1. The Social Mission of Education
Let us begin with the basic mission of education.
Many people have grown used to reducing education to “the transmission of knowledge”. This matches everyday experience: what those being educated need first of all is specialized knowledge that will get them a job and a living.
Yet education carries a mission far more fundamental than passing on knowledge, and it has two main parts.
First, at the level of individual development, education helps a person form the basic abilities for understanding the world, such as understanding language, knowing common sense, following logic and telling right from wrong. These basic cognitive abilities determine whether a person can find a footing in society.
Second, at the level of society, education must build a cognitive foundation shared by everyone, so that society can run stably. Especially since the twentieth century, as the division of labor has grown ever finer, making sure that citizens still share a common understanding with one another beyond their own professional skills has become a top priority.
In 1945 Harvard University published General Education in a Free Society, later widely known as the “Harvard Redbook”. It set out to answer the question of how a civil society could, through basic and even compulsory education, sustain its common cultural foundation, its capacity for reason and its public responsibility.
This kind of education came to be widely recognized as “general education”. The core purpose of general education is not to give learners professional skills, but to pass on a shared structure of knowledge and shared ways of thinking, and so to build a common understanding among the members of society. Without that common understanding, the foundations of civil society would cease to exist. A report on education thus became the educational manifesto of an era.
This point is even clearer and more important in today's age of artificial intelligence: when AI can do most of the work, the meaning of learning will return to cultivating individual cognitive ability and public awareness, rather than rote memorization and the accumulation of skills that machines may replace at any moment.
2. The Cognitive Predicament of the Intelligent Society
Now, in 2026, we stand once again at a crossroads of history. Last time the cause was the fragmentation of knowledge; this time it is artificial intelligence.
Today, AI is becoming everyone's assistant. It helps us search for material, read the literature, write, analyze data, generate plans and make predictions, and it even takes part in scientific discovery and complex decision-making. The coexistence of humans and machines has become an irreversible historical trend.
An even greater change is taking place at the level of society.
Today, people increasingly encounter the world through search engines, recommender systems, social platforms and generative AI. What we see and what we overlook; which information is amplified and which is obscured; which views look more credible; which social goals are turned into computable metrics — all of these processes are shaped ever more deeply by data, algorithms, models and platforms.
We are entering a new form of society in which AI is deeply involved. Here AI is not merely a tool but a kind of infrastructure: it is tightly coupled with the institutions through which society operates, and forms a basic mechanism of that operation. We can call this new form of society the “intelligent society”.
The intelligent society has an enormous effect on the individuals who live in it. This effect is not an anxiety about whether we can use AI tools; it is an effect on the individual mind. AI tools grow ever more powerful and can do more and more work; once AI has done a great deal of work in place of humans, the outsourcing of thought and dependence on AI will grow ever more serious, and the cognitive autonomy of the individual will face an enormous challenge.
Humanity has never faced such a situation before: we spent thousands of years learning to think rationally and more than a century building habits of independent thought, yet now we risk losing our cognitive ground because our tools are too powerful. Anyone may give up thinking because they no longer need to think, and so gradually give up the precious capacity to be a cognitive subject.
What matters most is that this “crisis of cognitive subjecthood” affects not only individuals' ability to think independently but also the foundations on which civil society runs: when large numbers of citizens lose the capacity to be cognitive subjects, the basis for citizens' participation in public affairs will be hollowed out, and public order will become fragile.
How to preserve the cognitive subjecthood of human beings is a fundamental question that all of humanity must face.
3. Cognitive Autonomy: The Mission of General Education in Our Time
The cognitive predicament of the intelligent age gives general education a new mission: on top of its traditional mission of building a shared public cognitive foundation, it must also preserve people's “capacity for cognitive autonomy”. By cognitive autonomy we mean the process by which an individual, in a complex cognitive environment, actively regulates their own belief formation, use of evidence, allocation of trust, limits of judgment and responsibility for action; the capacity for this active regulation is called the capacity for cognitive autonomy.
Cognitive autonomy is not simply a mental defense; it means actively embracing the arrival of a new era. It requires people to explore and use AI actively while keeping active control over their own cognitive processes. We need to know where information comes from, what evidence a conclusion rests on, how far a model can be trusted and where the limits of its abilities lie; we also need to judge which tasks can be delegated, which value choices should be made by humans, and who should bear responsibility for actions that humans and machines carry out together.
For the individual, this is a basic survival skill in the intelligent society; for society, it is a precondition for civil society to function normally.
This kind of education must therefore be general education: it must be compulsory, and everyone must receive it.
From this we establish a basic principle: the development of AI is bringing human society into the intelligent society; the capacity for cognitive autonomy becomes an essential foundation for individual citizens and for social stability in that society; and cultivating the capacity for cognitive autonomy becomes the mission of general education in our time.
4. AI as a Natural Training Ground for the Capacity for Cognitive Autonomy
Cultivating the capacity for cognitive autonomy is the mission of our time, and every subject should cultivate it deliberately. Language and literature education, for example, should train the discernment of sources and the judgment of meaning; mathematics and science education should train awareness of evidence and the ability to reason; history and the social sciences should train the analysis of positions, causal explanation and the judgment of responsibility.
Artificial intelligence, however, plays a unique role that other subjects can hardly replace: it offers a natural and comprehensive training ground for the capacity for cognitive autonomy. This training can proceed along two lines.
The first is understanding knowledge. Learners understand the basic concepts and methods of AI and where its abilities come from; they understand why a model can accomplish certain tasks and also why it makes mistakes; and they understand how data, models, platforms and feedback mechanisms enter social life and shape the spread of information, the allocation of trust and the choice of action.
The second is hands-on practice. Complex AI systems can be simplified into teaching tools that can be analyzed and practiced with. Through analysis and experiments, learners come to understand data, models and algorithms, and gain an intuitive grasp of how problems such as data dependence, uncertainty, overfitting and generative hallucination show up and why they arise.
Learning about AI is therefore a natural setting for cultivating the capacity for cognitive autonomy. AI education that takes cultivating the capacity for cognitive autonomy as its goal is what we call “AI general education”.
5. AI General Education: Definition and Goals
AI general education is a new mode of education that arises from the fusion of general education and AI education.
First, it is a form of general education: it inherits general education's mission of building a shared cognitive foundation and extends that mission to cultivating the capacity for cognitive autonomy. Seen this way, AI general education expands the territory of traditional general education; it is general education's advanced version for the intelligent age.
Second, it is a special kind of AI education. It teaches students the basic knowledge and basic methods of AI, but the ultimate goal of learning is not to master that knowledge and those methods; it is to cultivate the capacity for cognitive autonomy through the learning process. If that goal is achieved, it hardly matters if students forget the specific terms and methods; if it is not, the course has failed even if students' heads are full of algorithms and programming tricks.
In a word: only AI education aimed at “cognitive autonomy” can be called “AI general education”. The term must not be misused.
It should be noted that “cognitive autonomy” is the overarching goal of an AI general education course, but not its only goal. Every course needs an overarching goal that distinguishes it from other courses, and it also needs concrete, staged and achievable goals. AI general education is no different: it should design a multi-level system of goals, with corresponding principles of assessment, according to the actual circumstances of each group of learners. These goals cover knowledge as well as ways of thinking and cognitive abilities. However the goals are designed, though, cognitive autonomy must be the core: if this goal is not achieved, no matter how much else is done, the course cannot qualify as an AI general education course.
Many of the deviations seen in AI general education in practice stem from a blurred core goal, such as tendencies toward the instrumental, the technical, the activity-driven and competition-driven, and the cutting-edge and fragmented, or the tendency to treat ethics as an add-on. Such courses may be valuable “AI courses”, but they are not qualified “AI general education courses”.
6. Conclusion: Building a New Cognitive Foundation for the Intelligent Age
The Harvard Redbook of 1945 sought to establish for the society of its time the mission of general education as building a “shared cognitive foundation”. Today, faced with the arrival of the intelligent society, we too must answer a question of historical significance: now that machines are beginning to take part in human cognition, what should we preserve and develop for the next generation through education?
Our answer is: cognitive autonomy.
When answers come within easy reach, we must still cultivate the ability to ask questions;
When machines express themselves ever more fluently, we must still ask whether the evidence is sufficient;
When algorithms can offer predictions and recommendations, we must still judge whether they deserve our trust;
When humans and machines accomplish complex tasks together, we must still be clear about human value choices and ultimate responsibility.
The more powerful technology becomes, the more education must safeguard human judgment; the better machines become at generating answers, the more people need to keep the ability to ask questions, test evidence, define values and take responsibility.
This is the essence of cognitive autonomy.
The AI general education we are promoting aims to build this capacity for cognitive autonomy in citizens, and so to lay a new cognitive foundation for the intelligent society. This goes far beyond building a course or a set of textbooks: it concerns the stance in which a generation will enter the intelligent society, what a society will rely on to sustain reason, trust and cooperation, and whether humanity, faced with ever more powerful intelligent systems, can remain the master of its own cognition and action.
This is the mission of our time that AI general education carries, and it is also its most profound historical significance.