I am converting some of my long academic articles into short blog posts using AI. Here is another one.
Since the arrival of ChatGPT, discussions about artificial intelligence in education have exploded. Teachers are being shown how AI can generate lesson plans, provide feedback, create quizzes, tutor students, and personalize instruction. Schools are debating how students should use AI and how teachers can incorporate it into classrooms.
But there is a problem with much of this discussion.
We are trying to imagine the future of AI while keeping the traditional school almost completely unchanged.
The curriculum remains the same. Students are still grouped by age. Knowledge is divided into subjects. Teachers teach classes. Students complete assignments. Standardized tests measure achievement. AI is simply added to this existing structure.
This structure has been called the “grammar of schooling”—the deeply embedded assumptions and organizational arrangements that make schools look remarkably similar despite decades of reform.
AI gives us an opportunity to question that grammar rather than simply make it more efficient.
Personalization Should Mean Developing Unique Greatness
One of the most common claims about AI is that it can enable personalized learning. But most personalized learning still assumes that every student should reach the same destination.
Students may move at different speeds or receive different explanations, but they are still expected to master the same curriculum and achieve the same outcomes.
That is not genuine personalization.
A radically different approach begins with the recognition that children are already different. They have different strengths, interests, personalities, motivations, experiences, and abilities. Instead of using education to correct these differences and bring everyone toward the same standard, we could help each student develop what makes them uniquely capable.
The goal of personalized education should therefore be to help each student become uniquely great, not equally average.
This matters even more in the age of AI. Routine and repetitive abilities—whether physical or intellectual—are increasingly susceptible to automation. What becomes more valuable is the ability to do something unusually well: to bring distinctive knowledge, creativity, judgment, passion, and expertise to problems that matter.
From Solving Assigned Problems to Finding Problems
This also requires us to rethink project-based learning.
Traditional schooling is largely built around known problems. Teachers present questions; students are expected to find the correct answers. Even many project-based learning activities begin with a problem chosen by the teacher.
But life does not work that way.
Before people can solve meaningful problems, they must first find problems worth solving.
Students therefore need opportunities to notice needs, identify possibilities, ask questions, and decide where their particular strengths might create value. Three questions can guide this process:
Why this problem? Why you? Why now?
Once students identify meaningful problems, they can use their own abilities, collaborate with others, and work with AI to create solutions. Learning then becomes part of action rather than preparation for action.
Instead of learning something today because it might become useful someday, students learn because they need knowledge now to accomplish something meaningful.
Learning becomes doing, and doing becomes learning.
From Competition to Interdependence
This kind of education also changes students’ relationships with one another.
Traditional schooling places students in competition. Everyone studies largely the same things and is then ranked according to performance.
But when students develop different strengths and pursue different problems, difference becomes valuable.
A student discovers not only, “Here is what I can do well,” but also, “Here is something another person can do that I cannot.”
That realization creates genuine collaboration. Students contribute their strengths to other people’s projects while depending on others whose strengths complement their own.
AI can expand this network even further. Students can work with peers, experts, online communities, and intelligent systems. Learning becomes an exercise in human interdependence, rather than individual competition.
What Would Schools Look Like?
Once we take these ideas seriously, many familiar features of schooling become open questions.
Do all students need the same predetermined curriculum? Perhaps governments and schools should specify only a limited common foundation, leaving much more time for students to pursue their own interests and strengths.
Do students need to be grouped permanently by age? If learning is organized around interests, competency, and problems, students of different ages could learn together.
Do teachers need to spend most of their time delivering content? Teachers could increasingly become mentors, coaches, advisers, and designers of learning environments—helping students discover strengths, refine problems, overcome difficulties, and make good use of AI.
And do we still need standardized assessment? If students follow increasingly personalized learning journeys, assessment should document their growth, accomplishments, contributions, and ability to solve meaningful problems rather than simply compare everyone against the same test.
AI Is an Invitation to Reimagine Education
AI is extraordinarily powerful, but its educational impact depends on what we choose to do with it.
We can use AI to help teachers prepare traditional lessons faster. We can use it to help students complete traditional assignments more efficiently. We can build better tutoring systems to help students master the traditional curriculum.
All of these uses may be helpful.
But they leave the basic structure of schooling untouched.
The much greater opportunity is to use AI as a reason—and a resource—to rethink education itself.
Instead of asking “How can AI improve schooling?”, perhaps we should ask:
“If we were designing education today, with AI already available, would we invent schools the way they currently exist?”
The answer may lead us somewhere far more interesting than simply adding another technology to the classroom.
It may lead us beyond the grammar of schooling altogether.
Based on the original article:
Zhao, Y. (2025). Artificial intelligence and education: End the grammar of schooling. ECNU Review of Education, 8(1). https://doi.org/10.1177/20965311241265124
The article was first published online on July 23, 2024.
