In a new
article in ECNU Review of Education, Professor Yong Zhao posits that we’ve
been asking the wrong question—namely, whether AI is good or bad for education.
A familiar scenario that leads us to ask that question looks something like
this: a student uses generative AI to draft an essay, and a teacher discovers
it. What tends to unfold is a sequence of concern, policy memo, maybe a new
detection tool, and/or a ban on devices or AI.
In his reliably provocative manner, Zhao suggests that a far better question is whether the work we’ve been assigning in schools was/is worth protecting in the first place. Consider the weight of such a question in leadership team meetings, faculty meetings, admission offices, or boardrooms. When we consider what schools have long taught and rewarded—summarizing a text, producing a five-paragraph essay, answering a knowledge-based question, or organizing an argument (to name but a few), we must simultaneously consider to what degree such work constitutes something routine, something with heuristics that can be programmed reliably into algorithms. Naturally, it is easy to point to such ‘routine-ness’ as fundamental to human cognition, yet we’re also faced with the reality that the routine-ness results in it being able to be replicated algorithmically rather easily. This is an uncomfortable position for us to contemplate. Why is that?
The comfortable position is that the problem is elsewhere. What we hear typically is that the problem is technology and its misuse. The problem is not our curriculum, not our assessment, and not our belief of what counts as rigor (see Semmelweis Reflex, a cognitive bias). Zhao exposes the underbelly of that comfort. As he asserts, cheating, weakened effort, cognitive offloading, and policy backlash are not separate crises requiring separate fixes; rather, all are symptoms of the underlying reality that schools are still teaching and assessing in ways that can now be easily replicated by machines.
Much is demanded of us. If Zhao’s diagnosis is even partially correct, our standard response is inadequate by design. Tighter honor codes, better plagiarism detection, more cautious rollout plans, clearer guidance on acceptable use, and the like, are merely improvements to the existing model. Zhao distinguishes carefully between “improvement change,” which makes current practice more efficient without touching its assumptions, and “transformational change,” an approach that centers on whether the assumptions themselves still hold. He stresses that much of what schools currently call “innovation” is actually improvement change in disguise.
As to why schools don’t simply transform, Zhao offers an assessment that he shared with delegates at our April 2026 Leadership Conference: schools function as a kind of peace treaty, rather than being held in place primarily by pedagogical conviction and philosophy. They pursue a settlement among parties with different, sometimes competing interests — parents wanting recognizable markers of success, colleges wanting familiar credentials, boards wanting stability and comparability, communities wanting continuity. What this means is that things such as grouping by age, academic transcripts, grade point averages, and class rank (as examples) are not true pedagogical choices. By contrast, they are the terms of an agreement that permits otherwise divergent interests to coexist. When a school attempts to change a good deal of it at once, it is no longer a question of improving a school; it is akin to renegotiating a peace treaty.
Zhao submits that waiting for the whole system to move together should not be seen as an exercise in patience. To the contrary, it is a way of ensuring that nothing moves. His proposed alternative is what he calls the “courageous minority”: a relatively small number of teachers, students, leaders, and community partners who redesign learning inside the spaces they actually control — a classroom, an advisory, a capstone program, a school-within-a-school — without waiting for permission to remake everything at once. Here, Zhao draws on panarchy theory, the idea that small, adaptive subsystems can influence larger, more resistant ones over time, rather than the reverse. Large systems are resilient precisely because they resist direct transformation. Smaller ones can experiment, fail cheaply, adjust, and eventually offer the larger system something worth noticing.
Independent schools should consider Zhao’s work to be an invitation. We already possess something most large systems do not, which is the governance latitude and the mission clarity to protect a genuine experiment without asking permission from some distant authority. A capstone program built around real community problems. An assessment model that values process, revision, and judgment over polished, machine-assistable output. A faculty team given room to redesign a course around the questions AI cannot answer for a student, such as: What is worth pursuing? What do we owe each other? What have I decided, and why? These are not exotic proposals for independent schools.
Zhao’s challenge does not require schools to declare AI either villain or superhero. AI has not lowered the bar. What it has done, with uncomfortable clarity, is to show us to what degree what we teach and assess requires a courageous minority to redesign and revalue what learning is. Though we don’t need to upend things overnight (remember, mandates are unhelpful in this regard), the longer we wait to engage and deploy a courageous minority, the more opportunity we miss to make transformational change.
For reference: Yong Zhao, "Obsolete Schooling and the Courageous Minority: Rethinking Educational Change in the Age of Artificial Intelligence," ECNU Review of Education 9(3), 2026.
In his reliably provocative manner, Zhao suggests that a far better question is whether the work we’ve been assigning in schools was/is worth protecting in the first place. Consider the weight of such a question in leadership team meetings, faculty meetings, admission offices, or boardrooms. When we consider what schools have long taught and rewarded—summarizing a text, producing a five-paragraph essay, answering a knowledge-based question, or organizing an argument (to name but a few), we must simultaneously consider to what degree such work constitutes something routine, something with heuristics that can be programmed reliably into algorithms. Naturally, it is easy to point to such ‘routine-ness’ as fundamental to human cognition, yet we’re also faced with the reality that the routine-ness results in it being able to be replicated algorithmically rather easily. This is an uncomfortable position for us to contemplate. Why is that?
The comfortable position is that the problem is elsewhere. What we hear typically is that the problem is technology and its misuse. The problem is not our curriculum, not our assessment, and not our belief of what counts as rigor (see Semmelweis Reflex, a cognitive bias). Zhao exposes the underbelly of that comfort. As he asserts, cheating, weakened effort, cognitive offloading, and policy backlash are not separate crises requiring separate fixes; rather, all are symptoms of the underlying reality that schools are still teaching and assessing in ways that can now be easily replicated by machines.
Much is demanded of us. If Zhao’s diagnosis is even partially correct, our standard response is inadequate by design. Tighter honor codes, better plagiarism detection, more cautious rollout plans, clearer guidance on acceptable use, and the like, are merely improvements to the existing model. Zhao distinguishes carefully between “improvement change,” which makes current practice more efficient without touching its assumptions, and “transformational change,” an approach that centers on whether the assumptions themselves still hold. He stresses that much of what schools currently call “innovation” is actually improvement change in disguise.
As to why schools don’t simply transform, Zhao offers an assessment that he shared with delegates at our April 2026 Leadership Conference: schools function as a kind of peace treaty, rather than being held in place primarily by pedagogical conviction and philosophy. They pursue a settlement among parties with different, sometimes competing interests — parents wanting recognizable markers of success, colleges wanting familiar credentials, boards wanting stability and comparability, communities wanting continuity. What this means is that things such as grouping by age, academic transcripts, grade point averages, and class rank (as examples) are not true pedagogical choices. By contrast, they are the terms of an agreement that permits otherwise divergent interests to coexist. When a school attempts to change a good deal of it at once, it is no longer a question of improving a school; it is akin to renegotiating a peace treaty.
Zhao submits that waiting for the whole system to move together should not be seen as an exercise in patience. To the contrary, it is a way of ensuring that nothing moves. His proposed alternative is what he calls the “courageous minority”: a relatively small number of teachers, students, leaders, and community partners who redesign learning inside the spaces they actually control — a classroom, an advisory, a capstone program, a school-within-a-school — without waiting for permission to remake everything at once. Here, Zhao draws on panarchy theory, the idea that small, adaptive subsystems can influence larger, more resistant ones over time, rather than the reverse. Large systems are resilient precisely because they resist direct transformation. Smaller ones can experiment, fail cheaply, adjust, and eventually offer the larger system something worth noticing.
Independent schools should consider Zhao’s work to be an invitation. We already possess something most large systems do not, which is the governance latitude and the mission clarity to protect a genuine experiment without asking permission from some distant authority. A capstone program built around real community problems. An assessment model that values process, revision, and judgment over polished, machine-assistable output. A faculty team given room to redesign a course around the questions AI cannot answer for a student, such as: What is worth pursuing? What do we owe each other? What have I decided, and why? These are not exotic proposals for independent schools.
Zhao’s challenge does not require schools to declare AI either villain or superhero. AI has not lowered the bar. What it has done, with uncomfortable clarity, is to show us to what degree what we teach and assess requires a courageous minority to redesign and revalue what learning is. Though we don’t need to upend things overnight (remember, mandates are unhelpful in this regard), the longer we wait to engage and deploy a courageous minority, the more opportunity we miss to make transformational change.
For reference: Yong Zhao, "Obsolete Schooling and the Courageous Minority: Rethinking Educational Change in the Age of Artificial Intelligence," ECNU Review of Education 9(3), 2026.
Copyright © 2026