When AI can do everything, what can education still do?

This weekend I listened to a podcast episode about AI and education that I really liked. Before I introduce it, though, something from many years ago flashed into my mind. It was the summer after Year 4 of primary school, and I was staying at my aunt's house. One day in the park, my cousins and I were piling up stones in the sand when an old man crouched down and started talking with us. A lot of what he said left me wide-eyed at that age. One thing stayed. A song called Mice Love Rice was everywhere that year, and everyone loved singing the line "I love you, I love you, the way mice love rice." He said: what mice feel for rice is not love, it is stealing food. Real love should not be about grabbing and consuming. Later, two young men cycled past and called out "Professor X!". That was the first time I learned that the same thing can be seen, and questioned, from a completely different angle. It planted a very small seed in me.

The podcast I heard this weekend was from SV101. The guest was Jun Liu, founder of 7EDU and Leadways School. She has worked in education for more than twenty years, from China to the United States, and in the end, unhappy with the education she found in Silicon Valley, she started her own school. Her very first sentence made me stop: in the age of AI, what matters most is not solving problems but finding them. Hearing that, I suddenly understood. This was what the old man had been sharing with me all those years ago. A question can be redefined.

Finding the problem

We are trained from childhood to solve problems. But Jun Liu says this is out of date in the age of AI. AI solves problems faster than people do. What is genuinely scarce is: what deserves to be solved? Why does this matter? What is wrong with the current solution?

She gave the example of lunch. At school, students vote on what most needs improving about lunch. She also takes students to the supermarket to spot problems and think about how to fix them, then give the feedback directly to the manager. At first I took this as a classroom activity. Then I realised it is training agency.

What is agency? It is the belief that "I can see a problem, name it, and push for change." Not waiting for someone else to solve it, not complaining and leaving it there, but "maybe I can do something about this." That goes deeper than learning to analyse a problem. Analysis is a skill; agency is a belief. Yet for many of us, even what counts as a problem has always been defined by someone else. The boss says this is the problem, so we solve it. The exam says this is the key point, so we memorise it.

The trouble with systematic education

The podcast questioned something many people avoid touching: does traditional, systematic education really help a child's potential? Many creative people are not products of the system at all. They succeeded because they stepped outside its habits of thought.

Traditional education works like an assembly line. The goal is for every child to meet a certain "quality standard": maths up to standard, English up to standard, PE up to standard. What comes off the line is a standard part.

The podcast was not dismissing foundations. The maths curriculum at Leadways draws on competition-level thinking, and the humanities teaching is solid too. But the point is this: knowledge and skills are the foundation, not the purpose. The real world does not need standard parts. And the biggest problem with systematic education is not what it teaches, but the hidden message it sends: life has standard answers.

Stay in that system long enough and you start assuming, without noticing, that there must be a correct path, a standard way of doing things. In the real world that habit becomes painful, because the real world is full of problems with no standard answer. Often it is not even clear what the problem is. If we are trained from childhood to wait for a problem to be assigned and then solve it, then even with all the knowledge and skills in hand, we may not dare to ask: is this a question that matters to me?

Jun Liu says the AI era changes by the minute, and what children need is not "learn the system first, apply it later" but continuous learning and constant adapting. To me, adapting is not only learning new things. It is daring to question old systems and old authorities. That takes a kind of psychological safety: I am allowed not to know, allowed to make mistakes, allowed to challenge both the question and the answer. Systematic education does exactly the opposite. It rewards being right, punishes being wrong, and does not encourage questioning.

This reminded me of something else. Over the years I have met many exceptionally capable and intelligent people, and their emotional lives are rarely easy either.

The logic of meritocracy may earn plenty of rewards early in life, and it tricks us into believing that reason and intelligence can solve almost everything. But at the more complicated stages of life, facing relationships, loss, illness, death, that method starts to fail. What hurts more is this: when you find you have already done everything you were "supposed" to do, the good university, the good job, even a good relationship, and the distress is still there, and it cannot be "solved," the torment is all the greater.

This may be the most hidden cost of systematic education: it teaches us how to succeed, but not how to be.

On values

Jun Liu says Leadways puts particular weight on values education at primary age. Through games, children choose between and weigh different values: justice, kindness, honesty. They judge, scenario by scenario, whether an action fits a value, until the values become their own.

When a child does something wrong, the school does not say "you broke the rules." It asks, "which of your values did this go against?" She explains the difference. "You broke the rules" is external control; right and wrong are defined by an outside standard. "This went against your values" is internal drive; right and wrong are defined by your own value system.

Many people think self-discipline means forcing yourself to do things you do not want to do. But self-discipline is really this: your behaviour is consistent with your values. When you genuinely hold a value, keeping to it requires no "control." Someone who truly prizes honesty finds lying painful in itself. They do not need to stop themselves from lying.

The point of values education is not to tell children "these are the correct values," but to teach them how to weigh values against one another. Because the dilemmas of the real world are rarely a choice between good and evil. They are conflicts between one good and another.

When AI enters education, what can we still teach?

The podcast's view is clear: AI is the greatest tool for personalised education there has ever been. But it is a teaching assistant, not the teacher.

Leadways uses AI everywhere it helps: student tracking systems, personalised practice on Khan Academy, ChatGPT for summarising and extending. AI takes on a great deal of the repetitive work, which gives teachers more time to notice how each child is really doing. But the school is equally clear: the emotional connection between teacher and student is something AI cannot do.

If education were only the transfer of knowledge, AI could indeed replace teachers, and do it better. But the heart of education is not transferring knowledge. It is sparking curiosity, forming values, and building a sense of connection. The future may look like "AI gives the lesson, the teacher guides": AI handles the knowledge, and the teacher inspires, empathises, connects. That actually asks more of teachers, not less. A teacher can no longer be just a porter of knowledge. They also have to be the one who lights the lamp.

Speaking of that sense of connection, I want to return to the beginning. I do not know whether, in an environment that runs on efficiency, KPIs, and rational trade-offs, a scene like the one twenty years ago could still happen: a university professor seeing a few girls playing with stones, and choosing to crouch down and talk with them for a whole afternoon. Nor do I know whether, inside timetables and work schedules packed to the edge, there is still an idle afternoon that lets you wander into a park for no reason at all, and, in a child's life, accidentally become the "seed" moment they will treasure.

Greatness, like goodness, does not come from planning and calculation.