How Should High School Education Change in the Age of AI? Inquiry, Generative AI, and an Education That Grows Culture

Hello, this is Ryuta Hamamoto from TIMEWELL.

In my day-to-day work I sit down with business owners and the people on their shop floors to design how AI gets brought into the work. Lately, one question keeps coming back to me: by the time the students in high school today step out into society, what will the world look like? Generative AI has advanced startlingly fast in just a few years. Much of the work of "producing an answer," such as writing text, drawing pictures or building programs, is now handled by AI at a fairly respectable level.

I feel this shift is quietly shaking the premises of high school education. The ability to reproduce memorised knowledge accurately has long sat at the centre of academic achievement. Yet part of that ability is being taken over by AI. So what should high schools grow from here? The state holds the same concern, and a review of the Courses of Study, Japan's national curriculum standards, has begun to move. In this article, I want to first pin down where that policy stands using primary sources, and then, addressed to teachers on the ground, write out my own proposal. It runs a little long, but I would be glad if you stayed with me to the end. If you are curious about your own distance from AI as a teacher, checking where you stand first with the free AI literacy self-check will make the second half of this piece feel more concrete.

Where high school education stands right now

Let me start by organising how the state is moving. Once you can see this, the debate on the ground stops getting lost.

The Courses of Study are revised roughly once every ten years. Looking toward the next revision, on 25 December 2024 the Minister of Education referred the question "On the Ideal State of the Standards for Curricula in Primary and Secondary Education" to the Central Council for Education¹. A referral is when the minister formally puts a question to a council of experts: "Please debate how education should be designed from here, and give me your answer." Discussion proceeded from that referral, and on 25 September 2025 the Special Subcommittee on Curriculum Planning of the Central Council for Education compiled an "interim summary of issues"². This interim summary is a mid-stage document that provisionally gathers up thirteen rounds of deliberation, on the way toward a formal report and then a revision. Do keep in mind that neither its name nor its content is settled yet.

This interim summary lays out "three directions" that run through the revision debate. The first is to properly implement proactive, interactive and deep learning. The second is to embrace a diverse range of students. The third is to make all of this achievable on the ground rather than a rice cake in a picture. That third point stayed with me. Beyond raising ideals, a sense of realism comes to the front: design it so it does not lean on excessive burden for teachers. On the premise of not adding further to instructional hours, trim the volume of textbooks and create "margin," meaning breathing room, for teachers. Seeing such words lined up in an official document reads, to me, as a sign of alarm about exhaustion on the ground.

The points touching high schools are concrete too. The interim summary sets up a standalone item on making high school curricula more flexible, and it examines loosening the credit system and creating special curricula for students with a marked talent in a particular field. It also lays out a policy of fundamentally raising information-use ability in light of the development of generative AI, and positions that ability as "a foundation supporting inquiry-based learning." One thing to watch here is the name of the policy. The core programme for high schools is the "High School DX Acceleration Programme," commonly called DX High School. This is a different thing from the second phase of the "GIGA School Initiative," which puts a device in every student's hands (commonly called NEXT GIGA); the two get confused, but their aims differ. DX High School supports the cost of preparing the environment for high schools that take on a curriculum emphasising information and mathematics, together with practical inquiry that crosses the arts-and-sciences divide. Funded by a supplementary budget of 10 billion yen in FY2023, roughly 1,010 schools were selected in FY2024, with support of up to 10 million yen per school³. I take it as a symbol that the state has begun to put its back into developing digital talent in high schools.

Inquiry and generative AI are not in opposition

On the ground, I still sense a lingering air of talking about generative AI and inquiry-based learning as if they were opposites. The worry is that "if we let AI do the writing, students will stop thinking." That concern is natural, but I think the two can complement each other.

First, let me sort out the terms. Inquiry-based learning is often called PBL. It stands for project-based learning, a way of learning in which students set their own theme, research it, form a hypothesis, test it and present their findings, all as one connected project. In high schools, the "Period for Inquiry-Based Cross-Disciplinary Study" was fully introduced from FY2022 and has already begun to take root on the ground. Generative AI, on the other hand, is the umbrella term for AI that automatically produces text, images, programs and the like. These two, in fact, go well together.

The grounds lie in MEXT's own stance. The guideline on using generative AI (Ver. 2.0), which the ministry released in December 2024, states plainly that it neither uniformly bans nor mandates the use of generative AI in schools⁴. At its base sits the idea of being "human-centred." On that footing, it says this: precisely because it is a digital age where information is easy to obtain, we are called on to understand the meaning of learning, to grasp what each piece of information means, and to question the essence of a problem. It also writes that it is important for each person to become able to handle generative AI as a tool and to let their own talents bloom. At the same time, it honestly admits that it is extremely hard to fully prevent AI from producing mistaken content, the phenomenon known as hallucination.

Stand on this position and the way to combine inquiry and AI comes into view. Sparring over the theme, a partner to widen a hypothesis, a doorway into background reading, polishing prose. Use AI in these scenes and students can turn their time toward what really deserves thought. The crucial thing is not to make the AI's output the answer as it stands. Doubt what comes out, confirm it against a primary source, and restate it in your own words. That back-and-forth of verification is exactly the ability inquiry is meant to build. Rather than banning AI to force thinking, grow the "ability to doubt AI" while letting students use it. I think this fits the reality ahead better. For teachers who want to push students all the way into advanced making, the AI development guide for high school students should also be a concrete doorway.

In an age when AI produces answers, what work do people carry?

Let me step one further and think about the line between what AI can do and what only people can do. Once that line is set, what high schools should train comes into focus.

Today's AI is good at receiving an instruction and generating something. In software development, new ways of working that make use of this trait are spreading. There is a phrase, for example, "specification-driven development." It is a way of proceeding in which a person carefully defines in words the specification of what they want to build, meaning what should work and how, and hands most of the implementation to AI along that specification. What gets asked here is less the ability to write code one line at a time and more the ability to put into words, precisely, what should be built. The same change is happening in research. The phrase "AI for Science" points to efforts that treat AI as a tool to accelerate scientific research itself. Finding patterns in vast experimental data, narrowing down candidate materials: AI has begun to take on searches of a scale people alone cannot handle. I lay this movement out in an introductory guide to AI for Science as well.

Such technologies hold value in proportion to how far "adoption," meaning actual settling into real work, proceeds on the ground. What matters here is the accumulation of technique. A new tool does not take root if you touch it once and stop. Try it, fail, record the knack, and put it to use next time. That build-up becomes an organisation's strength. In high schools too, rather than ending generative AI as a one-off event, use it repeatedly inside inquiry, clubs and daily study, and have students share with each other the ways of using it that worked. That plain, unglamorous accumulation is, I think, what pays off after graduation. Incidentally, in the training we design for companies, we aim not to end at learning but to carry people to the point of actually moving something usable with their own hands within about two months. This is not an external statistic; it is only the target we set when we build a course. But the idea of moving your hands to make it stick applies to schools too, I feel.

So when AI takes on the generating, what remains for people? I think there are three things. One is the ability to frame a question. AI is good at answering a question, but it will not decide for you what ought to be asked in the first place. The second is value judgment. When several answers come out, which one to regard as good rests on that person's values and responsibility. The third is lived, bodily experience: touching things with your own hands, meeting people, the memory of failing. The interim summary too states plainly that it fully takes into account the importance of the bodily nature and lived experience unique to human beings. These three are, I think, abilities that high school of all places can thicken. If there are students interested in AI-driven research at university, showing them the further scenery of something like AI-driven development and science at university can also motivate a choice of path.

The ability to frame a "question" grows from cultivation and cultural capital

Let me take a slightly theoretical turn here. Where do good questions and sound value judgments come from? I think they grow from the cultivation a person has built up, the thickness of their foundation, so to speak.

A concept useful for explaining this is "cultural capital." It is an idea proposed by the French sociologist Pierre Bourdieu, referring to a person's assets other than money, such as knowledge, ways of speaking, cultivation and aesthetic sense. Bourdieu divided it into three. Things embodied as bearing, language and taste; things with a physical form, such as the books, paintings and instruments in a home; and things institutionally recognised, such as academic credentials and qualifications. The amount of this cultural capital is, in fact, related to academic achievement and to the path one takes, he argued. A famous example is the correlation between the number of books in a home and academic achievement.

This is not an abstract theory. In the detailed reference material attached to this referral, there was a piece of data that stayed with me⁵. If you take an elementary-school class of 35 as an example, the picture is that children of diverse backgrounds sit mixed together in the same room: roughly 12.5 children with few books at home and a tendency toward lower achievement, 4.1 with a tendency toward non-attendance, 1.0 who speak little Japanese at home, 0.8 with an unusual talent, and so on. This is of course a correlation, not a simple cause. Even so, the gap in the cultural capital of the home one is born and raised in can become a gap at the starting line. How schools face this reality is what is being asked. The "embracing of diversity" the interim summary holds up, and its move away from an answer-first mindset and the pressure to conform, connect at the root to this, I read.

That is exactly why, paradoxically, education in the age of AI should move in the direction of thickening cultivation. The lower the cost of generating an answer falls, the more the scarcity rises of people who can frame a good question and judge which of several answers to choose. What supports that questioning and judgment is cultural capital and cultivation. The STEAM education MEXT promotes can be read in this context too⁶. STEAM adds an A to STEM, which is science, technology, engineering and mathematics. That A does not point only to art; it is used in a broad sense that includes culture, life, economics, law, politics and ethics. Multiply the power of the sciences by the cultivation of the humanities and social sciences, and solve real-world problems. This design philosophy overlaps cleanly with laying the foundation for an age that handles AI as a tool. How a school turns such "learning centred on culture and cultivation" into a structure is a theme we would like to think through together as well. Our work for schools and educational institutions is gathered on WARP for Schools.

Japan's strength lies in a long, continuous culture

Finally, let me write the proposal I most want to convey. I would like high school education in the age of AI to be an education that grows the culture of this country called Japan.

Why culture? AI can be used the same way all over the world. Throw the same question at the same model and you get a similar answer back whatever country you are in. If so, competitiveness from here will be decided, I think, on that shared ground of how you use AI, by what a country or a region can multiply into it that is uniquely its own. And Japan has an inherent asset worth multiplying in: a long, continuous culture.

What symbolises that continuity is the Imperial House. Counting from the accession of Emperor Jimmu, held to be the first emperor, in the reckoning known as koki, the Japanese Imperial House is said to have a history of more than two thousand six hundred years; the year 2026 falls in the 2,680s by that count. Because koki is a count based on tradition, not all of it can be asserted as historical fact. Even so, within what can be traced as history, the Imperial House is known to have a continuous history of close to two thousand years, and Guinness World Records has recognised it as the oldest continuing monarchy in the world. That a single lineage has lasted this long is itself a fact of Japan's continuity with few parallels anywhere.

I want to be careful with how I handle numbers, but beyond the Imperial House there are many other verifiable proofs of continuity. Kongo Gumi, which handles temple and shrine construction, was founded in the year 578 and has carried its craft down for more than 1,400 years, known as one of the oldest companies in the world⁷. Horyu-ji temple in Nara is held to have been built around the early seventh century, in the fifteenth year of Suiko (607), and is a World Cultural Heritage site as one of the oldest surviving wooden structures in the world. The Tale of Genji, written in the early eleventh century, is one of the oldest works of long-form narrative literature anywhere. The Ise Shrine's regular rebuilding, by reconstructing the shrine buildings every twenty years, has carried its techniques on to the next generation without letting them lapse. These all show that Japan holds a long, continuous accumulation of culture and craft that is rare even by world standards.

What I want to say is that letting the next generation hold this continuity as their own context is the role of education. Kongo Gumi lasted 1,400 years because it kept handing its techniques from person to person, with the body involved. The Ise rebuilding is the same: the twenty-year cycle carries within it the wisdom of having the craft handed on by actually moving one's hands. This overlaps almost exactly with the "lived, bodily experience" and the "steady accumulation of technique" I described earlier. In an age when AI takes on the reproduction of knowledge, what people should inherit is precisely this tangible culture and craft.

What to do concretely? I do not think you need to overthink it. Invite the region's traditional industries and craftspeople into the classroom. Make local history and nearby shrines and temples a theme for inquiry. Read the classics, not only for exams but connected to your own questions now. Have generative AI look things up, confirm them with your own feet, and retell them in your own words. In learning like that, students come to know what strand of culture they stand within, and from there they build the foundation for their questions and their values. Young people in Japan, holding AI as a tool shared with the world, multiply into it the cultural capital that is uniquely Japan's, and create new value. That is the future I picture. If you would like to stand such learning up in a school as a structure, or to start by talking through teachers' own use of AI, please talk to our WARP team. Specialists who carried DX and data strategy at major companies walk alongside you month by month, designing it together in a form that puts no unreasonable strain on the ground.

To sum up

It ran long, so let me organise the key points.

• The state is moving toward the next Courses of Study. The referral of December 2024 and the interim summary of issues of September 2025 hold up three directions, deep learning, embracing diversity and achievability on the ground, along with "margin" for teachers
• Generative AI and inquiry are not in opposition. MEXT's guideline holds up human-centred use and neither uniformly bans nor mandates it. Growing the "ability to doubt AI" while letting students use it is the realistic path
• The more AI takes on generating answers, the more the value rises of the ability to frame a question, to judge value, and of lived, bodily experience. These are abilities high school of all places can thicken
• What underpins them is cultivation and cultural capital. The cheaper answers grow, the more the scarcity of people who can frame a good question rises
• Japan holds a long, continuous accumulation of culture and craft, as seen in Kongo Gumi, Horyu-ji and the Ise rebuilding. An education that hands this culture on to the next generation is, I believe, exactly what pays off in the age of AI

Technology is a tool, not an end. For what purpose, and with what values, do we use that tool? What decides that will, from here too, be people. That is exactly why what the three years of high school should grow is a person's foundation. The student in front of you, handling AI while creating something new with pride in their own country's culture. I would be glad to build that future together with teachers on the ground.

References and primary sources

Footnotes

1. Referral to the Central Council for Education, "On the Ideal State of the Standards for Curricula in Primary and Secondary Education," 25 December 2024 (MEXT). https://www.mext.go.jp/b_menu/shingi/chukyo/chukyo3/101/index.html ↩

2. Special Subcommittee on Curriculum Planning, "Interim Summary of Issues," 25 September 2025 (MEXT). https://www.mext.go.jp/content/20251020-mxt_kyoiku01-000045486_06.pdf ↩

3. "High School DX Acceleration Programme (DX High School)" (MEXT). https://www.mext.go.jp/a_menu/shotou/shinkou/shinko/mext_02974.html ↩

4. "Guideline on the Use of Generative AI in Primary and Secondary Education (Ver. 2.0)," 26 December 2024 (MEXT, Elementary and Secondary Education Bureau). https://www.mext.go.jp/content/20241226-mxt_shuukyo02-000030823_001.pdf ↩

5. Points (detailed version) of "On the Ideal State of the Standards for Curricula in Primary and Secondary Education (Referral)," 25 December 2024 (MEXT). https://www.mext.go.jp/content/20250327-mxt_kyoiku01-000039494_2.pdf ↩

6. On promoting STEAM education and other cross-disciplinary learning (MEXT). https://www.mext.go.jp/a_menu/shotou/new-cs/mext_01592.html ↩

7. Kongo Gumi official site (the world's oldest company, founded in the year 578). https://www.kongogumi.co.jp/ ↩