Reading the MEXT White Paper 2026 (Part 2): How AI Can Reshape Teachers' Work and Learning

Hello, this is Ryuta Hamamoto from TIMEWELL.

In Part 1, I used the FY2025 MEXT White Paper, published on 24 July 2026 and commonly called the "MEXT White Paper 2026," as a starting point to organise the issues that schools face today¹. Teachers' long hours, the shortage of applicants, the rise in bullying and non-attendance, and the wobble in the foundations of academic achievement and learning itself. The further you read, the more each of them feels like a problem of structure that no single teacher's effort can solve.

In Part 2, I want to think about what we can do against those issues. In my day-to-day work at TIMEWELL I design how companies bring AI into their work, and at the same time I am involved with the education field as an Associate Professor (Project) at Shinshu University. Moving back and forth between these two positions, what has come into view is this: AI need not be a tool that takes work away from teachers. It can be a tool that gives teachers back the time to do what they wanted to do in the first place. Working from the facts the white paper presents, let me write out the countermeasures one at a time. It runs a little long, but I would be glad if you stayed with me. 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 rest of this piece feel more concrete. And if you have not yet read Part 1, starting from Reading the MEXT White Paper 2026 (Part 1) will make the thread easier to follow.

First, the problem of "time" and "who is the lead" that the white paper laid out

Before getting into the countermeasures, let me briefly confirm the facts that also serve as the ground for Part 2. If this is vague, the debate over what to do just spins its wheels.

Feature 1 of this white paper is titled "Toward schools where teachers can face each child one by one: realising both ease of work and worth in work," and it states plainly that teacher work-style reform and better treatment are the most important issue for education administration². In surveys of working conditions by boards of education, the average monthly overtime in-school hours come to about 31 hours for elementary-school teachers, about 40 hours for lower-secondary teachers, and about 33 hours for upper-secondary teachers². Through recent efforts, the share of teachers who stay within 45 hours a month has risen across all school types, and the share exceeding 80 hours a month has fallen, yet the tendency toward long hours still remains. On the hiring side too, the multiplier for teacher recruitment exams fell to a record-low 2.0 times for elementary schools², and a "teacher shortage," in which the necessary instructors cannot be secured, has been pointed out.

In response to this situation, the state also moved policy in a big way. The revision to the Act on Special Measures Concerning Teachers' Salaries and related laws, enacted in June 2025, provides for raising the teaching-adjustment allowance in stages from 4 percent of the monthly salary to 10 percent by FY2030². As a full-fledged improvement in treatment concerning teachers' pay, this is the first since 1974, and the white paper describes it as a salary improvement roughly fifty years in the making². Alongside this, a revision to the standards law for compulsory education fixed a lowering of the class-size standard in lower-secondary schools from 40 to 35². This is a review about forty years in the making.

What I want to stress here is that these are countermeasures at the "entrance," namely treatment and headcount. Providing money and people is of course important, but that alone does not automatically increase the very time teachers have to face children. How to create time, and, with the time created, who to make the lead of learning. Everything from here on is where AI comes in, I think.

Countermeasure one: lighten administrative work with AI and win back time to face students

The first countermeasure is the most pressing problem, "time." A teacher's day is not made of lessons alone. Writing newsletters, letters to guardians, tallying surveys, meeting materials, organising interview records: this clerical work takes up more time than one might imagine. This routine clerical work is precisely the area where generative AI can most easily show its strength.

What matters here is the separation of work that the white paper's guideline sets out. Within its work-style-reform guideline, the state sorted the work that schools and teachers have carried into three categories². Work that should be handled outside the school; work in which people other than teachers should actively take part; and work that is the teacher's work but whose burden can be lightened. Much of the work in this third category can have its groundwork taken on by AI. For example, have AI produce a draft of an event newsletter and let the teacher do only the fact-checking and the final adjustment. Have AI organise the free-text responses of a survey and let the teacher read the trends and concentrate on judgment. Uses like these greatly reduce the burden of building from zero.

The white paper itself backs this direction. In the chapter on education DX, it sets a goal of introducing next-generation school-administration systems to every municipality, and states that it is advancing a project to accelerate school-administration DX³. Building the foundation that smooths document creation and information sharing digitally is already moving as national policy. On generative AI as well, it revised the guideline on use at the primary and secondary stage to Ver. 2.0 in December 2024, laying out the idea of human-centred use⁴.

What I feel on the ground is that the biggest reason people cannot take this step is a vicious circle: "there is no time, so there is no room to learn the very tool that would create time." That is exactly why the first step can be small. Try handing just one piece of administrative work, say the grade newsletter, to AI. Share the ways of using it that worked in the staff room. This steady accumulation eventually lightens the burden of the whole school. In the development programmes we design for companies too, we do not try to change all work at once; we take a point where the effort pays off and make it stick. This is not an external statistic but the aim of our own course, yet the idea of starting small and spreading out applies to schools just as it is, I feel. If a school would like to design AI use as a structure from the ground up, a form of month-by-month companionship by specialists, such as WARP, also becomes an option.

Countermeasure two: from teaching to coaching

Once time is created, what to do with it. This is the second countermeasure. I think it means gradually moving the weight of the teacher's role from someone who teaches toward someone who supports.

Much of today's lesson centres on the teacher standing at the front explaining knowledge and students receiving it. If we call this teaching, then from here I see it as raising the proportion of coaching, in which the teacher walks alongside the process by which students themselves frame questions, research and think. Explaining basic knowledge, and part of the repeated practice tuned to each student's level of mastery, are things AI has become able to take on individually. If so, teachers are freed from explaining the same content to everyone at once, and can spend more time facing the child who is stuck one on one, or drawing out motivation.

This direction overlaps with the debate on the next Courses of Study that the white paper introduces. Citing the interim summary of issues compiled in September 2025 following the referral of December 2024, the white paper states that it places the implementation of proactive, interactive and deep learning as one of its cores⁵. The idea is to return the lead of learning to students. In the background there is a fact that cannot be ignored. Japan's children kept a world-top level in all three fields in the international academic survey PISA 2022, and maintained a high standard in TIMSS 2023 as well⁶. On the other hand, a survey tracking change over time within Japan shows a decline in scores in four of five subjects, and the share of children who feel they are good at maths and science, and their study time at home, are also falling, the white paper points out as an issue⁶. High in the amount of knowledge, yet the will and confidence to learn are thinning. What fills this gap is coaching that faces each student one by one, I think.

What I do not want to be misunderstood here is that this is not a story of handing education wholesale to AI. Even if AI supports individually optimised practice, seeing why that child is stuck and what words let them move forward is work that only people will be able to do from here too. AI does not replace the teacher; it creates the margin for the teacher to see each student. In that margin, devote yourself to coaching. This, I think, is the substance of the shift from teaching to coaching.

Countermeasure three: make AI-native ability and the power to build the foundation of learning

The third is about the very abilities students acquire. The coming generation grows up in a world where AI is a given. I call this generation "AI natives," people for whom AI has been close at hand from birth. For them, what matters is neither fearing AI nor blindly trusting it, but the ability to handle it skilfully as a tool.

The white paper too shows this direction clearly. In the education-DX chapter, it holds up a policy of fundamentally raising information-use ability in light of the development of generative AI, and states that it is examining this within the debate on the next Courses of Study⁵. Concretely, adding a domain that handles information to the Period for Integrated Study in elementary school, creating a subject in lower-secondary school that handles information and technology, and enriching the Information subject in high school have come up as points at issue⁵. That said, these are still at the interim-summary stage, and neither their names nor their content is settled. This is a point I want to hold accurately. The environment that forms the foundation is also being prepared: the one-device-per-student that the GIGA School Initiative handed out is being renewed over several years from FY2024, and the share of schools that have secured the necessary network speed rose to 63.9 percent as of December 2025³.

What I want to add is that the substance of the ability to handle AI skilfully will change from here. In the world of software, a way of proceeding is spreading in which a person carefully defines in words the specification of what they want to build and hands most of the implementation to AI. What gets asked here is less the technique of writing code one line at a time and more the ability to put into words what should be built. Widening the view further, AI is ceasing to be an existence only inside the screen. Tied to robots and devices, the realm of so-called physical AI, which moves in the real world, has been growing too. This is not a fact written in the white paper but only my own reading of the wider current of technology, yet in an age of AI with a body like this, I think the value of the experience of actually moving your hands to make something, testing it and fixing it rises rather than falls. Redesigning inquiry and making in the classroom by tying them to AI, this new tool. That, I believe, becomes the foundation-building for the AI-native generation. If there are students who step into full-fledged AI-driven learning and research at university, showing them the further scenery of something like AI reform in university education can also motivate a choice of path.

Countermeasure four: connect people from the private sector with the world outside school

The fourth is the countermeasure of opening the school. To distribute the burden on teachers and connect learning to society, borrowing the power of people outside the school is indispensable, I think.

Here too, the white paper's three-way sorting of work is a clue. The state clarified the work in which people other than teachers should actively take part, and pointed toward greater use of support staff and specialist personnel². In the plan to improve the fixed number of staff as well, a policy of strengthening the placement of clerical staff and of teachers who handle student guidance is included². In other words, the state too has begun to review the very premise of running a school with teachers alone.

For people who have built practical skills and expertise in the private sector to join this current holds great meaning, I think. People who have been involved in data, digital, management or manufacturing at companies take part in lessons and inquiry as part-time or external personnel. Or the path for people with experience as working adults to take up a teaching post is widened. When such connections increase, not only can the burden on teachers be distributed, but students' chances to touch the real workings of society also increase. Someone who knows the field speaks of how the term "AI-driven" is actually used in real work. That becomes learning that textbooks alone cannot convey. Since how far social-recruitment and special-licence systems are treated concretely in this white paper differs in expression by edition, it is surest to confirm the details in MEXT's official materials. What I am stating here is the author's view that the connection points between schools and the private sector should be increased.

I myself stand on both the field of companies and the lectern of a university, and I feel there is still a large distance between these two worlds. The taken-for-granted of handling AI skilfully in companies does not readily reach schools. Conversely, the wisdom of learning that education has built up is not put to use in company development. Bridging this distance is exactly the work that will hold value from here, I think.

Countermeasure five: protect culture and history as an occasion to learn properly

The fifth is a countermeasure of a slightly different character. The talk of AI has continued, but here let me write about what I want to cherish precisely because it is the age of AI.

To begin, let me get the facts accurate. As far as this MEXT White Paper 2026 shows, there is no statement that fields such as tradition, culture, history or morality are being reduced⁷. On the contrary, the current Courses of Study are organised in a direction of expansion: deepening understanding of tradition and culture through Japan's language culture, cultural properties, annual events, traditional instruments, martial arts and washoku cuisine, and so on⁷. Morality too has been fully implemented as a special subject since FY2018 in elementary school and FY2019 in lower-secondary school, which amounts to expansion through becoming a formal subject⁷. In the previous revision, subjects such as Modern and Contemporary History and Civics were newly created in high school⁷. As for the next Courses of Study, deliberation continues at the interim-summary stage of September 2025, and how the instructional hours of each subject will turn out is not yet settled⁵. So the claim that culture and history are being cut cannot, at least from the current white paper, be drawn. I want to make this line clear.

On top of that, from here is the author's view. I think that precisely in the age of AI, we should protect the occasions to learn culture and history properly, and indeed thicken them. The reason is simple. AI can be used the same way all over the world. Throw the same question at it and a similar answer comes back whatever the country. If so, competitiveness from here will be decided, on that shared ground called AI, by what a country or a region can multiply into it that is uniquely its own. Japan has an inherent asset worth multiplying in. It is the culture handed down over a long time. Our country holds an accumulation of culture and craft continuous to a degree rare even by world standards. Handing this thickness on so that the next generation can hold it as their own context. That, I think, is a value only people can inherit in an age when AI takes on the reproduction of knowledge.

What can be done concretely is not difficult. Make the region's history and traditional industries a theme for inquiry. Read the classics, not only for exams but connected to your own questions now. Have generative AI do the preliminary research, confirm it with your own feet, and retell it in your own words. Use the tool of AI and learning like this becomes, if anything, easier to deepen. Keep the facts accurate, and on that footing choose your values for yourself. Growing this stance leads to an education that protects and grows the identity of the country, I think.

Countermeasure six: with AI, "amplify" culture, work-styles and jobs

Finally, let me write the one idea that runs through all the countermeasures so far. I capture it in the word "amplify," meaning to magnify, to make resonate more widely. This is not a term from the white paper; I use it as a word for my own way of thinking.

The debate over AI tends, if you are not careful, to turn into a story of subtraction, "people's jobs will be taken away." Yet the countermeasures we have looked at so far are, every one of them, not subtraction. Handing administrative work to AI is not to take away the teacher's time but to increase the time to face children. Having AI take on individually optimised practice is not to take away the joy of teaching but to create the margin to see each student. Looking up culture and history with AI is not to replace tradition with a machine but to deepen learning and hand it on to the next generation. AI can be a tool that amplifies, rather than replaces, the workings of people. This is the idea I have put into the word amplify.

The exhaustion of teachers we saw in Part 1 is, to repeat, not any individual's lack of effort but a problem of structure. That is exactly why it should be solved not by appeals to spirit but by a mechanism that creates time. And at the centre of that mechanism, place AI. Turn the time created toward coaching, toward inquiry, toward learning that hands culture on. Create margin with technology, and in that margin concentrate on what only people can do. When this cycle begins to turn, the school will surely become a richer place for teachers and children alike.

Of course, for one school, let alone one teacher, to shoulder this is unreasonable. That is exactly why I hope you will borrow outside power well. If you would like to stand up AI use in a school as a structure, or to start by consulting on how teachers face 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 of Part 2.

• Feature 1 of the MEXT White Paper 2026 treats teacher work-style reform and better treatment as the most important issue, and shows policy moves such as working conditions, the fall in recruitment multipliers, the revision of the salary-measures law and the shift to 35-student classes in lower-secondary school. These are entrance-level countermeasures; a mechanism to create time is needed separately
• The starting point of the countermeasures is AI use in administrative work. From the work in the white paper's three categories whose burden can be lightened, hand the groundwork to generative AI, ride the current of next-generation school-administration DX, and win back time to face students
• Use the time created for the shift from teaching to coaching. Fill the gap between the proactive, interactive and deep learning the interim summary holds up and the issue of a thinning will to learn, through companionship for each student
• As the foundation for the AI-native generation, value the fundamental raising of information-use ability, the ability to put a specification into words, and the experience of making with your hands. That said, the next Courses of Study are at the interim-summary stage and not yet settled
• By opening the school and connecting with people from the private sector, distribute the burden and connect learning to society
• On culture and history, the white paper carries no statement of reduction, and the current direction is, if anything, one of expansion. On top of that, the author's view is that, precisely because it is the age of AI, they should be protected and grown as occasions to learn properly
• Running through all of this is the idea of amplifying the workings of people with AI. Create margin with technology and concentrate on what only people can do. This cycle enriches the school

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. I would be glad to receive the issues the MEXT White Paper 2026 laid out not as pessimism but as a starting point for design, and to build the countermeasures together with teachers on the ground.

References and primary sources

Footnotes

1. "FY2025 MEXT White Paper" (MEXT, published 24 July 2026). https://www.mext.go.jp/b_menu/hakusho/2026/mext_00004.html ↩

2. FY2025 MEXT White Paper, Feature 1, "Toward schools where teachers can face each child one by one: realising both ease of work and worth in work" (MEXT). https://www.mext.go.jp/content/20260724-mxt_soseisk01-000050974_5.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

3. FY2025 MEXT White Paper, Part 2, Chapter 9, "Promoting education DX and strengthening information dissemination using ICT" (MEXT). https://www.mext.go.jp/content/20260724-mxt_soseisk01-000050974_15.pdf ↩ ↩²

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

5. 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 ↩ ↩² ↩³ ↩⁴

6. FY2025 MEXT White Paper, Part 2, Chapter 2 (descriptions of academic achievement and learning conditions, Figure 2-2-1 and others) (MEXT). https://www.mext.go.jp/content/20260724-mxt_soseisk01-000050974_8.pdf ↩ ↩²

7. FY2025 MEXT White Paper, Part 2, Chapter 2, Section 1 (descriptions of education on tradition and culture, and moral education) (MEXT). https://www.mext.go.jp/content/20260724-mxt_soseisk01-000050974_8.pdf ↩ ↩² ↩³ ↩⁴