Hello, this is Hamamoto from TIMEWELL.
"There are 2.2 billion people in the world who cannot connect to the internet. 750 million still have no electricity. Under these conditions, AI risks becoming an enabler of inequality." With this single line, Boral Batangel, founder and CEO of AI Academia Asia, set the tone of the entire session and sharpened every conversation that followed[^1].
The "AI x Education for Well-being in the Intelligent Age" session at SusHi Tech Tokyo 2026 brought together University of Tokyo President Teruo Fujii and AI Academia Asia CEO Batangel, whose organization is rolling out AI-powered education across Mongolia, Azerbaijan, Tajikistan, and other countries[^1][^3]. As someone who runs TIMEWELL with the mission of "democratizing the right to challenge," I have long believed that education is the most decisive battleground for democratization. If the benefits of AI accumulate in the hands of a few, society's overall well-being will inevitably erode. Conversely, if we can build an era in which education lets everyone wield AI, the future becomes one of the most exciting prospects imaginable. This session reinforced that conviction.
In this article, I unpack the core arguments raised during the dialogue and translate them into practical takeaways for Japanese educational institutions, companies, and local governments — all framed through the "democratizing the right to challenge" lens. For a fuller picture of what Tokyo is staking out as Asia's innovation hub, I also recommend reading my keynote report and the piece on smart city evolution from the same SusHi Tech Tokyo 2026.
Setting the Stage — SusHi Tech as a Forum for the Future of Education
SusHi Tech Tokyo 2026, held April 27–29 at Tokyo Big Sight, is one of Asia's largest global innovation conferences. While the program centers on three pillars — startups, cities, and technology — several sessions tackled "education in the AI era" head-on.
The conversation around AI education is not just about reforming schools. It probes a more fundamental question: as AI becomes social infrastructure, how should humans learn and how should we hone our judgment? The dialogue between the president of Japan's flagship university and an entrepreneur from an emerging economy approached that question from sharply different angles. The room was packed with educators, policymakers, and corporate HR leaders, and you could feel just how seriously the audience is taking AI-era talent development.
What struck me most was the quality of the audience questions. Teachers asked the practical question of "how do we engage with students who hand in homework solved by ChatGPT?" Corporate HR managers raised strategic concerns like "where should we direct our reskilling budgets?" The session never drifted into abstraction; it stayed anchored to the realities of the field, and that grounded energy gave the entire dialogue its texture.
Sobering Numbers — The Architecture of AI Inequality
The figures Batangel opened with carried real weight.
| Metric | Figure |
|---|---|
| People without internet access | Approximately 2.2 billion |
| People without electricity access | Approximately 750 million |
| Major AI-developing nations | A duopoly of the U.S. (Silicon Valley) and China (Beijing) |
AI development is concentrated in the Silicon Valley–Beijing duopoly, and the rest of the world is structurally dependent on infrastructure built elsewhere[^1]. Conversations about "AI boosting productivity" implicitly assume access to AI in the first place — yet roughly a third of the world has not even reached that starting line.
Mongolia's numbers stood in sharp contrast. With a population of 3.5 million, the country has 7 million smartphones — two devices per person on average. Internet penetration sits at 95%. And 75% of the population is under the age of 37[^3]. AI Academia Asia has trained 10,000 teachers in this same country, building an environment in which roughly 20,000 children in remote regions now receive education[^3].
A nation of 3.5 million with 7 million smartphones and a population that is 75% under 37 — the inverse of Japan's demographic shape. When the population pyramid differs this dramatically, the very design philosophy of education policy has to shift. We tend to treat "Japan's education system" as the universal default when debating policy, yet the idea that young countries like Mongolia are actually the experimental frontier of AI education struck me as genuinely fresh.
There is a simple structural truth at play: "the more severe a country's challenges, the stronger its motivation to adopt AI." Countries like Japan, where the existing education system still functions reasonably well, paradoxically tend to be conservative toward new technology. In Mongolia, the urgency of "the old methods can no longer keep up" is what drives bold investment in AI.
Looking for AI training and consulting?
Learn about WARP training programs and consulting services in our materials.
Batangel's Provocation — "AI Sovereignty"
The single most thought-provoking concept of the session was "AI Sovereignty." Batangel's framing was sharp[^1]:
"For a small country like Mongolia, depending 100% on English-based large language models means linguistic and cultural subordination. The vast majority of global AI systems are trained on English data, and non-English speakers are marginalized at the system-design stage."
This is not a technology argument — it is a civilizational one. In an era when AI becomes social infrastructure, whether AI works in your own language is bound up with national independence itself. Mongolia is now investing in the development of small-scale language models trained on Mongolian.
This is far from a distant problem for Japan. Major LLMs such as GPT, Gemini, and Claude have steadily improved their Japanese-language support, but their cores are still trained on English and Anglophone cultural data. Japanese-specific honorific systems, deeply context-dependent communication patterns, and industry-specific terminology — unless Japan develops AI capable of handling these natively, Japanese culture risks being gradually marginalized inside AI systems.
A concrete example: the Japanese word "sumimasen" can mean apology, gratitude, getting someone's attention, or expressing humility, depending entirely on context. An English-based model that simply maps it to "I'm sorry" cannot accurately handle the nuanced communication that defines Japanese society. As AI agents start to genuinely embed themselves in customer service and internal communication from 2026 onward, that gap will become a non-trivial factor in business quality. "AI sovereignty" is less a national-strategy issue like trade controls and more a deeply practical challenge of defending the meaning of your own language.
President Fujii's Response — A "Layered Strategy"
President Fujii's reply was calm and intellectually grounded. He argued that the realistic path forward is "developing technology that builds culture- and language-specific adaptation layers on top of English-based large models"[^1][^2].
In other words, you do not need to develop the entire foundational large model in your own country. By layering local adaptation layers on top, you can secure linguistic and cultural sovereignty. It is a pragmatic approach that balances the global with the locally specific.
This "layered strategy" maps almost perfectly onto Japan's national AI strategy. We do not need to build every foundation model in-house. The realistic winning move is for Japanese startups and academia to jointly build adaptation layers specialized in the Japanese language, Japanese culture, and Japanese industry knowledge. Japan-grown AI players like Sakana AI, Preferred Networks, and ELYZA are already implementing this layered strategy. Borrow the foundation, then compete on the layers above — that is a healthy strategic position for the entire Japanese startup ecosystem.
The same idea applies directly to small and mid-sized businesses and local governments. Building your own foundation model is unrealistic, but investing in a "business adaptation layer" specialized in your own workflows, customer knowledge, and internal vocabulary delivers a strong return. RAG (retrieval-augmented generation) and fine-tuning are essentially the SMB-scale version of this same layered strategy.
Implementation at the University of Tokyo — The Matsuo Lab AI Course
The scale of the AI education that President Fujii showcased was striking. Professor Yutaka Matsuo's AI courses reportedly enroll 10,000 to 20,000 students per year[^4]. That is the largest AI education program at any university in Japan.
The president's stance was unambiguous: "A society without AI is no longer avoidable. Rather than a precautionary approach, what matters is trial and error to understand AI's possibilities and capabilities." Universities, in his view, should actively build environments in which students can use AI safely and creatively.
Shifting from "education that prevents use" to "education that enables mastery" — the fact that Japan's most prestigious university is leading this transition carries enormous weight. While many regional universities and high schools remain stuck in "ChatGPT bans," the position the University of Tokyo is taking should serve as a role model for Japanese education as a whole.
Equally important is the fact that Matsuo Lab's courses are open across faculties and majors. The design crosses the humanities-sciences divide and frames AI as part of the liberal arts. When thinking about AI-era education, the deeper insight is that we must reposition AI from "a specialized skill for computer science students" to "a social literacy for everyone." That repositioning is the real essence of university redesign.
A New Faculty After 70 Years — "Tokyo College of Design"
The biggest surprise of the session was President Fujii's announcement of the "Tokyo College of Design," a brand-new faculty — the first new one at the University of Tokyo in 70 years[^2].
"Design" here does not refer narrowly to aesthetic design. It encompasses "every system that shapes future society — social systems, AI learning systems, healthcare systems, international rule-making, climate-change responses."
The design choices are bold:
- Fully English-language instruction on a four-term calendar (aligned with global standards)
- 1.5–2 years of mandatory off-campus activities (with 2.5–3 years on campus)
- Fully residential during the first year (for integrating multinational students)
- BA (4 years) + Master's (2 years) = 6-year integrated program
This is a fully globally oriented faculty design. Batangel underscored the quality of UTokyo's education when he noted that "more than 50% of the AI Academy team has received Japanese engineering education"[^1]. Talented young Mongolians studying at UTokyo's Design College and returning home to expand AI education — that kind of cycle is exactly what will strengthen Asia's AI-education ecosystem.
We should not overlook the symbolic weight of "the first in 70 years." Japan's flagship university is breaking a 70-year silence to launch a new faculty. This is more than an organizational reshuffle — it is a signal that Japan's higher-education system is entering its next phase. The message I took home was: it is too early to give up on the idea that "Japanese universities never change."
Batangel's Field Wisdom — "Start from the Fundamentals"
Batangel's curriculum-design philosophy was deeply practical[^3]:
"In an age where new AI tools appear daily, the fundamentals of algorithms, engineering, and design matter more than any specific advanced technique."
This observation is critically important. Many educational institutions teach surface-level skills — "the latest LLM use cases," "prompt engineering" — but the philosophy that a rock-solid foundation lets you adapt whenever new tools emerge is the right long-term educational investment.
We have the same debate at TIMEWELL when designing training programs. Do we teach "how to use the latest AI tools," or do we teach "the fundamentals of problem discovery, validation, and decision-making"? We always choose the latter. Tools change every few years; the fundamentals last twenty.
Concretely, what really moves the needle in corporate AI training is not the time spent teaching ChatGPT mechanics, but the time spent forcing employees to ask themselves "where are the bottlenecks in our work?" and "what is the essential nature of those bottlenecks?" Once the fundamentals are solid, the optimal move is selectable regardless of the tool. Without that foundation, layering on tool training simply means redoing the curriculum every time a new AI emerges.
President Fujii's "Lifelong Learning Platform" Vision
In response, President Fujii emphasized the necessity of "lifelong learning"[^2]. Four years of university education cannot keep pace with the speed of technological evolution. Universities should function as continuous learning platforms for working professionals, he argued.
For Japanese universities, this is a major shift. Historically, universities have been "places that teach students aged 18 to 22." But in the AI era, the standard becomes professionals in their 30s, 40s, and 50s engaging continuously with universities. The University of Tokyo signaling its intent to lead this shift is, I think, deeply significant.
Most Japanese companies have operated on a model of new-graduate hiring → in-house education → lifetime employment. But that structure is breaking down. We are entering an era where individuals design their own careers and continuously acquire necessary skills through universities and specialized institutions. "The boundary between universities and companies is dissolving" — that, in my view, is the dominant education trend of the next 10–20 years.
For SMBs in regional Japan, I want to emphasize one thing: you do not have to teach everything in-house. The mainstream design going forward is to outsource continuous learning programs offered by universities, local governments, private schools, and consulting firms — and embed them into employees' career paths. Insisting on building all training in-house will leave you unable to keep up with the pace of technology.
The Importance of "EQ" in AI Education
A sharp question came from a student during the session, asking about the role of psychological education (EQ) in the AI era.
Batangel's answer was outstanding. AI Academia runs an "EQ and AI" class for children aged 9–12[^3]. The idea is to develop technical and emotional intelligence simultaneously. Precisely because technology becomes more advanced, uniquely human capabilities — emotional resilience, relationship-building, and judgment — become more valuable.
This is something we feel daily at TIMEWELL as well. The more AI automates work, the more value flows to deep human dialogue, the subtleties of emotion, and trusted relationships. EQ education is not a "luxury good" in the AI era — it is a necessity.
The need for EQ education becomes especially acute for executives and managers. In an age where AI presents decision options, humans still make the final call. The quality of the decision-maker's emotions, intuition, and ethics determines the direction of the entire organization. "Can you handle your own emotions before using AI?" — that is on track to become the most important theme in management training.
Personal Reflection — "Democratizing the Right to Challenge" and the Democratization of Education
As I listened to this session, I kept overlaying it onto TIMEWELL's mission. Our vision of "creating the world's No.1 infrastructure for taking on challenges" is fundamentally about opening the right to challenge to everyone.
Batangel's work is precisely the democratization of educational opportunity. Twenty thousand children in Mongolia's remote regions are now learning through AI Academia's programs[^3]. Knowledge access transcends geographic and economic constraints. To me, that is the most upstream layer of the "challenge infrastructure."
Whether someone wants to launch an AI startup or build a new business, none of it begins unless a critical mass of people in society "understands AI." Education is the most upstream form of democratization.
Mongolia x Japan Collaboration Possibilities
What this session made concrete for me was the potential for Mongolia x Japan collaboration in education.
| Country | Strengths |
|---|---|
| Mongolia | 75% youth population, high AI-receptiveness, 95% internet penetration, proven track record in mother-tongue AI education |
| Japan | Expertise in aging-society responses, growing internationalization needs at educational institutions, the new UTokyo Design College platform |
Mongolia brings youth and openness to change; Japan brings educational quality and institutional rigor. Bidirectional knowledge circulation is genuinely possible.
A model in which talented young Mongolians study at UTokyo, and in which AI Academia's education programs are adapted for employee training at regional Japanese governments and SMBs — this kind of "small country x Japan" talent-circulation model could become a key to solving Asia's AI-talent shortage as a whole.
"Nomadic Culture x AI" — A One-of-a-Kind Combination
Some of the Mongolia-specific examples Batangel shared were fascinating. Among them: drone-based livestock management and AI-driven nomadic-life support[^3]. These are use cases that no other country could possibly produce — they are uniquely Mongolian.
This is a textbook example of creating unique value through "culture x AI." Japan can apply the exact same logic to mass-produce its own "Japanese culture x AI" use cases. Tea ceremony x AI, sushi craftsmanship x AI, shrine rituals x AI, anime x AI, agriculture x AI, traditional crafts x AI — at these intersections, products that no global competitor can replicate become possible.
When local governments and SMBs adopt the mindset of multiplying their "local cultural assets" by AI, they can produce services that no big company in Tokyo or overseas could ever build. That is another dimension of democratizing the right to challenge. "Don't imitate the leading regions — win on your own context" becomes a realistic option in the AI era.
Wrapping Up — AI-Era Education Is "Judgment x Curiosity x Fundamentals"
Here are the three elements of AI-era education I took away from this session:
- Judgment — the ability to make decisions in ambiguous conditions
- Curiosity — the intrinsic motivation to keep absorbing new technologies and cultures
- Fundamentals — the foundational principles of algorithms, engineering, and design
Cultivating these three requires a trinity of universities x industry x international cooperation. When players from different layers — UTokyo's Design College, AI Academia Asia, and private-sector enablers like TIMEWELL — connect, AI-era education infrastructure starts to actually function.
A line from Batangel still lingers with me: "Whether AI becomes a tool that widens inequality or a tool that achieves equality depends entirely on how we design education at this very moment"[^1].
For everyone who challenges and everyone who learns — universal access to AI. Building the infrastructure for that is work I want to keep doing, from Japan to the world. The image of UTokyo's president and a Mongolia-born entrepreneur sitting side by side at SusHi Tech Tokyo 2026, having this dialogue, will, I believe, remain a symbolic scene of Asia's AI education future.
For more context from the same SusHi Tech Tokyo 2026, my keynote report (Governor Koike x Prime Minister Takaichi) lays out Tokyo's overall city strategy, and smart cities evolving with citizens walks through real-world AI implementations on the urban-infrastructure side. Looking at education, cities, and national strategy in three dimensions makes it much clearer where Japanese companies should place their own bets.
If you are wrestling with how to design "the mechanisms that develop people for the AI era" — internal AI training, reskilling programs, regional talent-development initiatives — TIMEWELL's AI consulting service WARP offers tailored consultations. You can start with a 30-minute online consultation.
References
[^1]: YouTube. "AI x Education for Well-being in the Intelligent Age." https://www.youtube.com/watch?v=ti3lBO_eg0U [^2]: University of Tokyo Official Website. https://www.u-tokyo.ac.jp/ [^3]: AI Academia Asia. https://www.ai-academy.asia/ [^4]: Matsuo Laboratory, University of Tokyo. https://weblab.t.u-tokyo.ac.jp/