Participating in a Gemini for Education Ideathon, I was more surprised by the 'differences in school environments' than by the generative AI itself
On July 11, 2026, I participated in the "Gemini for Education Ideathon 2026" held in Sapporo.
It was a Google for Education-related event where school staff, board of education members, university students, and others gathered to think about how generative AI can solve challenges in the educational field.
However, what remained in my mind after it ended was not the quality of the ideas.
More than the discussion about what generative AI can do, there is a significant difference in the "environment where generative AI can be used" depending on the school.
That fact still weighs on me.
What kind of event is an ideathon?
Some people may not be familiar with the term "ideathon."
It is an event where multiple people share ideas on a specific theme and work in a short time to shape them into concrete solutions. The theme this time was the use of generative AI in the educational field.
On the day, after participants introduced themselves to each other, we first brought up the problems we each feel in our daily work. Lesson preparation, teaching material creation, school administration, information sharing, and responding to students. Depending on the position, the challenges that arise are quite different.
From there, groups chose one challenge and organized it according to a worksheet.
What is the problem?
Why is that problem occurring?
How to solve it using Gemini, etc.
How will the field change as a result of the solution?
Finally, the flow was for each group to present the ideas they had thought of.
In the advance information, categories such as "school administration use," "class use," and "teaching material development" were set, but at the Sapporo venue, it was not a format divided by category, but a format where each person could freely decide on a theme.
It became a place to bring in things that had been prepared in advance
At the venue, too, finished products that had been prepared in advance appeared. They had a level of perfection that would not come out of a few tens of minutes of discussion.
Preparing in advance is not a bad thing. Rather, I think it is a sign of seriousness.
However, I thought it was an ideathon because we think about it with the people gathered there, so I was a little confused. If you bring in things you have made, it becomes closer to a contest or a pitch. When people who want to think on the spot and people who want to show what they have made are in the same room, it becomes difficult for the conversation to mesh.
This is not a complaint to the organizers.
The thing that surprised me the most about participating
At the event, there was an opportunity to hear from people who are already using generative AI in the educational field.
What surprised me there was that in some schools, the use of Gemini was much further ahead than I had imagined.
It was not at the stage where some teachers interested in generative AI were trying it out personally. Chromebooks are distributed to students, they use Google Workspace accounts, and they use the devices daily under the school's management. There was also talk that in environments for educational institutions, children and students under 18 can also use Gemini in a managed form.
Of course, not all schools are in the same state. But at least some schools were already moving from the stage of debating "whether generative AI should be allowed to be used" to the stage of thinking about "how to incorporate it into education within an environment where it can be used safely."
I felt that this difference was quite large.
"Having one knowledgeable teacher" is not enough to keep it going
When it comes to the educational use of generative AI, attention tends to be focused on how to write prompts and convenient ways to use it.
However, participating this time, I felt that there was a problem before the prompts.
Are there devices that students can use? Are school accounts prepared? Are the services that can be used and the services that cannot be used organized? Is the information that can be entered and the information that cannot be entered shared?
If teachers just create individual accounts and try them out at their own discretion, it will not become a school-wide initiative.
Even if there is one teacher who is knowledgeable about generative AI, if that person is transferred or removed from the assignment, the utilization itself will stop.
In schools, the moment a person in charge changes, questions like "What did you do last year?", "Who should I check with in this case?", and "Where are the previous materials?" arise all at once. Even if the materials themselves remain somewhere, it is not clear which one to look at.
The use of generative AI will be the same if left alone. The moment the knowledgeable person is gone, the method itself will disappear.
Introducing it does not mean distributing accounts
When you hear "use Gemini at school," you might imagine preparing an account and using it once in class.
But introducing it to a school in the true sense must be more than that.
Devices, accounts, permissions, information management, lesson design, teacher training, and who to consult when in trouble. Only when that is included can it be used continuously.
Generative AI is convenient. Creating a draft of teaching materials, organizing sentences, and proposing multiple ideas. Its speed is incomparable to working with humans alone.
On the other hand, it can also provide incorrect information, and a sentence that sounds plausible is not necessarily correct.
I think it is necessary to learn how to use it in a managed environment, rather than letting them use it freely or banning it entirely.
The next day, an article about Norway came out
The day after the event, such an article appeared in the Yomiuri Shimbun.
Kari Nessa Nordtun, the Minister of Education of Norway, which is considered an advanced country in digital education, stated regarding Japan's digitalization of education that if excessive emphasis is placed on equipment and teaching materials, there is a risk that major learning problems will occur, and she does not want Japan to repeat the mistakes they made. The country introduced one device per person early on.
In June 2026, Prime Minister Støre announced at a press conference a policy to make generative AI unusable in elementary schools in principle.
In Japan, in the same month of June, a related law was passed to make digital textbooks official textbooks.
At almost the same time that I was surprised that "schools with a well-prepared environment are amazing," the country that had finished preparing the environment had begun to turn back.
However, the saying "the Nordic countries have returned to paper" is quite simplified and widespread in Japan. Regarding Sweden's policy change, there are also points that the actual goal was to guarantee the quality of textbooks and that it was not to exclude digital. It does not lead to the story that Japan should also stop because the Nordic countries have turned back.
Even so, I thought it was quite heavy that a country that had finished preparing the environment was saying that they had made a mistake in how to use it after preparing it.
Nothing starts without an environment. However, what lies beyond the prepared environment is another matter.
Conclusion
No matter how interesting the usage ideas are, if there is no environment where students and teachers can actually use them, the ideas will remain just ideas.
Conversely, in schools where devices, accounts, and management systems are in place, even somewhat rough ideas can be tried out for the time being.
I feel that this gap will become quite large in a few years. It made me feel a bit anxious.
Of course, it is not a simple matter of just rushing to introduce it. This is even more true when looking at the example of Norway. Each school has different circumstances, and caution is required regarding the information handled.
Even so, if we continue in a state of not using it because nothing has been decided, rules will never be created. Try it on a small scale, find the problems, and fix the operations. I believe that the educational use of generative AI can only move forward through that repetition.
It was an event called an ideathon, but for me, rather than a place to take home ideas, it was a place to be shown the 'current location' of generative AI use in educational settings.
Before thinking about what can be done with generative AI, I intend to first check it at my own site.