On November 25, 2025, we held a programming workshop utilizing generative AI at the “Meta Support Campus” operated by the Ehime Prefectural Board of Education. Using Microsoft Copilot, which requires no login, participants created web-based applications written in HTML.
Previously, participants needed to learn languages like HTML and JavaScript or master authoring tools. However, by leveraging generative AI, creating web apps became possible in no time.
For this workshop, we prepared an environment where “HTML could be pasted and run immediately.” Children experienced pasting HTML code generated by AI and seeing it run.
First, as a warm-up, we ran a 10-question quiz about generative AI using Kahoot. This quiz was also created using generative AI (ChatGPT 5.1). Next, to broaden their ideas about what they could or wanted to create, they “played with” (tried out) various programs generated by AI. Then, they each worked on their own projects. Diverse creations emerged, ranging from games to multiplication training tools.
Thunder rumbled loudly that day, startling the children, which inspired one participant to create a “training app to overcome fear of thunder.” Since we held a similar workshop in October, some participants commented, “I had an idea for a project I wanted to make after the last workshop, but I couldn’t remember it.” While it’s unfortunate they couldn’t recall the specific project, it was very heartening to hear they had thought about “projects they wanted to create” after the workshop. This showed they were considering how to apply programming in their daily lives.
We have consistently worked to enable children to use Scratch, OctoStudio, or our original tools like Programmable Battery as “one means of expression” through programming. While fostering “computational thinking” through programming is important, we place even greater value on children developing an interest in computers and programming as tools to realize what they want to create and express.
Using generative AI has made it possible to create works without learning programming itself. Therefore, it remains unclear to what extent programming can cultivate “computational thinking” or “programming-like thinking.” On the other hand, when considering it as one means of expression for children, generative AI can be seen as a powerful tool for realizing their own creative ideas.
When actually using generative AI to create works (web apps), there are times when it works well, and times when it doesn’t work no matter how many times you interact with it. However, by leveraging the interactive nature of generative AI and persisting through trial and error, it feels like it’s becoming easier to gradually create “works that suit your taste.” This is likely because the AI learns individual preferences from the interaction history. In any case, the difference between “when it works well” and “when it doesn’t work well” is strongly felt. This might be an issue with how prompts are formulated—what’s known as prompt engineering.
So, in this situation where things sometimes work and sometimes don’t, what are we learning? In conventional programming, the language used is fixed, and the same code always produces the same result. With generative AI, however, the same prompt doesn’t necessarily yield the same response. What are we learning through these interactions with generative AI?
Constructivism emphasizes the importance of “creating things that spark one’s interest.” Using generative AI allows anyone to greatly expand the range of things they want to create or are interested in making. In this case, “what to create” becomes more important than “how to create it.”
What are we learning—including children—not only when we create what we intended, but also when unexpected creations emerge through the serendipity of generative AI? I hope to explore this slowly through workshops.
Works created by children using generative AI at the workshop, along with pieces I tried making myself, are posted here.