TOKYO – The power of artificial intelligence is being put to use in Japan for diverse purposes, from helping children at risk of skipping school to teaching English conversation, and even keeping the stories of atomic bomb survivors alive.
In 2023, Toda city in Saitama Prefecture, near Tokyo, launched a pilot program to see if AI could predict the likelihood of truancy and provide children with preemptive support.
The AI was trained on data such as absences and health check records, generating numerical predictions for each student to support teachers’ decisions.
On a computer screen, students’ names appear alongside a color-coded risk level — red, pink, orange, or yellow — indicating the likelihood they would skip classes.
To generate these scores, the AI analyzed personal information such as attendance, tardiness, academic survey results, health checkups, nurse’s office visits, and bullying records, drawing on patterns from students who had previously been absent.
During the trial, the AI flagged 1,193 students as being at risk. Of those, schools were advised to prioritize support for 265. At one elementary school, the AI’s alerts helped identify 10 students for extra help who would have been overlooked otherwise. The trial concluded in fiscal 2023.
Access to the predictions was limited to principals and administrators. Keio University professor Makiko Nakamuro, who sits on the city’s education advisory board, called the effort promising but cautioned against becoming over-reliant on the technology.
“By linking objective data, we have the potential to prevent absenteeism,” she said. “But privacy concerns must be addressed, and parents and students need clear explanations.”
Toda city also issued guidelines banning discrimination or unfair treatment based on AI results.
While the initiative may help in efforts to tackle truancy, it also raises the sensitive issue of allowing technology to rate children by numerical scores. For busy teachers, the system could be useful in flagging students at risk of skipping school who might otherwise go unnoticed.
Some municipalities are also testing AI to identify children vulnerable to abuse or bullying.
But like teachers, AI is far from infallible and can miss warning signs. Experts stress that AI must be supplemented by human judgment, and only after parents and children are given clear explanations.
Several universities are now using AI to act as an English conversation partner. Waseda, Kansai, Chuo, Meiji, and Kyushu universities have adopted the system, along with junior high schools in Gifu.
Developed by Tokyo-based startup Equmenopolis, the AI chats with students for about 20 minutes, analyzing grammar, vocabulary, pronunciation, response time, and even facial expressions. It scores performance on a six-point scale and offers feedback.
In one session, the AI asked: “Do you have any movie recommendations?” A university entrance exam candidate replied in English, and the AI responded naturally with follow-up questions. The student later said they felt less nervous speaking to AI than to a person.
Yoichi Matsuyama, Equmenopolis CEO and a visiting researcher at Waseda University, said the system solves a long-standing classroom challenge.
“It’s difficult for a human teacher to speak with every student,” he explained. “AI allows students to practice at their own level.”
Advocates say the technology could help Japanese students overcome one of their biggest weaknesses in English: speaking confidence.
The Kanagawa prefectural government is using AI to assist memory preservation. In March 2025, the prefecture will trial the “AI Storyteller System” at the Kanagawa Peace Memorial Hall in Yokohama.
The system allows visitors to ask questions and hear recorded responses from atomic bomb survivors. The testimony of 93-year-old Nagasaki survivor Hiroshi Nishioka was cut into about 130 topics. When asked a question, the AI selects and plays back the most relevant recorded answer.
During a demonstration, elementary school students asked, “Wasn’t it scary?” and “What was life like during the war?”
The AI replied with Nishioka’s own words and video clips. One 11-year-old participant said: “I was surprised, it felt like I was talking to a real person.”
Nishioka was 13 when he was exposed to the atomic bombing, about three kilometers from the hypocenter. He hopes the project will keep his memories alive.
“There are fewer and fewer survivors,” he said. “I want to leave an accurate testimony while I still can.”
The Kanagawa government plans to offer the system to schools and make it available at the memorial hall.
The municipal government of Hiroshima, the other city devastated by atomic bombs during the closing days of World War II in 1945, is developing a similar device for its Peace Memorial Museum.
Developers stress that the AI will never invent or alter what the subjects said. It can only replay actual survivor testimony — a safeguard designed to respect the sensitivity of the subject.