The second installment of the DiTT Symposium series, “Considering Education in the Age of AI.” We welcomed nine speakers—representing companies developing AI-based educational materials and cram schools that have adopted them—to discuss AI and education.
(My notes ended up being quite long, so I’m posting this in two parts: Part 1 and Part 2.)

　The panelists were Makoto Nakashita, COO of atama plus; Naoki Inaba of Z-Kai Educe; Naonori Endo, CEO of Meits; Naoya Tsujimura, CEO of Weblio; Motoki Kamino, CEO of COMPASS; Hiroaki Yuasa, CEO of UNIVERSCHOOL;and Hideki Kawai of Kawai Juku. I served as the moderator.
　
　First, the individual presentations.
Founded in April 2017, “atama plus” uses AI to analyze each student’s strengths, weaknesses, and learning progress to create “personalized lessons,” thereby minimizing study time.By significantly reducing the time needed to master foundational academic skills, it frees up time to cultivate the “life skills” necessary for the 21st century. Traditional learning methods involve repeatedly reviewing problems where students have struggled. In contrast, atama plus breaks down problems into smaller components to identify the root cause of these “stumbling blocks” and provides appropriate lessons.It has demonstrated significant results, such as an average improvement rate of 50.4% among students who attended the approximately two-week winter intensive course.

Z-Kai, widely known for its correspondence education business, established Z-Kai Educe in 2015 to oversee its classroom-based operations. The program emphasizes “researching, thinking, and expressing ideas independently.”
　Z-kai’s University of Tokyo Individual Tutoring Classroom Pleiades has adopted atama+. While AI-powered learning materials are used to convey information and lesson content, instructors are responsible for helping students set goals and maintain motivation. atama+ visualizes learning history, making it easier for instructors to provide feedback. AI handles the “teaching,” while instructors handle the “coaching.” It’s not a choice between “AI or people,” but rather “both AI and people.”

　Meits has been exploring tutoring methods since 2010. Initially, it used “group” instruction with a 1:12 instructor-to-student ratio, but in 2012, it shifted to “individualized” instruction with a 1:3 ratio.Furthermore, in 2013, we introduced iPads and switched to “individualized” instruction with a ratio of 1:6 to 1:12. By delegating part of the instructor’s role to the iPad, we achieved one-to-many individualized instruction. Based on this expertise, we developed educational applications. Our Eiken app, designed specifically for cram schools, automatically generates a list of areas needing improvement based on answer history and provides guidance on what to study.Teachers focus exclusively on answering questions and monitoring progress, serving as coaches to boost student motivation. As a result, a single instructor can now handle up to 24 students, and the center achieved an 86.7% pass rate for the Eiken Level 3 exam.At the time of development, the app supported only reading and listening sections, but through a partnership with Weblio and the integration of AI, it now also supports speaking and writing. In addition to making pronunciation instruction—which was previously difficult—possible, students have also begun to study more independently.

　COMPASS, the developer of Cubina—the world’s first AI-powered educational material—was founded in Silicon Valley in 2010. The company researched learning efficiency and developed “Nano-Step Learning,” which provides personalized learning tailored to each individual. It transformed traditional analog teaching materials into AI-powered ones. The system provides real-time feedback based on learning history and adjusts difficulty levels accordingly.AI teaches, while teachers foster motivation. With the extra time created by efficient learning, students engage in active learning that leverages cutting-edge technology. By conducting workshops using drones, 3D printers, VR, and other tools, the program sparks students’ desire to learn and transforms their “I want to” into “I can.”Through the use of AI, we aim to realize both adaptive learning and active learning. When introducing AI, it is crucial to clearly define “what we want AI to solve.”

Uni-Ba High School presents a new style of learning that utilizes the latest technology. The introduction of AI-based teaching materials is highly effective for providing each student with an optimal learning experience.Lessons begin with a video introduction, followed by hands-on exercises using Meits and Cubina. Students now spend more than twice as much time solving problems within the same time frame. Thanks to Cubina, test scores rose from 10 to 70 points in just two weeks. As teachers’ roles shifted from teaching to coaching, the amount and quality of student interaction improved.Students now have extra time, allowing for student-centered, inquiry-based active learning classes, such as career education.

Kawai Juku faced challenges such as varying levels of understanding among students and the limitations of providing individualized support within its existing system. As a solution, the school decided to introduce AI-powered learning materials and jointly developed COMPASS and high school math curriculum.
　This has created an environment where students can learn according to their individual styles and paces. In the classroom, a variety of learning styles can be observed, including students studying alone, those learning through discussion, and those teaching one another. The burden on instructors has also been reduced.Since instruction can be tailored based on learning history, more time is available for coaching, and conversations with students have even expanded to include life advice. Students have also begun to proactively study other materials on their own due to the reduced time required for core subjects.
(Part 2 will cover the discussions.)