Tokyo Metropolitan schools also fully implement AI education: The Light and Shadow of AI Utilization in Education “The brain becomes lazy and can't escape” — Shocking research results presented by MIT

Table of Contents

1. Introduction: The Current State of AI and Learning

2. Overview of MIT Research and Experimental Protocol

3. Mechanism of "Metacognitive Laziness"

4. Deployment of Generative AI "Tokyo Metropolitan AI" in All Metropolitan Schools

5. Comprehensive Comparison of Benefits and Risks

6. Comments from Japanese and Overseas Experts

7. Curriculum Design Guidelines for the AI Era

8. 10 Actions Parents and Learners Can Take Starting Today

9. Proposals for the Future: Co-evolution of Technology and Humanity

10. Conclusion

1. Introduction: The Current State of AI and Learning

As of 2025, generative AI is permeating educational settings as a "second brain," handling not only translation, summarization, and programming support but also essay and report writing. The push for AI adoption is fueled by the demand for "higher results in less time," a trend accelerated by the shift to online learning during the COVID-19 pandemic. However, there is growing concern about learners outsourcing their thought processes entirely to AI.

2. Overview of MIT Research and Experimental Protocol

The research team at MIT Media Lab divided 54 students from five universities near Boston into three groups and assigned them regular essay writing tasks over four months.

• AI Group: Used prompts with ChatGPT-4o and utilized the generated text almost as is

• Search Engine Group: Gathered information using Google and wrote independently

• Analog Group: Conducted library research and wrote entirely by hand
  
  

Participants wore EEG headsets to measure brain waves during and after tasks. Results showed that the AI group experienced an average 27% decrease in beta (focus) and gamma (integrative thinking) bands, with slow recovery after task completion. 83% could not accurately reproduce what they wrote two weeks later.Nextgov/FCWNature

3. Mechanism of "Metacognitive Laziness"

The study named this phenomenon **metacognitive laziness**. Humans naturally require energy to "keep thinking." When AI takes over the thinking process, the brain suppresses activity as an "optimization," making it difficult to return to its original active state. While this may be efficient in the short term, it could lead to "cognitive muscle deficiency" in the long term.Laptop Mag

4. Deployment of Generative AI "Tokyo Metropolitan AI" in All Metropolitan Schools

In May 2025, the Tokyo Metropolitan Board of Education fully launched a dedicated generative AI platform, "Tokyo Metropolitan AI," available to all 256 schools and approximately 140,000 students and staff. The platform supports GPT-4o-mini and above, ensuring safety through filtering and tenant separation. It is being introduced to support individualized learning and improve school operations.Press Release Distribution No.1｜PR TIMESTokyo Education Portal

4-1. Purpose of Introduction

1. Individual Optimization: Support for exploratory learning and supplementary lessons

2. Reduction of Teacher Workload: Automation of document draft generation and test question creation

3. Data-Driven Lesson Improvement: Optimization of teaching methods through learning data analysis

4-2. Voices from the Field

• High School Teacher A: "Organizing materials for exploratory learning is time-saving, but I'm worried about dependency on copy-pasting."

• Junior High School Student B: "Having AI summarize helps me understand quickly, but I'm worried if I can explain it in my own words."

5. Comprehensive Comparison of Benefits and Risks

6. Comments from Japanese and Overseas Experts

"AI is like car navigation. If you completely lose the ability to read maps, you'll get lost the moment you 'abandon' it."
— Dr. Natalia Kosmyna (MIT)Nextgov/FCW

"AI utilization is inevitable, but teachers need to become 'cognitive weight trainers,' deliberately adding weight in their lessons."
— Dr. Nicola Jones (Nature Correspondent)Nature

"Tokyo Metropolitan AI is the world's largest school tenant. That's why I want to design a schedule with 'rest days' where AI is not used."
— Ryota Sato (Tokyo Metropolitan Government ICT Promotion Officer)Press Release Distribution No.1｜PR TIMES

7. Curriculum Design Guidelines for the AI Era

1. Three-Stage Utilization Model

   • Pre-Stage: Generate rough ideas by hand or orally

   • Mid-Stage: Request AI for summarization and organization, then compare and review

   • Post-Stage: Re-explain and reconstruct without AI to solidify understanding
     
     

2. Weekly "No AI Day"
   Designate a day within the school schedule to completely refrain from using AI, focusing on discussions, experiments, and fieldwork.
   
   

3. Prompt Literacy Education
   Make training in crafting quality questions mandatory. Transform AI from a "black box that provides answers" to an "interactive thinking partner."
   
   

4. EEG Feedback Learning
   Experimentally use EEG devices to visualize one's concentration level and develop a program to enhance metacognition.

8. 10 Actions Parents and Learners Can Take Starting Today

1. Always write a conceptual memo without AI for the first and last 15 minutes of homework

2. Verify three sources of information that support the AI-provided answers

3. Use reading aloud and handwritten summaries to aid memory retention

4. Engage in family discussions to debate the validity of AI answers

5. Make long-form reading from paper books a weekend habit

6. Balance typing and handwriting

7. Challenge yourself with creative outputs using generated images and audio

8. Play a "spot the error game" by intentionally inserting errors into AI responses to develop critical thinking

9. Save learning logs to visualize understanding "before and after AI intervention"

10. Always add credits when using AI-generated content on social media

9. Proposals for the Future: Co-evolution of Technology and Humanity

AI can become the ultimate partner in expanding learning, but depending on how it is used, it can also be a double-edged sword that stifles thought. The large-scale introduction in Tokyo Metropolitan schools serves as a social experiment in "AI-assisted learning," leading the world in public education. MIT's warning suggests the need for us to design a "blueprint that prevents the human brain from slacking."

• To Policymakers: Institutionalize AI usage limits and "brain training" curricula together

• To School Administrators: Incorporate prompt design and metacognitive evaluation into teacher training

• To Parents: Set rules for AI use and offline learning time at home

10. Conclusion

MIT research quantitatively demonstrated the "cost of convenience." While Tokyo Metropolitan AI holds the potential to reduce educational disparities, without mechanisms to prevent metacognitive laziness, it will not lead to true academic improvement. The key isthe design ability to distinguish and reintegrate parts entrusted to AI and parts where humans think through painstakingly.

##HTML