AI Reduced My Lesson Preparation from '2 Hours to 20 Minutes'

I still remember the sigh I let out.

It was Friday evening. The clock in the staff room had passed 7:00 PM. My weekly plan was still only half-filled. Writing the objectives for next week's Japanese, math, social studies, and independent living classes, thinking through the flow, and considering support strategies.

I thought, 'Is it going to be like this every week?'

Back then, it was normal for lesson preparation to take two to three hours a week. Now, there are days when I can finish the core parts in 20 to 30 minutes. It has changed dramatically. However, there were quite a few failures before it changed. I want to write about them honestly.

The first time I used AI to create a weekly plan was about a year ago.

I thought, 'This might be useful,' and tried it immediately. I had ChatGPT generate a Japanese lesson plan tailored to the children in my special needs class.

The moment it appeared, I was a little excited.

It certainly produced text that looked the part. It included the unit objectives, the flow, and even considerations. But as I read on, a feeling of 'Wait a minute?' gradually crept in.

This isn't for my kids.

The support strategies written were somewhat textbook-like. 'Prepare visual cues,' 'Confirm repeatedly.' They weren't wrong. But the faces of the children in front of me didn't come to mind.

For example, for a child who has difficulty reading text, I individually adjust the font, line spacing, and even the background color. For a child who has trouble anticipating activities, I use a timer and procedure cards together. Those 'sweet spots' unique to each child were completely absent from the AI's output. That's only natural. I'm the only one who knows them.

I thought I could use it as is. That was my first mistake.

The week after I pasted the AI's output directly into my weekly plan, the class became disjointed.

In one scene, I had anticipated the flow, thinking, 'The children should do this here,' but in reality, they gave a completely different reaction. It's common for lessons not to go as planned, but at that time, there was an uncomfortable feeling that 'this plan isn't something I actually thought of in the first place.'

After class, I thought while blankly erasing the blackboard.

I am teaching a lesson that the AI thought of.

That disconnect continued for a while.

The turning point was changing how I used it.

I stopped asking it to 'give me the answer.' I switched to 'creating a draft to think about together.'

For example, when making a weekly plan, I first write the objective myself. I write a sentence in my own words, such as 'I want this child to learn this this week.' I give that to ChatGPT and ask, 'If I were to think about a 50-minute flow with this objective, what kind of activities could I consider?'

I read what it produces and select what to keep, deciding 'this is unusable' or 'this is interesting'.

When I do that, I strangely feel like the lesson becomes my own.

This shift in how I use it aligns with the guidelines from the Ministry of Education, Culture, Sports, Science and Technology (MEXT).

In December 2024, MEXT published the 'Interim Guidelines on the Use of Generative AI in Elementary and Secondary Education (Ver. 2.0)' (*1). In it, teachers using generative AI for 'lesson preparation and teaching material research' is positioned as a useful application. On the other hand, the issues with 'using AI output as-is' are also organized, and the teacher's independent judgment and involvement are treated as a prerequisite.

It is not about 'prohibition' or 'anything goes,' but rather a question of 'how to use it.'

I think that is exactly how it feels on the ground as well. Whether you use what the AI produces as-is, or use it through your own judgment. That difference is reflected in how the lesson feels.

From the 2025 fiscal year onwards, each local government is also moving forward with the development of guidelines regarding AI use. The number of tools has also increased. ChatGPT, Gemini, Copilot. As the options have expanded, it has become necessary to decide for myself 'what to use and how.' I feel that having that axis will become even more important from now on.

Currently, I use AI for the 'objectives' and 'brainstorming teaching methods' that form the core of my weekly plans.

The time it takes has changed significantly. Weekly plan creation, which used to take over two hours, now often takes only 30 to 40 minutes for the actual consideration time, now that I am used to it.

With the time I saved, I started writing notes reflecting on how the children are doing. As records of support accumulate, the next weekly plan becomes more concrete. I feel like a virtuous cycle like that has been gradually forming.

The reason I couldn't immediately tell my colleagues 'try using this' was because of my initial failures. I couldn't recommend it when it wasn't going well. But since the way I use it changed, I have gradually been able to talk about it.

In the paid section that follows, I have included the actual prompt examples I use and the template for creating weekly plans as they are.

I have arranged them so you can just copy and paste them and rewrite the information for your own class. I have prepared them for both special needs classes and regular classes. I made them for those who say, 'I understand the philosophy, but what should I do tomorrow?'

■ Why a 'template' for prompts is necessary

At first, I was lost every time about what and how to ask.

Even if I entered 'think of a lesson plan,' all that came out was generic content. I understand now why that is. The information I was giving the AI was too little.

When I properly verbalize and provide the reality of my class, the cognitive characteristics of the children, and the objectives of the unit, the output changes drastically. It was from that realization that I created a 'template' for prompts.

■ Prompt Example 1: Creating a draft for a weekly plan (for special needs classes)

Please copy the following as-is and rewrite the parts inside 【 】 to use it.

私は【小学校・特別支援学級】の担任です。
今週の【国語】の授業について、展開のアイデアを一緒に考えてほしいです。

■ 今週のねらい
【例：「ひらがなの『を』と『お』の書き分けができる」など、1〜2文で】

■ 子どもの実態（個人が特定されない範囲で）
【例：「文字を視覚的に捉えることが難しい子が1名。活動の切り替えに時間がかかる子が2名。】

■ 授業時間
【45分 or 30分など】

■ 使える教材・環境
【例：タブレット1人1台、ホワイトボード、手順カードあり】

上記をふまえて、授業の展開案（導入・展開・まとめ）を3パターン出してください。
私が取捨選択するので、バリエーションを重視してください。

Tips for use

While comparing the three options provided, I decide, 'I'll just use this activity part' or 'This flow doesn't fit my class.' Don't try to use everything. It's fine to just take the best parts.

■ Prompt Example 2: Deepening Support Strategies

A prompt to use when the lesson plan is set, but you feel that 'the consideration for this child is still thin.'

以下の授業展開で、【活動の見通しが持ちにくい子ども】への支援の手立てを
もう少し具体的に考えてほしいです。

■ 授業展開（先ほど決めたもの）
【展開の概要を貼り付ける】

■ その子の具体的な困難さ
【例：「次に何をするかわからないと不安が高まりやすい。口頭説明だけでは伝わりにくい場合がある。」】

支援の手立てを5つ提案してください。
私が実際に使えそうなものを選ぶので、多様なアイデアを出してください。

Instead of just writing 'provide visual cues,' the more specifically you describe the difficulties, the more concrete the resulting strategies will be. This is the one thing I most wanted to tell myself when I first started using it.

■ Prompt Example 3: Roughly Creating a Weekly Plan (For Regular Classes)

私は【小学校3年生】の担任です。
来週の週案のたたき台を一緒に作りたいです。

■ 各教科の単元（来週の予定）
・国語：【例：「もうすぐ雨に」第2時】
・算数：【例：「わり算」導入】
・理科：【例：「植物の育ち方」観察】
・その他：【道徳、体育など】

■ クラスの状況
【例：「グループ活動が得意。書く活動に時間がかかる子が複数いる。」】

各教科について、ねらいと主な学習活動のアイデアを出してください。
私がそれをもとに肉付けして週案を完成させます。
あくまで「たたき台」でいいです。

Using this prompt, the framework of a weekly plan comes out in 10–15 minutes. From there, I add and revise it myself. The reason I've been able to keep it within 30–40 minutes in total is because this flow has been established.

■ Weekly Plan Template (Special Needs Class, 1 Week)

Below is the structure of the template I actually use. Please copy and paste it into Google Docs or Word to use it.

【週案テンプレート：特別支援学級用】
対象期間：　年　月　日（月）〜　日（金）
作成日：
担当：

━━━━━━━━━━━━━━━━━━
■ 今週の重点目標
（クラス全体として大切にしたいこと）
→

■ 個別の重点目標
・[子どもA]：
・[子どもB]：
・[子どもC]：
※個人名ではなく「Aさん」「Bさん」など

━━━━━━━━━━━━━━━━━━
■ 各教科の計画

【国語】
・単元名：
・ねらい：
・主な活動：
・支援の手立て：
・使用教材・ICT：

【算数】
・単元名：
・ねらい：
・主な活動：
・支援の手立て：
・使用教材・ICT：

【生活】
・単元名：
・ねらい：
・主な活動：
・支援の手立て：
・使用教材・ICT：

━━━━━━━━━━━━━━━━━━
■ 今週の見通し・懸念点
→

■ 週末の振り返り欄
（金曜に記入）
・うまくいったこと：
・次週に活かしたいこと：
・個別の気づき：

I created the 'Weekend Reflection' section to apply records of support to the following week's plan. The records I write can be used for the next prompt. By setting up that kind of flow, the accuracy of the input to the AI improves with each passing week.

■ Checklist: AI Utilization Flow for Weekly Plan Creation

□ 自分の言葉で「今週のねらい」を1〜2文書く
□ 子どもの実態（個人が特定されない範囲）を書き出す
□ プロンプトに情報を入れてAIに投げる
□ 出てきた案を読んで「使える／使えない」を仕分ける
□ 使える部分だけ取り出して週案に組み込む
□ 特別支援が必要な子への手立てを自分で確認・加筆する
□ 週末に振り返り欄を記入する
□ 気づきを翌週のプロンプトに反映する

This flow takes a little time to get used to at first. But if you continue for 2–3 weeks, it will become second nature. That's how it was for me.

■ Points I Want to Share as Precautions

When inputting into AI, I make sure not to include information that can identify individuals, such as a child's name or school name. I consciously continue to stick to expressions like 'Student A' or 'a third-grade boy.'

Also, AI output is just a 'draft.' I make the final decision myself. Even if the AI says, 'This fits this child,' I am the one who knows the child. I make sure never to let go of that responsibility.

Summary

My first step in using AI was a failure. But because I failed, I thought seriously about 'how to use it.' Are there any other teachers who have thought, 'I tried using it, but it didn't quite feel right'? I think that feeling is probably correct.🐱

Having a prompt template changes things significantly. Let's try and error together.🐱✨

References/Sources

*1: Ministry of Education, Culture, Sports, Science and Technology (MEXT), 'Provisional Guidelines on the Use of Generative AI in Elementary and Secondary Education (Ver. 2.0)' (December 2024)
https://www.mext.go.jp/a_menu/other/mext_02412.html

• Science Portal, 'Supporting Appropriate Use of Generative AI in Educational Settings, Guidelines and Supplementary Materials Released One After Another' (March 2025)
  https://scienceportal.jst.go.jp/newsflash/20250312_n01/

note tag

#FieldCaseStudy, #TeacherInnovation, #AIUtilization, #ClassroomImprovement, #before-after, #SpecialNeedsEducation, #Prompt