Lessons for Raising Children Who Don't Just Leave Everything to Generative AI

I used generative AI in a 5th-grade Japanese language unit titled "How to Make Writing Persuasive." This time, the style involved students directly operating the generative AI themselves.

The Reality of the Lesson

First, students learn the structure of an opinion piece by referring to examples in the textbook:
・Claim
・Evidence
・Anticipated counterargument
・Thoughts on the counterargument
・Conclusion
After that, we consider questions such as, "What effect does repeating your claim in the conclusion have?" and "Does the way a claim is conveyed change depending on whether or not there is an anticipated counterargument and thoughts on that counterargument?"

"Now, I'd like us to write opinion pieces, but let's use AI for that. I'll go first."
The prompt I entered was as follows:

I share the generated text with the students and have them read it. The structure seems to follow the instructions, but they start pointing out various flaws in other areas. To introduce just a part of it, this is the text the AI produced...

"It's not 'often listen,' it's always like that."
"We wouldn't say 'refreshing activities.'"
"Instead of 'beneficial,' 'good' is fine."
"For 'conveying information,' 'announcements' or 'messages' is fine, right?"

And so on. So, I confirm with the students:

"As you can see, we probably can't just use what the AI produced as is, so please correct what it generated."
"Can we add to it, not just correct it?"
"Of course."
"Can we have the AI rewrite it before bringing it into Word?"
"Of course."

From here, the students finally operate the AI (tomoLinks) themselves. The learning process is as follows:

1. Have the AI create an opinion piece using the prompt from earlier.

2. Read the generated opinion piece.

3. Paste the generated opinion piece into Word and turn on [Review]-[Track Changes].

4. Correct parts where you think, 'This is wrong,' or add parts where you think, 'This is missing.'

5. Submit the corrected version using the assignment feature in Teams.

As for what happened with what the students submitted...

I've blurred it, but I think you can tell it turned bright red.

Is this okay for 'writing an opinion piece'?

For this practice, the first criticism likely to fly in would be, 'Is this really okay for learning how to write an opinion piece?' 'Isn't this just correcting what the AI wrote? Can you really call this writing?'

First, I should note that this unit is about learning "how to make an opinion piece persuasive," and the direct goal is not just to have them write. The goals of the unit are as follows:

◎ Be able to think about the structure and development of the entire text so that it becomes a logical piece of writing. (Thought, Judgment, Expression B(1)b)
○ Be able to understand how words relate to each other within sentences, word order, the connection between sentences, the structure and development of stories or texts, and the types and characteristics of stories or texts. (Knowledge, Skills(1)f)

Therefore, I believe there is no problem at all with having students use AI to write texts and then analyzing them to learn the items above. But setting that aside, let's think about this a bit more deeply. Is it wrong for children to use AI when writing opinion pieces?

It is true that the guidelines also list "submitting or entering works generated by generative AI as one's own work, such as for various contests, reports, or essays" as an example of what is considered inappropriate.

My practice is not that simple; the students put a lot of work into it. If I were to illustrate it, it would look something like this.

Some might criticize this, saying, "No, that's still not okay. An opinion piece is only an opinion piece if you write it yourself from scratch."
That may be true, but please think about it for a moment. Doesn't it work like this for adults?

Suppose there is a tedious task. If you use generative AI for it, it saves a tremendous amount of time. You could even have the AI complete it entirely. That is praised as "improving work efficiency" or "realizing work-style reform," isn't it? Why is it okay for adults to be praised for this, but not for children?

I may have phrased that a bit provocatively, but we are going to be asked similar questions by children countless times from now on.

"If generative AI can translate everything, why do we have to learn English?"
"If generative AI can instantly summarize any long text, why do we have to learn how to summarize?"
"If generative AI can write a book report in an instant, why do we have to write it ourselves?"

How will you answer those questions? How will you conduct your classes? My class is one answer to those questions.

Why it turns bright red

As you have seen, the Word files submitted by the children are all bright red. They have put a tremendous amount of work into them. There is no child who submitted what the AI output as-is. The same goes for children who revised their prompts and had the AI output multiple times. Why does this happen?

I think there are several reasons, but the biggest one is that the children had the intention that "this is my opinion piece" and "I am going to make this my own opinion piece."

The part where I first gave the example of "Elementary schools don't need morning assemblies" and had the AI generate an opinion piece was very exciting. Since it was an argument brought into a space where many students were already thinking, "Morning assemblies are a hassle" and "Why do we have to do them?", they were able to experience what it means to put their own opinions into writing. That is why I think they were not satisfied with what the AI produced and offered various revision suggestions.

In other words, the children were convinced of the task in this class. No one said, "If generative AI can write an opinion piece in an instant, why do we have to revise what it generated?"

Next, the fact that "employing the perspectives and ways of thinking of Japanese language arts" had taken root was also a major factor. Since this is a class that has been using AI since April and has been critiquing what the AI generates every time, they seem to be quite particular about words and writing.

However, more than anything, the biggest factor in "employing the perspectives and ways of thinking of Japanese language arts" being rooted is (though it's a bit self-serving to write this) that I have been conducting proper classes on a regular basis. Because of my position, I inevitably have many experimental attempts. If that is the case, it is all the more important to conduct classes properly, otherwise the class would fall into chaos.

Summary

I don't know what the outcome of the discussions by the "Council on the Use of Generative AI in Primary and Secondary Education" will be. Since diverse opinions are being expressed from various standpoints, the final report will be extremely difficult to compile, and there is no doubt that from my perspective, there will be parts of the finished product that I am not convinced by.

However, that is the same for every committee member. What is important is to state one's own thoughts clearly. Also, in my case, as the only teacher currently in the field, I consider it my duty to share "what kind of generative AI utilization practices I am conducting" in real-time.

So, that is why I am writing this note, but honestly, it is hard to understand just from this. If I were a researcher, I would compile this into a paper, but in my case, I would really like you to see the actual classes. You are welcome to observe my weekday classes at any time, provided they do not conflict with school events, so please feel free to contact me.

For those who think, 'Well, weekdays are a bit difficult...', please join this seminar!