From The Newspaper (155) Generative AI (96)
From The Newspaper (155) Generative AI (96)
[From his experience of ChatGPT Web Course] (No.31)
Actually, there were 27 questions in the game that he had made with Python with the help of ChatGPT. Although, technically, he ordered ChatGPT to make ‘around’ 20 questions. It made 27.
Anyways, as he had a small meeting with some members of his guide group, he brought his laptop computer and asked one of the members to try to play the game.
The first word quiz offered her “calm”, “offers”, “discoveries”, “Each”, “visit”, “of”, “a”, “new”, “and”, and “sense”.
She was confused and couldn’t answer well. She gave a wrong answer.
The next one was “Three”, “are”, “Morioka”, “must”, “Great”, “Noodles”, “try”, “The”, “of”, “Dishes”.
She also couldn’t answer the second quiz correctly either.
She complained saying, “I have Pre 1st grade of Eiken, and maybe these questions aren’t good.”
Before this meeting, he had made an ‘Easy Version’. He ordered ChatGPT to use A1 level words. He offered the easier version to her.
They were something like:
“is”, “Morioka”, “This”, “City”.
“nice”, “It”, “a”, “city”, “is”.
“local”, “We”, “eat”, “food”.
“There”, “old”, “is”, “an”, “castle”.
She got angry, so he stopped this experiment.
At home, he checked some other questions from the other 27 word quizzes.
They were :
“paths”, “line”, “and”, “rivers”, “beauty”, “Walking”, “offer”, “seasonal”, and “the”.
“is”, “ironware”, Nambu”, “known”, “elegance”, “a”, “for”, “traditional”, and “crafts”.
“leaves”, “in”, “and”, “city”, “golds”, “color”, “reds”, Autumn”, “the”, and “vivid”.
They were not super difficult.
But, he asked ChatGPT to make questions on the level of B2.
This time there were 15 questions. A1 level ones were 20 pieces. So, he understood that ChatGPT somehow has a sense of ‘roundabout’, like humans. Maybe if he had ordered in the first place to make ‘20 questions’, not ‘about 20 questions’, it would have made just 20.
As the level B2 sentences were different from the first ones, he started to do the next stage.
In the process so far, ChatGPT suggested something for the next stage.
He remembered one of them. In the experiment for her, she complained that she couldn’t return after realizing her mistake and had to reset. It was troublesome. He asked ChatGPT, “Can I put an additional button on the game in order to return one process from the former click.”
ChatGPT praised :
“Good question.
The Undo (go back one step) button is very effective for sentence-reordering practice.
And the best part is — it’s easy to implement 👍
If you add the code below directly to your current GUI version, it will work as is.
(Explanation assumes you are running it in IDLE.)
① First, the idea behind “Undo” (very important)
In the current game, you have:
・selected_words (the order of the words the learner selected)
・buttons (the word buttons)
The Undo function does only three things:
1.Remove the last selected word from the list
2.Re-enable the button for that word
3.Update the displayed sentence
That’s all there is to it.”
Then ChatGPT added how to do these in detail and their codes. But, he had a hard time.
(To be continued)
No.641
January 23rd,2026
