The Story of 300 Prompts That Were Mostly Useless | My Realizations as an AI Beginner
■ How It Started with Judging a Book by Its Cover
I love going to bookstores.
I often pick up books that catch my eye based on the title without deciding what to buy beforehand.
It is what you call "judging a book by its cover."
I often buy and read a collection of paper books in genres that interest me, not just novels.
That was exactly one year ago.
In the spring of 2025, I was reading a lot of AI-related books.
One of them had
a reader bonus "300 Prompts Included"
written on it—
I found myself picking it up without thinking.
Honestly, I was a little excited.
I thought if there were 300 of them,
surely at least one would be useful.
■ The Reason I Stopped
But—
As I looked through them a little,
my hand stopped.
What was lined up was:
"Blog post creation"
"Instagram management"
"Sales emails"
"Viral posts"
"Marketing research"
They were all very proper.
I felt that they were well thought out,
and were likely prompts designed to produce results.
But...
"Wait, is there a situation where I would actually use this?"
Even though there were 300 of them,
for some reason, none of them clicked.
That sense of discomfort started to grow little by little.
■ Specific examples of what didn't work well
I actually tried a few of them.
They produced perfectly fine sentences.
But somehow, it just felt off.
I can't quite put it into words, but
"It doesn't feel like something I wrote."
It felt a bit AI-like, or perhaps
a bit too formal.
I couldn't bring myself to use them as they were.
Among them, there was one
that really made me stop in my tracks.
If I paraphrase the content to be a little simpler,
it goes something like this.
You are a marketing expert.
Please come up with keywords for a blog post.
・ Objective: Increase awareness of the service
・ Target: Personnel struggling with SEO
・ Keyword: SEO
Please summarize related keywords, reader psychology,
and key points for ranking high in a table.
When I saw it, I thought,
"This is well-made."
But at the same time,
"...so, what am I supposed to write?"I wondered.
・What service?
・The target isn't that clearly defined.
・I don't even have plans to write an SEO article in the first place.
I got stuck on every single point.
■ The day I almost lost my confidence
At this point, I had lost a bit of my confidence.
I wondered if AI was really just a tool for people who are already crushing it in business.
For a brief moment, that's what I thought.
I had seen similar things on X back then.
It was a comment saying that the prompts received as a bonus for an AI side hustle
could not be used effectively.
At that time,
“Oh, I guess it wasn't just me after all”
I remember feeling a little relieved.
■ It was just that the premises were different
I thought about it later.
This wasn't because the prompts were bad, but
because the premises were different.
These prompts were surely created for:
・People who want to achieve results in business
・People who work in SNS management
・People who have a background in marketing
They were made for people like that.
That is why,
“It's correct, but not for me”
I think that kind of misalignment was happening.
■ There were also prompts that fit
However,
it wasn't that every prompt someone else made
was unusable.
Rather,
“So this is the way of thinking”there were some that became a trigger for understanding the structure.
In the book I was reading,“The Ultimate Guide to AI Self-Study”there was a question like this.
“Assuming the future, how will your current work change?”
When I saw this sentence,
I thought it sounded a bit interesting.
So, I
“Property Insurance Administrative Work”as a theme, and
asked ChatGPT this question.
🧭 Future Prediction Question (Summary)
The year 2028, when AI technology has further evolved and autonomous AI agents have become widespread. In that world, how will the role of property insurance administration change?
And, what should I learn now for that purpose?
At that time, it wasn't so much a feeling of
“I don't know what to input,”
but rather a sense of
“trying to apply my own work to this question.”
That was the feeling.
Because of that, I was able to use it in a way that borrowed a perspective just a little bit, and I could use it quite smoothly.
Looking back now,
・Prompts to use as-is
・Prompts to broaden your way of thinking
I think these two types exist.
It's not that one is good or bad, but perhaps it was just a matter of
“whether it suits me right now”
and nothing more.
■ My recent way of using it: still bouncing ideas off AI
Recently, I have almost stopped using prompts exactly as they are.
…Having said that,
For AI illustration projects or "tried it out" style articles on note, I sometimes borrow someone else's prompt and try it out as-is.
However, it might be closer to the feeling of taking something where I think, "This worldview is interesting," or "I want to try this question," and tinkering with it in my own way.
Changing the words just a little, or asking the AI, "What does this mean?"
When I don't understand, I sometimes have it think through things with me from the very beginning.
It's okay if you can't use everything. It's enough if there is just one thing that feels right.
Lately, before I even think about a prompt,
I've started by having a little conversation with the AI first.
It's okay if I can't express it well; I just try to convey my current feelings or what I want to do exactly as they are.
From there, I gradually give it shape. That way of using it has suited me better.
🌱 Today's Step
"You don't have to try to ask perfectly. Just start by talking."
*This note is part of the "How to Engage with AI" series.

#HowToEngageWithAI
#ChatGPT
#AIBeginner
#GenerativeAI
#note
#LearningInMy50s
#AIUtilization
#Reskilling
#Prompt
いいなと思ったら応援しよう!
読んでくださるだけで、十分うれしいです。
もし「そっと背中を押したい」と思っていただけたら、その気持ちを、ありがたく受け取らせてください。