AI Beginner Subscribes to ChatGPT and Claude. Realizing the Importance of 'Prompts' While Creating LINE Stickers [Day 2]

It is Day 2 of the AI Challenge 365.

In yesterday's article,

I wrote about the reason I started this challenge:

"How much can I change if I, someone who was bad at AI, continue to use AI for 365 days?"

And the first challenge I chose is "AI × LINE Stickers"

Today, I actually used AI to create LINE stickers and proceeded to the point of submitting them for review for sale.

However, I made one big decision before that. **I subscribed to both ChatGPT and Claude.**Trying to do it for free, I made no progress at all

At first, I thought I would start without spending as much money as possible.

Even though I am aiming for monetization with AI, I haven't earned a single yen yet.

I also had feelings about whether it was right to spend money from the start in such a state.

So, I was trying various things with the free plan first.

However, when I actually started using it, the limits were reached faster than I thought.

Especially when creating or modifying images while interacting with AI repeatedly like this time, there were situations where I couldn't proceed with the work as I wanted.

"I want to try more, but I have to stop here..."

That's how it felt.

Even though I decided to "seriously use AI for 365 days," it would be counterproductive if I reduced the number of times I used it because I was worried about the limits.

So, I decided to think of this as self-investment to study AI, not a waste of money.

And I upgraded both ChatGPT and Claude to their respective $20/month plans.

Is it necessary to have two all of a sudden?

There is also that question.

Honestly, I don't know yet.

So, that is also part of the experiment.

I will actually use both, and

"What is the difference?"

"Which one suits me better?"

"Is it necessary to subscribe to two in the first place?"

I will verify these points in the future. Trying to create LINE sticker images with ChatGPTOnce I subscribed, I immediately proceeded with creating the LINE stickers.

This time, I tried generating images for LINE stickers using ChatGPT.

What I thought at first was,

"If I tell it the image I want to create, the AI will create it nicely."

But when I actually tried it, it wasn't that easy at all.

Looking at the generated image,

"Something is different."

I give instructions again.

"Still different."

I modify it again.

I repeated this many times.

Here, a word that cannot be avoided when using AI appears. What is a "prompt"?

What is a "prompt"?

. In yesterday's article, I also wrote, "Prompt? Generative AI? How should I give instructions?" Prompt

Prompt

I am currently struggling with that "prompt."

Simply put, a prompt is

an instruction or question to tell the AI "what you want it to do."

For example, for image generation, "Create an image of a cute dog." This is also a prompt.

However, if you only do this, there will be a lot left to the AI side.

What kind of dog?

What kind of expression?

What kind of pose?

What about the background?

Is it an illustration or realistic?

Should I include text?

Even if I have the finished form in my head, the AI cannot see it.

Therefore, if I don't convey it specifically, an image different from what I thought will come out.

This time, I remade the image many times and realized

"It is better to give instructions to the AI more specifically than I thought."

I should have asked the AI how to create "good prompts" from the beginning

And halfway through, I realized one more thing. I was inputting the prompts I thought of myself into the AI from the beginning. But when I think about it,

I should have asked the AI how to create prompts themselves.

For example, I should have asked at the beginning, "I want to create an image for LINE stickers, what information should I put in the prompt?" If I did that, I could have been told by the AI side what to specify, such as character features, expressions, poses, composition, background, and style.

If I had started generating images after creating a prompt based on that, I might have been able to reduce the number of times I had to redo it like this time.

This was a pretty big discovery for me as a beginner.

If you don't know how to use AI, you should just ask the AI how to use it.

I ended up actually experiencing what I wrote on Day 1:

"If you don't know, ask the AI."

While redoing the images many times, I was finally able to shape them into LINE stickers. Even so, I completed the LINE stickers and submitted them for review

Even so, I completed the LINE stickers and submitted them for review

And today,

It doesn't mean I've been able to sell them yet. I don't even know if they will pass the review. I actually submitted the created LINE stickers for review.

I actually submitted the created LINE stickers for review.

Of course, even if they go on sale, there is no guarantee they will sell.

But from the state of saying "I'll try something using AI" until yesterday, today I was able to actually create a product with AI and proceed to the application to put it out into the world.

As for the amount, it is still 0 yen.

Current revenue: 0 yen.

Even so, it is a step forward for me.

Today's learning

Today's learning

What I felt most today is that

AI is not a "magic tool that will create anything in one go if you ask it casually."

At least for me right now, I lack the ability to accurately convey my image to the AI.

That's why I ended up redoing it many times. But thinking about it the other way around, it also means that

And if there is something I don't understand, I don't need to worry about it alone. I should just ask the AI, if I improve how I give instructions, there is a possibility that I can use it more efficiently.

if I improve how I give instructions, there is a possibility that I can use it more efficiently.

"How should I give instructions?"

This feeling was not something I had as someone who felt uncomfortable with AI until yesterday.

On the second day, I subscribed to ChatGPT and Claude. I remade images with AI many times, learned the importance of prompts, and finally submitted the LINE stickers for review. After finishing Day 2

After finishing Day 2

Revenue is still 0 yen.

Rather, it is a start from a negative balance due to the subscription fees for AI.

But when I look back after 365 days, I want to be able to say that this subscription was truly a "self-investment."

And whether the LINE stickers I created this time

will pass the review,

will pass the review,

and whether they will sell if I actually put them on sale.

I will record the results from here on out.

How much can an AI beginner change if they challenge themselves with AI for 365 days?

It's only the second day.

I will continue to try using AI tomorrow as well.