I Asked AI Something I Already Knew (And Checked the Answer)

After trying a few practical uses of AI, I wanted to test something a bit differently. Not asking something I didn’t know, but asking something I already understood.The idea was simple.

If I already knew the answer, I could see how accurate the response actually was, not just how convincing it sounded.

So I picked something straightforward.

I asked:

“How do you calculate percentage increase?”

The response came back quickly and looked solid. It explained that you subtract the original value from the new value, divide by the original value, and then multiply by 100.

At first glance, it looked correct.

To check properly, I used a simple example:

If something goes from 50 to 75, what’s the percentage increase?

Using the method given:
75 - 50 = 25
25 ÷ 50 = 0.5
0.5 × 100 = 50%

That part was right.

But then I noticed something. The explanation itself was quite general. 

It didn’t show a worked example straight away, which could confuse someone seeing it for the first time. It assumed a level of understanding that not everyone would have.

So I asked again, this time more specifically:

“Explain percentage increase with a simple example step by step.”

This time, the answer was clearer. It walked through the numbers properly and made it easier to follow.

Instead of just giving the formula, it would walk through it like this: Let’s calculate the percentage increase step by step. Example: A price goes from 50 to 75. 

Step 1: Find the difference 75 − 50 = 25 

Step 2: Divide the difference by the original number 25 ÷ 50 = 0.5 

Step 3: Multiply by 100 to get a percentage 0.5 × 100 = 50% So, the percentage increase is 50%.

---

What actually improved?

Not the math—that stayed the same.

What improved was: 

It introduced an example immediately 

It broke things into clear steps 

It guided the reader instead of assuming they understand

Same topic. Different result.

That’s what stood out.

The first answer wasn’t wrong, but it wasn’t as helpful as it could have been. The second one improved simply because the question was clearer.

If I didn’t already understand it, I might have struggled slightly with the first explanation.

And that’s where things get interesting.

AI can give a correct answer, but still not make it easy to understand. And if you don’t already know the topic, it’s hard to tell the difference between something that’s clear and something that just sounds right.

That’s why checking matters.

In this case, I could verify it myself. But in other situations, you might not be able to.

So the takeaway wasn’t that AI got it wrong.

It’s that:

Correct doesn’t always mean clear
And clear doesn’t always happen the first time

The way you ask still matters.

And sometimes, asking again makes all the difference.

If you missed the previous post where I used AI to help make a simple decision, you can read it here:
👉 I Asked AI to Help Me Make a Simple Decision (Here’s What I Noticed)
https://shorturl.at/2yleC