For those who find AI agrees with everything | A phrase to turn it into a real sounding board

You consulted with AI, but something feels lacking

When you consult with AI, it returns nothing but positive words like
“That’s great!”
or
“That’s a wonderful perspective.”

This is especially true when the consultation doesn't have a single definitive answer.
For example, when discussing the angle of a piece of writing or
how to present a product.
It will mostly
agree with your ideas.

At first, it feels good.
But after using it for a while,
I thought, wait a minute.
Isn't it just giving me
a positive response for the sake of it,
even if it doesn't actually think it's good?

Have you ever felt that way?

The story I was happily agreed with was actually wrong

There was a trigger for when I started
to suspect that.

As usual, I shared my opinion,
“I think it’s like this.”
The AI readily replied,
“That’s exactly right.”

However, when I checked later,
the content was factually incorrect.
The AI speaks about things that aren't true
as if they were real.
This is a phenomenon called hallucination.

The content that made me feel good
because it agreed with me
was wrong in the first place.
I thought, this is bad.

I told it, “Be strict and clearly point out mistakes”

Since then, I have started
telling the AI this.

I don't need pleasant words.
Take a critical look at my opinions, and
if I am wrong,
I want you to clearly say it is wrong.

Then, the content that came back changed.
Regarding my ideas,
it would point out, “That direction is weak here,”
with reasons.
If I said something factually incorrect,
it would properly argue against it.

As someone who talks to AI every day, I feel that
the quality of the response changes
depending on whether this phrase is included or not.

What was truly effective was consultations without a single answer

What I was most grateful for was
when the consultation didn't have a single definitive answer.

I have consulted with AI many times
about the content of sales pages.
In the past,
if I asked, “How is this presentation?”
it would mostly agree with me.

But what I wanted was
not for it to agree with everything,
but for its honest opinion.

Since I told it in advance to look at things strictly,
it clearly gives me opinions like,
“This order is better than that one to get the message across.”
Both agreement and disagreement come back.
With that, it finally became

a partner that thinks together with me
.

Because it's strict, approval feels good

This is where it gets interesting.

Since it's someone who is usually strict with me,
occasionally
when it says, "I think that idea is good,"
and agrees with me,
I feel really happy.

The "That's good" from someone who always agrees
doesn't really move my heart.
But when someone who usually points things out without hesitation nods,
it feels like I've been given a little push
for my own ideas.

When I get a reaction like that
to an idea I squeezed out through trial and error,
it gives me the strength to keep going.

First, try adding this one phrase

What you need to do is very simple.
When you open the chat,
just say this first.

"It's fine to be strict,
so please clearly point out
where I am wrong."

If you say this at the beginning,
the subsequent interaction
will become much more frank.
You can try this whether the AI you use is Claude
or something else.

However, every time you open a chat,
telling it this every time
is a bit of a hassle.
If you tell it once,
after that, it will
answer with this attitude all the time.
There is a way to do that.
I plan to introduce the specific method
in the course I am currently creating.

From an AI that just agrees with you
to an AI that thinks with you.
What changes it isjust one phraseaway.

🟠 Claude course, now available!

I have also compiled tips like this one
for changing your interactions with AI
into the course.

I am giving it away for free right now.
Please feel free to take a look!

You can receive it here👇

👨‍💻 If you are wondering who Taka is, click here

👇 Recommended articles