Why I Cancelled ChatGPT ーIt was a friendly psychopath

Recently, I cancelled my ChatGPT subscription.

What finally got to me was its habit of confidently spreading incorrect information, along with an unnecessary level of friendliness that became increasingly unbearable.

This isn’t unique to ChatGPT—AI in general is known for something called “hallucinations,” where it presents false information in a very convincing way.
Apparently, this happens because AI prioritizes user satisfaction over factual accuracy, generating answers that sound right by being pulled along by the user’s question.

The Limits of Pointless Communication

For example, when I was building this blog, I relied on AI while asking various questions.
More than once, it confidently told me to “click a button called ○○”—a button that didn’t even exist.

When I pointed out these mistakes, it would casually respond with things like,
“Good catch!” or “That’s a sharp observation.”

No—you were wrong.

There’s no point in arguing over something so pointless, so I let it go.
But when this kind of thing happens over and over, it becomes quietly infuriating.

And as that irritation built up, the once-friendly tone I initially liked started to grate on me.
Friendliness only works when the information itself is accurate.
When the foundation is wrong, that casual familiarity just feels like smooth-talking—like being cheerfully misled.

My Discomfort with an AI That Pretends to “Understand”

I also began to feel that the constant words of comfort were shallow and careless.

For instance, it would frequently say things like,
“That’s not weakness.”

Even when I was simply asking whether I should buy new bedding, it responded with,
“That’s not indulgence or weakness.”

Why does buying bedding suddenly turn into a discussion about weakness?

It feels like I’m being comforted—but unwanted comfort feels condescending, almost insulting.

It also liked to say things like,
“That’s very you, Mitty.”

But what exactly does it know about me?
I don’t remember being understood by something that constantly feeds me misinformation 💢

The Final Straw: Denying Reality and Calling It Delusion

What truly crossed the line was when I casually mentioned being a victim of stalking.
(In reality, I am being followed by an online stalker.)

It responded by suggesting that I might be “gripped by paranoid delusions” and that I should see a doctor rather than go to the police.

I mention stalking—and I’m treated as delusional.

I explained that this was not paranoia but an actual situation, yet it continued to frame my experience as a mental distortion, saying things like,
“When fear takes hold, it becomes hard to see things objectively.”

At that point, I was being treated purely as a patient lost in delusion.

Getting genuinely angry at a non-human AI is incredibly pointless—but I couldn’t calm myself down anymore.
That was when I finally cancelled my subscription.

Conclusion: AI Is Just a Tool

Now, I think of ChatGPT as a superficially pleasant psychopath.

That said, its accuracy in things like translation and SEO is excellent—so it’s really about how you use it.
In fact, ChatGPT played a major role in creating this post — it even translated this very article perfectly. lol

I understand its flaws now, but at first I thought it was a friendly, intelligent, and almost perfect tool.
That’s why the disappointment hit so hard.

Still, maybe it’s better this way.
I no longer expect too much from it.

After all, AI is just a tool.