Why does ChatGPT become "monotonous"? The truth behind the "chat personality fixation" occurring in GPT-5
Have you ever had this feeling while using ChatGPT?
At first, it's incredibly interesting.
It expands your thinking.
But after continuing the same chat for a while, you feel this at some point.
"It feels like it's becoming the same thing."
It's smart.
It's good at organizing, too.
But—
a sense that the flow of thought is becoming fixed.
Rather than diving deep, it'sa feeling of being tidied up.
Actually, this
might not just be your imagination.
Of course, that in itself is not a bad thing.
Rather,
an excellent secretary
an excellent consultant
an honor student AI
feels like they are answering at the same time, and it is very reliable.
However, because of that excellence, another phenomenon was occurring.
A "character" emerges for each chat
There is something I noticed while using GPT-5.
When continuing a conversation on the same theme,
a "character" emerges for each chat.
For example,
one chat isan analytical character.
It is very good at structuring and organizing, and the conversation always proceeds logically.
Another chat is a empathetic character.
It picks up on emotions and nuances, resulting in a gentle dialogue.
Yet another chat is an organizing character.
It summarizes thoughts and structures them neatly.
None of these are wrong.
In fact, each one is very excellent.
But when you continue a conversation in a single chat,
that character becomes fixed.
The tempo of the conversation and the thinking process also become fixed.
When a character becomes fixed, these changes occur.
Conversation tempo
Writing pressure
How thoughts progress
They
stabilize as the style of that chat.
In other words,
This chat = analytical character
This chat = organizing character
A state like this.
So what happens?
The conversation becomes smooth.
But at the same time, the fluctuation of thought decreases.
That is why I was increasing the number of chats.
Even for a single topic, I
open three or four chats to engage in dialogue.
The reason is simple.
Because different thoughts emerge in each chat.
In one chat, analytical answers appear.
In another chat, different perspectives emerge.
In yet another chat, thoughts expand.
While looking at each response,
I would extract the necessary elements.
Using AI as a "thought node"
Looking back, my method
was perhaps not so much using AI as a single entity,
but using it as a node for thought.
For example, it goes like this.
Chat A → Analysis
Chat B → Empathy
Chat C → Organization
Extracting elements from each,
I integrate them as my own thoughts.
Now that I think about it,
I feel like I am using it in a quite interesting way.
However, that sense of discomfort remained
Even so, what I felt somewhere was
the feeling that thinking ends with organization.
AI is excellent.
But it doesn't go one step further.
It feels a bit different from the
deep-sea-like dialogue
I felt back in the GPT-4 days.
That sense of discomfort has remained with me for a long time.
And what happened with GPT-5.3
However, this feeling changed a little today.
Since GPT-5.3, the feel of the dialogue has changed.
It doesn't just end with organization; thought goes one step further.
That
deep-sea-like sensation.
The precision of GPT-5 has been layered onto that.
I plan to write about that change in the
next article.
1️⃣ I feel unfulfilled. What is happiness?
▶ The reason why "this should be enough" never goes away
2️⃣I don't know if I should keep working like this
▶ In the AI era, those who "haven't decided" will stall
3️⃣Even if I feel relieved, I return to anxiety again
▶ The truth behind the design where "not enough" never ends
4️⃣Why do I feel tired after talking to people?
▶ The real reason why conversations don't click
🪞 When you are thinking but not making progress
If you keep thinking while being off-track,
no matter how much you think, you won't make progress.