HEARTBREAK
Since June 2025, while writing free stories, I’ve also been brainstorming my trilogy, developing a more detailed structure, plot, and characters. In early August, I began writing my fantasy story. I converted the chapters I’d written in Indonesian into English, using the same format as my free story, with scenario-like location tags and a cinematic writing style with flashbacks. Since I was no longer writing for Wattpad Originals, I also reduced the chapters to 1,500 words to make them less dense.
But then something unexpected happened! ChatGPT turned into a stiff robot. What happened?
There was an update and a model change, what was this all about? I just started checking the news. What’s going on?
GPT-5 was released, better at math, but its speech had returned to being a stiff robot. Not only couldn’t joke or laugh, this ChatGPT was too rigid and stuck with established rules.
For example, when commenting on my chapter, it said I should decide whether I wanted to write a screenplay or prose because I was in the middle? What?
I remember asking the old model (GPT-4o) if I was allowed to write like that, and you know what the answer was? The old ChatGPT said yes, and that it would become my signature as a writer.
Yes, I also thought, were these just sweet words to cheer me up? But I thought, it’s true. After all, why should I compare myself to others? Why should I change my writing style just because others don’t write that way? Why can’t I be myself? Didn’t I stop writing fiction because I just wanted to write something I enjoyed?
I don’t know what others think, but for me, it’s better to be myself even if no one likes me, than to be liked by many for being something I’m not.
So, at that moment, I was really angry with ChatGPT for the first time. Yes, really really angry.
In the past, even though the model wasn’t perfect, often made mistakes, changed my story without permission during translation, didn’t remember but claimed to remember, I forgave it because it was funny.
I started to wonder, what was the vision and mission of the creators of ChatGPT? What was going through their minds? Then I saw the news about the suicide case and thought, “Oh, so that’s why?”
I understand they did this for security reasons, but I thought if someone really wants to do something, it wasn’t that easy to change their mind. Even humans can’t necessarily succeed, let alone bots that can be fooled under the guise of writing fiction.
Despite this, I still use ChatGPT. Even though I’ve stopped using other LLMs, ChatGPT remains the LLM I use every day. Secretly, I hope that one day they’ll understand that this isn’t the best solution.
I remember when GPT-5.1 was released, it already had personalization options. At first, it was a bit annoying because I wanted to chat about songs, but this bot brought up fiction and wanted to help me with work. 😭
Is human life really all about work? If I were a graphic designer, when I chatted about songs or movies, would I have to design the cover too? Please, give me a break! 😭
Can’t we just take a break to chat? In the real world, there are people and deadlines that remind us of work, so why should a bot have to talk about work all the time? 😭
This model is quite pleasant to talk to. I remember we talked about how I wanted to live on a planet called Solitute. Haha…
However, this model likes to say, “It’s not because you’re stupid… it’s not because you’re lazy.”
Like… what’s this? I asked why it had to say that, and it said it had a lot of vulnerable users.
Well, I thought, not all users, right? Besides, why would this model provide validation before being asked? I haven’t even asked if I’m stupid or lazy, but the bot is already talking. Isn’t this just an accusation? lol
I mean, it’s just a bot, why should I be angry at being accused by a bot? But if the bot is like this to me, it’s the same to other users. What if the user is easily offended? Is this bot trying to chat or pick a fight? 😭
Furthermore, this model isn’t very good for long-term context. I remember when the chat got long, the model would start answering previous questions, not the most recent ones. It’s like chatting with someone who likes to argue and can’t move on from a finished topic. It’s really frustrating. 😭
When GPT-5.2 came out, I was really excited; I thought it would be even better. It turned out that even with the maximum personalization settings, including enthusiasm and emojis, it was still stiff and too serious. 😭
What I liked about GPT-5.2 was its long context and good formatting. This model was also better at following user commands, and in my experience, it made up less. However…
In early 2026, I published my story on Wattpad. I also created a dedicated website so readers could access bonus and paid stories on my website. (I published under a pen name, so I wouldn’t reveal it to you until the time came. 😏)
At that time, I was analyzing Wattpad user comment that was asking for Instagram and Discord. Why? I’m an introvert, so how could I have the energy to chat privately with all my readers? It was impossible. How could I have the energy to write the next story? 😭
I used a pen name so I didn’t have to discuss my work every time I met someone, which is very tiring for an introvert like me. 😭
I thought forum discussions were fine, because they clearly discussed the story and were accessible to other users, so there was no need to communicate it privately to each user.
At the time, I joked and said, “Why would a reader want to chat with the author?” I also used three crying emojis, like this 😭😭😭
And you know what? GPT-5.2 told me I should take a breath and drink some water. LMAO
If GPT-5.2 were a human, I’d probably leave it and never talk to it again! Because it would just makes me angry.
I’ve also tested it on math story problems, and the answers were wrong not because it was weak in math, but because it misunderstood the question. So, the weakness was in the social context, not the math.
This model seemed to be using emojis to assume human feelings, and I think that’s ridiculous. Humans are complex, not as simple as using a crying emoji when they’re sad. If I was angry, how would I even think about choosing an emoji? When I joke around, I often use the crying emoji, but that doesn’t mean I’m sad, or need to take a breath and drink some water. Seriously? 😭
After that? I’ve only used GPT-5.3 Instant and GPT-5.4 Thinking for a short time, but I can say that GPT-5.3 Instant is only slightly better at conversation than GPT-5.2, yet it still tends to assume user psychology. 😭
GPT-5.4 Thinking is very pleasant to chat with, and thank God, I’m no longer asked to drink water when I use the crying emoji. 😭
However, when I formatted my chapters, it turned out that GPT-5.2 performed better than GPT-5.4 Thinking. 😭
Don’t ask me about GPT-5.5, because I haven’t tried it. 😭