The table is small. The computer is grimy. The white mouse has already been through a few years.
On the screen, a very serious question: what would be the influence of cheese on the happiness of a penguin that learned Python?
The chat replies with the composure of someone who has just received a scientific commission.
The young man doubles over laughing. The next round is going to need pizza.
A hypothetical scene about a quite concrete use: playing with a conversational tool.
For him, that is a game
There is no overdue delivery. There is no presentation for tomorrow. He opened the chat because he wanted to have fun.
He invents an absurd challenge, watches the reply, and changes the rules for the next round. He asks for a formula for something that has no measure. He mixes academic language with an impossible story. He sees how far the machine’s solemnity will go along with the joke.
As in a mobile game, the urge to continue comes from the experience itself. One round opens another. Each combination can get more absurd and funnier.
Maybe someone looks over and asks: “But what’s this good for?”
He’s laughing. That part is already working.
Playing is also using
Software can be a work tool, a space for creation, or entertainment. The same interface can mean different things to different people.
On ChatGPT, for example, prompting unexpected replies can be part of a playful interaction with the GPT. That does not automatically turn the conversation into an attempt to attack a system.
There is research on this kind of use. Nikghalb and Cheng analyzed 372 Reddit posts about ChatGPT and identified different forms of playful interaction, including play, challenges, and attempts to fool the model. The study describes that sample of posts; it does not measure all users, nor does it grant unrestricted permission to use. Read Interrogating AI: Characterizing Emergent Playful Interactions with ChatGPT.
Fun doesn’t need to be disguised as productivity to deserve a place in the discussion about the product.
The wall shows up after the collision
Now imagine that, in the middle of the game, the tool changes its tone:
“You should not go beyond the limits.”
Which limits? The young man only changed the penguin’s profession.
“You should not explore the model.”
Explore its possibilities? Or exploit a vulnerability? The sentence mixes two different things and leaves the person trying to guess which of them was attributed to their conversation.
That is the irony: demanding respect for a boundary without making the boundary understandable.
A rule needs to explain the conduct it restricts. A technical name like “guardrail” can identify a protection, but on its own it does not teach the user where the limit is or why that interaction encountered it.
Who chooses the next round?
Paying for the service does not eliminate its rules. It also does not turn every conversation into an obligation to produce something useful according to the provider's criteria.
The point is being able to understand the rules and choose the purpose of use. Curiosity, humor, and experimentation are part of that choice.
In this post, the absurd formulas belong to the play. They are not scientific discoveries; the fun lies precisely in noticing the difference between an answer that looks rigorous and a conclusion supported by evidence.
The young man doesn't need to turn in a spreadsheet before shutting off the computer.
Maybe he just wants to find out what happens when the penguin asks for a refund on the pizza.
If you use AI to play, invent challenges, or test unexpected answers, share it through the post's private story form. At what moment did the fun work — or did the tool interpret your play as something else?