Give a fictional hotel receptionist a missing suitcase, a power cut and a guest who refuses to explain why he’s carrying a wedding cake.
There’s your evening’s entertainment.
You could write the receptionist’s opening line yourself, ask a chatbot to answer as the guest and see where the exchange goes. Perhaps it produces something funny. Perhaps the guest immediately confesses to an international conspiracy and ruins a perfectly promising scene. You get to reject that bit.
This is a modest use for conversational AI, which is part of its appeal. Nobody needs to claim that it transforms relationships. You’re making something up because making things up can be enjoyable.
For adults curious about AI companions but uncomfortable with the grand promises surrounding them, a creative project offers somewhere concrete to begin.
Start with a scene you actually want to finish
A blank instruction such as “entertain me” gives the software most of the decisions. Give it a situation instead, and you have something to work with.
Try a dinner between two old friends. One has secretly applied for the other’s job. Neither wants to mention it before the food arrives.
Write the first exchange. Then ask the chatbot to continue as one character while you write the other. Keep the instructions narrow: no confession yet, no sudden emergency, no convenient telephone call that lets somebody escape.
The pleasure is in deciding what stays unsaid.
If the dialogue becomes too polished, ask for an interruption or a misunderstood remark. If every response explains a character’s feelings, cut the explanation. People can argue about who ordered the wine when they’re furious about something else entirely.
You don’t need ambitions to publish. A page that makes you laugh is a reasonable result.
I’d also resist asking the chatbot whether your writing is brilliant. Praise is easy to generate and difficult to spend. Ask a question with an answer you can inspect: “Where does this conversation become repetitive?” or “Which line reveals the secret too early?”
That leaves the final judgment with you. A hobby gets rather thin when the machine supplies the work and the applause.
The 2026 figures leave plenty of room for plain entertainment
Pew Research Center’s 2026 figures suggest that amusement deserves more attention in the conversation about chatbot use.
The striking comparison is among adults aged 30–49: 36% reported using chatbots for entertainment, compared with 6% for companionship.
|
US adult age group |
Fun or entertainment |
Creating or editing images/video |
Companionship |
|
18–29 |
33% |
34% |
7% |
|
30–49 |
36% |
34% |
6% |
|
50–64 |
20% |
18% |
2% |
|
65+ |
7% |
5% |
1% |
Source: Pew Research Center, 2026, chatbot activities by age. Survey conducted 17–23 February 2026. Percentages cover all adults within each age group. Activities overlap and must not be added together.
These are reports of use, not evidence of creative achievement or improved well-being. Nor do they identify how many respondents tried fictional dialogue.
They do give us a reason to widen the discussion beyond digital friendship.
Someone might spend an evening devising characters without wanting a simulated relationship. Another person might enjoy both. It’s worth knowing which experience you want before choosing a product, particularly where payment is involved.
For an adult comparing character-based services, Joi ai is one example whose homepage presents character creation and chat within an adult-oriented offering. A useful comparison would examine how those functions work, what costs extra and whether the available controls suit the intended activity. Its presence in that category doesn’t establish a psychological benefit.
Keep your expectations specific enough to assess. “Can I enjoy writing an exchange with this character?” is answerable. “Will this make me feel understood?” brings a much larger hope to the purchase.
Let the conversation produce something outside the app
Atsugiham’s article on creative hobbies at home puts writing, digital art and photography alongside activities such as painting and gardening. That’s a useful context for AI experiments: another way to spend free time making something, rather than an obligation to adopt a new lifestyle.
The best version, in my view, leaves you with something you can enjoy after closing the conversation.
For example, take a photograph of an ordinary object at home. Write a fictional history for it. A chipped cup could have belonged to a disastrous seaside café. An umbrella might have passed between strangers who kept missing the same train.
Ask the chatbot for competing explanations, then choose one and rewrite it yourself. Keep the detail you like. Throw away the grand revelation about the umbrella saving civilization.
Or bring the activity to a friend. Exchange fictional postcards from invented holidays. Each person writes the reply to the other’s increasingly implausible travel problems. AI suggestions can sit in the background while the shared joke belongs to you both.
There’s no requirement to turn this into a disciplined practice. Save a paragraph, print a postcard or abandon an experiment that bores you.
Still, an ending helps. “Finish this scene” gives the evening a shape. “Keep talking until I feel satisfied” is harder to recognise when it arrives.
A fictional character doesn’t need your private correspondence
Creative play offers a useful privacy advantage: you can invent the material.
You don’t need to upload a former partner’s messages to write an uncomfortable dinner. You can give the characters different names, different circumstances and a problem that never happened.
That protects other people’s information and gives you more freedom as a writer. You’re no longer checking whether every sentence accurately represents an actual argument.
Before sharing anything personal, read the service’s explanation of storage, deletion and data use. Don’t assume that deleting a conversation and deleting an account mean the same thing. Adult-oriented products also belong with adults.
Emotional expectations deserve similar care.
MIT Media Lab and OpenAI’s 2025 controlled chatbot-use study involved 981 participants over four weeks and more than 300,000 messages. Its publication summary reports worse measured outcomes among people who voluntarily used the chatbot more, while assigned experimental conditions produced no significant effects. Usage duration wasn’t randomly assigned, so the association doesn’t establish that longer use caused those outcomes. The study examined a particular chatbot setting over a short period, rather than every companion service.
The Guardian’s 2025 reporting on heavy chatbot use and loneliness also highlighted uncertainty about cause and effect.
None of that turns a fictional conversation into something shameful. It does argue against treating enjoyable interaction as proof of emotional benefit.
You can notice more ordinary things: whether you spent what you intended, stopped when you wanted and enjoyed the activity itself.
Then return to the hotel receptionist. The electricity is back. The suitcase has appeared. The guest is still standing at the desk, holding a wedding cake with somebody else’s name on it.
Give the receptionist a better question.