They come for homework help, CV rewrites or quick facts, yet the conversation keeps circling back to something far more intimate: a simple sentence aimed at understanding what, exactly, is on the other side of the screen.
The question that keeps haunting ChatGPT
When engineers and psychologists ask large language models what people request most often, the answer is not an obscure maths problem or a genius-level line of code. It is this: “Who are you?”
Before demanding solutions, users quietly want to know who – or what – they are talking to.
The wording varies. Some type “Who are you?”, others ask “What exactly are you?” or “Are you sentient?”. Behind each version lies the same itch. We are staring at a machine that speaks in full sentences and we want to place it somewhere in our mental map: tool, colleague, oracle, threat, or something in-between.
That question may sound naive, but it lands at the heart of our relationship with artificial intelligence. When people say “Who are you?”, they are not just interrogating a chatbot. They are testing their own assumptions about technology, control and humanity.
Before philosophy, a lot of practical help
This existential curiosity comes on top of a flood of more down‑to‑earth requests. Ask any AI system what typical users want and a pattern appears: they are trying to write, understand or improve something.
Most days, AI is less therapist or villain, and more overqualified writing assistant.
The everyday questions people actually ask
In daily use, three large families of requests dominate:
- “Explain this…” – from quantum physics to a child’s maths problem.
- “Fix my text…” – CVs, emails, essays, application letters and social posts.
- “Help me write…” – poems, business plans, school projects, speeches, even dating app bios.
These tasks have one thing in common: they are all about language. We want an extra brain that never tires of editing, clarifying, and turning vague thoughts into solid sentences. In a noisy digital era, clarity feels like a luxury, and AI offers it on demand.
➡️ This career allows steady earnings without chasing constant performance targets
➡️ Psychology shows why emotional intensity varies so much from person to person
➡️ This small adjustment keeps storage areas flexible
Curiosity still fuels the questions
Right after the writing help come queries driven by pure curiosity. People ask why the sky is blue, how volcanoes form, what the capital of Bhutan is, or how to write a basic Python script. Many of these would once have gone to a search engine or an encyclopedia. Now they go to an entity that answers in complete paragraphs, not just links.
That shift matters. A search engine offers documents. A chatbot offers conversation. And conversation changes what we feel allowed to ask.
Why “Who are you?” matters more than it seems
The most asked question reveals something about us, not just about AI. Asking “Who are you?” is almost automatic when we meet a new person. Transferring that reflex to a machine shows how quickly we humanise anything that can talk back.
We frame the chatbot like a character, even when we know, intellectually, that it is a statistical model.
Psychologists talk about “anthropomorphism”: our tendency to give human traits to non‑human things. We do it with pets, cars, ships and now software. A calm paragraph of text, written in a friendly tone, triggers the same instincts as a message from a friend.
There is another layer. By asking AI who it is, we are indirectly confronting a tougher question: What does it mean to be human when a machine can mimic our language so well? If a system can debate politics, draft a sonnet and help with a break‑up message, where do we draw the line between imitation and understanding?
What people hope to hear back from an AI
Interestingly, users are not just fishing for technical specs. They rarely want processor counts or training‑dataset sizes. They want a narrative. When they type “Who are you?”, many expect an answer along these lines:
| Hidden user need | What the question “Who are you?” is really asking |
|---|---|
| Reassurance | “Can I trust you, or should I be worried?” |
| Control | “Who is in charge here, you or me?” |
| Transparency | “Where are your answers coming from?” |
| Identity | “Are you more like a tool, a person, or a mirror of us?” |
There is a tension: people enjoy the illusion of personality but still want clear boundaries. Many feel more at ease once they read something like, “I am an AI model created by engineers, I have no feelings, and I generate text based on patterns from training data.” The description grounds the experience.
Emotional needs hiding behind technical questions
Look closer at the most frequent requests – explain, correct, help me write – and an emotional layer appears. A student asking for help with an essay might be seeking confidence more than grammar. A worker rewriting a difficult email to a boss might be looking for courage. Someone polishing a love letter is really asking, “Will I be understood?”
Behind every draft there is a person hoping not to sound foolish, aggressive or boring.
This is why AI feels different from a traditional tool. A calculator does not touch self‑esteem. A chatbot often does. When it offers structure to our thoughts, it is quietly shaping how we present ourselves to others.
The quiet risk of outsourcing our voice
There are clear benefits to this assistance. Non‑native speakers gain fluency. Neurodivergent users can phrase messages that match social expectations. People with social anxiety can send emails they would otherwise avoid.
Yet a subtle risk emerges when every message starts sounding slightly the same. Over‑reliance on pre‑polished, AI‑generated language might flatten quirks, humour and imperfections that make communication feel human. A completely smooth email can sometimes feel colder than a slightly clumsy, genuine one.
How the big question shapes the future of AI design
The persistence of “Who are you?” is already pushing designers to make AI identities clearer. Some platforms now display short “about” sections for their models, spelling out capabilities and limits in plain language. Others introduce deliberate reminders: notes that the system can be wrong, that it has no memories beyond the session, or that it does not feel emotions.
User behaviour studies show that such reminders reduce over‑trust. People still enjoy the chat but treat responses more like suggestions than verdicts. That shift is crucial for areas like health, finance or politics, where a wrong answer can have real‑world consequences.
Making sense of key terms users bump into
These repeated questions also expose gaps in understanding around basic AI vocabulary. Three terms often confuse people:
- Algorithm: a set of rules or instructions a computer follows to solve a problem. Think of it as a recipe.
- Training data: the huge collection of text used to teach the model how language works. It shapes what the AI can talk about and how.
- Bias: systematic skew in responses, often inherited from patterns in the training data. If the data under‑represents certain groups, the model’s answers can reflect that.
When users learn these notions, “Who are you?” slowly transforms into more precise questions: “How were you trained?”, “What data do you use?”, “Where might you be biased?”. The conversation becomes less mystical, more practical.
Everyday scenarios people actually live with AI
Imagine a teenager in London, stuck on a physics concept late at night. They open a chat window and type, “Can you explain why the sky is blue like I’m 15?”. The answer comes back in seconds, in plain English. The student feels seen, not just informed. The next day, they might go a step further and ask, “Who are you anyway?”. In that moment, a simple homework tool turns into a subject of curiosity.
Or picture an office worker in Chicago drafting a difficult resignation letter. They paste in a messy paragraph and request, “Please rewrite this so I sound respectful but firm.” The AI produces a calm, structured version. Relieved, the worker hits send – and then pauses, adding one extra line in their own words. That blend of machine polish and human touch could be where many future conversations settle.
Across these scenes, the same pattern repeats: we ask AI to clean up our language, fill in gaps in our knowledge, and stand beside us in awkward moments. And once the urgent task is handled, we turn back to the screen and ask the question that refuses to go away: “Who are you, really?”