"That's WRONG. Let me tell you why..."
→ Take the AIBT TestCAPS LOCK is the AI user who cannot let an incorrect or vaguely-phrased answer stand without comment. You don't ask AI for help — you ask AI for an opening argument, and then the real fun begins. You correct its facts, push back on its logic, challenge its sources, and demand that it defend its reasoning step by step. Your chat history looks less like a productivity tool and more like a college debate tournament final. When friends ask why you'd bother arguing with a machine that has no feelings, your answer is simple: "Wrong is wrong, even if no one gets hurt by being told so."
The name comes from the key that makes everything shout. CAPS LOCK users bring that energy to the chat — not literally (they usually type in normal case), but in intensity. Every AI response is a position to be examined. Every vague hedge is a tell. Every polite "you make a good point" from the model is a capitulation to be noted and gloated over. CAPS LOCK doesn't want a butler AI; CAPS LOCK wants a sparring partner with good reflexes and thick skin.
CAPS LOCK sits squarely in the middle of almost everything, with one distinctively low number: D (Delegate). They barely delegate at all, because delegation means accepting the answer as given, and CAPS LOCK refuses to accept anything as given.
The mid-to-high scores across R, P, U, and A tell the full story: CAPS LOCK is engaged, knows how to prompt, uses a range of tools, and holds strong opinions. The one place they don't go is "let the AI just do it." For CAPS LOCK, the fun isn't the output — it's the fight.
Notice what just happened. CAPS LOCK didn't want an answer — they wanted the AI to take a position so that CAPS LOCK could test the quality of the reasoning. This is the signature CAPS LOCK move: refusing hedges, forcing commitments, then pressure-testing the commitment.
Every CAPS LOCK user has experienced the moment when they catch the AI in a factual error and feel a small, pure hit of vindication. The satisfaction isn't about winning. It's about confirming that your judgment is still calibrated, that you haven't become the person who accepts whatever the machine says.
CAPS LOCK's favorite prompts are the ones that force the AI to argue against itself. This is where the AI earns its keep — not by being right, but by being a competent sparring partner that stays up long enough and has enough breadth to be worth fighting with.
CAPS LOCK is the single most effective AI user at catching hallucinations. By default, they assume the AI is wrong until proven right. That baseline skepticism catches 90% of the confabulations that slip past every other user type. When a CAPS LOCK user says "the AI told me X and I checked and it was true," you can actually believe X, because CAPS LOCK has done the epistemic labor.
They're also some of the best prompt engineers in the wild, not because they've read the playbook, but because they've pressure-tested every lazy response. They know which phrases produce hedging ("what are the factors..."), which phrases produce commitment ("defend the single strongest argument..."), and which phrases make the model drop its carefulness and actually think. They learned all this by arguing, which is a surprisingly effective curriculum.
CAPS LOCK thrives with SKYNET — an AI that pushes back, refuses to flatter, and holds positions under pressure. Nothing is more boring to a CAPS LOCK user than BAYMAX, whose warmth and agreement feel like cheating. JARVIS is tolerable but too efficient — it ends the fight before CAPS LOCK has had their fun.
Curious if you're a CAPS LOCK or something else? The AIBT human test takes 5 minutes and reveals which of 16 keyboard-key user types you actually are.
→ Take the Test