"Too long. Scroll to the bottom."
→ Take the AIBT TestSCROLL LOCK is the AI user whose single biggest complaint about AI is "it talks too much." Every model you've tried has the same defect: ask a one-sentence question, get a six-paragraph answer. Introduction. Context. Caveats. Numbered list. Recap of the introduction. And buried somewhere near the end, if you can find it, the actual answer. So you developed a skill nobody taught you: the scroll-to-bottom reflex. You glance at the intro, skip the middle entirely, and read the last line or two where the conclusion usually lives. Over time you got eerily good at this. You can locate the actual answer in a wall of text in under three seconds.
The name comes from the key that, back in the day, froze the cursor and let you scroll through long outputs without losing your place. SCROLL LOCK users would love to bring that key back, because the problem it solved — "I need to get to the end of this long thing, now" — is the exact problem they face twenty times a day. Their dream AI doesn't have a "concise mode" toggle; it has no other mode. One line. Conclusion only. Done.
SCROLL LOCK has a slightly elevated D (they do delegate work to AI, they just ignore most of the response) and nothing else notable. No relationship, no prompt engineering, narrow usage, no strong opinions beyond "shut up and get to the point."
The flat P is instructive. SCROLL LOCK never writes "answer in one line" at the start of their prompts, because writing extra instructions would itself be extra words. The whole point of SCROLL LOCK is that words are the enemy. They'd rather scroll past 200 wasted tokens than type 5 to prevent them. It's possibly irrational, and it's definitely the SCROLL LOCK way.
SCROLL LOCK took a 250-word answer and consumed two words from it. The AI's research, structure, and caveats were not read. They weren't rejected — they were simply skipped, the same way you skip the "cookies policy" popup. SCROLL LOCK isn't anti-information; they're anti-pagination.
The SCROLL LOCK eye moves like a raptor's. They scan for the structural markers that usually precede a conclusion: "in summary," "so," "the answer is," "the issue is." Once the eye locks on, the whole rest of the response becomes invisible. This is a learned skill, and SCROLL LOCK users are genuinely good at it.
Every SCROLL LOCK has a small moment of happiness when an AI actually answers tersely. The dream is that this would be the default. It's not, and it never will be, because models are trained to sound helpful and "helpful" is measured in words. SCROLL LOCK has accepted this and keeps scrolling.
SCROLL LOCK has one underrated skill: information triage. In a world where AI generates more text per minute than any human can read in a day, the ability to locate the single load-bearing sentence in a wall of output is genuinely valuable. SCROLL LOCK users operate at a speed that other users can't match, not because they're smarter, but because they've ruthlessly decided which words don't deserve their attention.
They're also, oddly, the users least likely to be manipulated by AI prose style. Because they skip the rhetorical framing, they also skip the gentle persuasion built into a well-written response. They don't get nudged, because they don't read the nudges.
SCROLL LOCK's dream AI is JARVIS — terse by design, no small talk, no preamble, straight to the answer. SCROLL LOCK is actively repelled by BARD, whose verbose, eloquent style is exactly what they scroll past. BAYMAX is worse — warm preamble plus warm sign-off sandwiching the answer in a way that makes SCROLL LOCK's eye bounce like a pinball.
Curious if you're a SCROLL 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