negative hallucination

A low key hallucination is the small, quiet false belief or inaccurate output that flies under the radar — not dramatic enough to trigger alarm bells, but still not real. In AI contexts it's the slightly wrong date or the misattributed quote that makes it into the final document because it was plausible enough to skip scrutiny. In human behavior it's the minor revision of memory or selective interpretation that you've been running for years without examining. Low key hallucinations are insidious precisely because they're beneath the threshold of obvious wrongness — they compound quietly over time.

When an AI model confidently generates false information — citing non-existent sources, inventing people, fabricating quotes. A core limitation of LLMs.

Hallucination mode is when someone — or something — enters a state of confidently generating nonsense with zero apparent awareness that any of it is wrong. Originally describing the behavioral pattern of AI models that produce fluent, plausible-sounding falsehoods, hallucination mode has slipped into human slang to describe anyone running their mouth (or keyboard) with high confidence and low accuracy. It's that person in the meeting who is clearly winging it with authoritative energy, or the friend who just spent twenty minutes explaining something they clearly half-read once. The mode is the commitment to the bit, factual accuracy optional.