Hallucination
Level 1
When a model states something false with complete confidence.
Term 6 of 8 in Language models
In plain language
Because a model is predicting plausible text rather than retrieving verified facts, it can invent a citation, a date or a quote that simply does not exist. It has no internal signal that it is guessing.
Think of it like this
Like a very well-read person who would rather give a confident answer than admit they cannot remember.
Why it matters
Hallucination is the single most important reason to verify AI output before you act on it.
Next terms in Language models
- Large language modelA model trained on enormous amounts of text to predict what words come next.
- TokenA small chunk of text that a model reads and writes, usually part of a word.
- PromptThe instruction and information you give a model.
- Context windowThe maximum amount of text a model can consider at one time.