factual errors

A factual error is the simplest failure to picture: the model gets a checkable fact plainly wrong. It says a river runs through the wrong country, gives a famous equation the wrong constant, dates a war to the wrong decade, or mangles a number that anyone could look up. These are not subtle matters of opinion — there is a right answer, and the model missed it while sounding entirely sure.

Why these slip through is instructive. Facts that appeared rarely, inconsistently, or wrongly in the training text are weakly held; the model blends conflicting memories into an average that may be true of nothing. Numbers and proper names are especially fragile, because a single digit or letter changes the meaning yet barely changes how the sentence flows. The fix is rarely cleverer prompting — it is grounding the answer in a trusted source the model can quote from.