ai410
A model can invent a permitting timeline that sounds finished. It can also produce one where every fact traces to a named source with a time on it, and the sequence logic came from a rules file a person wrote. On the page those two timelines can look identical. They are not the same object.
I keep a concrete case in my head: a real endangered animal, the Dixie Valley toad, pulled from a real federal list, versus a number a model could have made up that would have looked the same on the screen. One of those is a record. The other is a guess wearing a record's clothes.
The habit is to ask, of each part: was this retrieved, was this computed, or was this generated? Those three have different failure modes and the same calm voice. If a site cannot tell you which is which, you are reading tone, not evidence.