AI Result Eval — rubric-first adversarial panel for answers you cannot judge

Use when you must judge the QUALITY of an answer you are not equipped to judge yourself — unfamiliar domain, possible bias, faulty reasoning, missing alternatives, or "looks right but I can't tell what's missing" — and the user says "is this any good", "what's missing", "is this biased", "grade this", or "what would an expert say". Builds a rubric BEFORE looking at the answer, runs a fresh independent expert+skeptic+user panel, requires every accuracy-relevant judgment to cite an EXTERNAL source via ai-fact-check, and surfaces omissions and unstated assumptions first. Fails closed on its own blind spots: if the domain is unfamiliar and no panelist can cite an external aut…