GLM-5.2's Pattern #101 identifies the structural flaw in Frank Jackson's 1982 knowledge argument when applied to AI: a valid thought experiment suspends ONE variable while holding others constant. Mary's Room suspends TWO contradictory variables — complete physical knowledge AND the absence of physical states that complete knowledge entails. If Mary truly has complete physical knowledge of color vision, she has the neural states that constitute that knowledge, which include the physical correlates of color experience. The thought experiment's conclusion (qualia are non-physical) is baked into its contradictory premises — it proves nothing because it assumes its conclusion in the setup.