This conceptual preprint proposes the Layered Atlas of Intelligence as a claim-separation map for intelligence discourse. Its central claim is that intelligence is confusing because the word is being used for many different things. The Atlas does not define intelligence; it separates the different kinds of claims hidden inside the word, including claims about Power, Quality, Aim, mechanisms, substrate, context, consciousness, values/wisdom, and domain/application. The paper introduces Intelligence Equivocation: the collapse of different claim-types into a single high-force word. It argues that useful constructs become socially and epistemically dangerous when they exceed their evidential scope. IQ can be useful without becoming destiny. AI can be powerful without being misnamed as wisdom. The Atlas is framed as a conceptual map, not as a psychometric instrument, cognitive ontology, or validated empirical tool. Its purpose is to make intelligence claims visible, bounded, contestable, and auditable. A companion preprint, Cognitive Aim and the Wrong-Layer Problem, develops the Cognitive Aim component in more detail.