Six Load-Bearing Assumptions: An Assumption Audit of Classical Organization and Process Models Under AI

Practice has issued a verdict on the classical canon of organization and process models: that artificial intelligence makes much of it obsolete. The verdict has not been tested. This note proposes and demonstrates a way of testing it. Rather than asking which models are "outdated", it reduces each model to the single assumption without which it does not function, and asks whether AI leaves that assumption intact. Six such load-bearing assumptions carry most of the canon. A model does not fail as a whole; it fails at one point, and it matters greatly which point. I therefore distinguish three kinds of failure: a world condition ceases to hold (W), a substantive claim of the theory becomes false (T), or a design recommendation loses its justification while the framework itself survives (E). Three demonstration cases show the audit at work - Galbraith's information-processing view (E), statistical process control and Six Sigma applied to AI-produced output (T/E), and structural contingency theory, which holds. Across the cases a survival criterion emerges: the models that survive are those formulated functionally rather than personally - they specify which function must be performed, not who performs it. The note claims novelty only for the audit format, the failure typology, and the criterion; every individual verdict has antecedents, which are cited.