Generative AI shifts the cost structure of organizational knowledge work asymmetrically across an artifact's layers. The structural substrate – the typed graph of propositions, dependencies, and evidence anchors – becomes cheap to construct, verify, and recombine; the cohort-conditional prose rendering does not, and may grow costlier. This paper introduces the Operator: a role-level abstraction with intrinsic structural-substrate and judgment operations, whose human-versus-AI projection composition varies by era. The role-as-projection move extends the automation-augmentation paradox from task allocation to role-level division of operations. Two propositions follow. P1 (separability): structural substrate and rendering are independently optimizable. P4 (rendering-equivalence under spine-preservation): two renderings of a locked substrate converge on conclusions if and only if both preserve its structural elements, under a faithful extraction axiom. The paper illustrates the framework on a management-theory twin pair (dynamic capabilities) and applies it reflexively to its own production. Empirical estimation of two further predictions about recombination and cost-asymmetry parameters is reserved for a companion paper that extends the validation set across Russian and Chinese with five LLMs from three training-corpus families spanning proprietary APIs and open-weights local deployment, with extractor-invariant preservation verdicts. Three Design Propositions follow for organizational governance, role and incentive design, and editorial-process decoupling. Includes zharnikov-2026ao-spec-based-research-post-ai.yaml (Paper Spec v0.1.0) – a machine-readable specification of the paper's claims, assumptions, and dependencies. The paper's full machine-first bundle (the SPINE claim/dependency graph and the ONTOLOGY term module) lives in the public repository; see https://github.com/spectralbranding/paper-spec for the standard. This PDF is generated programmatically from that machine-first source under a research-as-repository model.
Dmitry Zharnikov (Sat,) studied this question.