Earned Standing, Not Fluent Appearance: An Anatomical Review of Generative AI as an Epistemic Technology and Its Consequences for the Ontology of Generative Artificial Intelligence

This review anatomises Baum, McCarthy and Hannigan's chapter on generative artificial intelligence (GenAI) as an epistemic technology. The chapter's central claim is that GenAI does not merely extend what organisations can know; it changes the conditions under which they can trust what they treat as true. Outputs lacking Toulminian warrant are miscategorised as warranted knowledge because fluency, confidence and coherence mimic the ordinary cues of epistemic standing. The risk is dyadic at the point of use and network-level once miscategorised claims travel, accumulate authority and reshape institutional memory. Using epistemic networks (epinets) and a GenAI-supported strategic-planning cycle, the authors map four work modes — authenticated, automated, augmented and autonomous — onto paired dyadic and network risks, and derive stage-specific stewardship levers. The review reconstructs this argument joint by joint, states its strengths and limits, and places it against twelve competing or adjacent models of organisational knowing. It then traces the ontological and epistemological verticals that the chapter opens, and asks what the argument does to the ontology of GenAI itself: neither a retrieving instrument nor a knowing agent, but a proposition engine whose standing must be earned inside an infrastructure. Review article. Dr. Syed Muntasir Mamun, ORCID 0000-0001-6845-2853.