The Appearance of Knowing: Why Epistemic Technologies Demand New Organization Theory

The familiar claim that generative AI drives the cost of cognition toward zero gets the problem backward. What AI cheapens is the appearance of knowledge: plausible claims that circulate as if warranted. Drawing on Epinets, this essay argues that large language models enter organizations as proposition-producing systems without the epistemic states their outputs normally imply: belief, justified knowledge, recognized ignorance, and accountability. That is the condition. The organizational danger begins when such systems are miscategorized as humanlike knowers. Their outputs can acquire epistemic authority before epistemic responsibility is established. Error becomes harder to detect because it arrives without the signals that normally make wrongness visible: hesitation, qualification, and acknowledged uncertainty. Correct and incorrect outputs can feel the same, making correction-seeking unlikely. The essay traces this progression from plausible output to miscategorization, authority capture, and epistemic misalignment, and then asks how organizations can govern epistemic standing in hybrid human–AI systems.