A philosopher observing a duck grooming its feathers confronts a deceptively simple question: does the behaviour express agency or merely biological maintenance? The vignette illustrates a structural problem that recurs across biological, institutional, and computational systems. Many systems display learning and optimisation, yet these capacities alone do not constitute agency. This article develops a distinction between epistemic maintenance systems, which stabilise behaviour within inherited normative standards, and generative epistemic agents, which possess the capacity to reconstruct those standards when epistemic frameworks become unstable. Drawing on Kantian accounts of normative authorship and debates in the philosophy of technology, the article argues that epistemic responsibility arises only where agents participate in determining the standards governing their cognition. Three forms of generativity—procedural, epistemic, and institutional—are distinguished, and epistemic systems are classified along two structural axes that yield four configurations with distinct governance implications. The analysis extends to distributed epistemic infrastructures in which authority may become embedded in systems whose governing standards lack identifiable authorship. Contemporary “generative AI” systems exemplify procedural rather than epistemic generativity. The article concludes by proposing a generativity criterion for evaluating the governance of epistemic infrastructures in algorithmically mediated knowledge environments.
Building similarity graph...
Analyzing shared references across papers
Loading...
Peter Kahl
Lexicon Pharmaceuticals (United States)
Lexmark (United States)
Building similarity graph...
Analyzing shared references across papers
Loading...
Peter Kahl (Sun,) studied this question.
www.synapsesocial.com/papers/69b8f162deb47d591b8c655f — DOI: https://doi.org/10.5281/zenodo.19030057