Technology Design for Epistemic Niche Construction in the AI Era

Generative AI is restructuring the epistemic environments in which humans think, learn, and produce knowledge. This paper argues that epistemic niche construction---the process by which human agents build and modify the representations, tools, practices, and values that sustain their knowledge work, and are cognitively reshaped by what they build---provides a productive framework for understanding this restructuring and for guiding technology design in education. To motivate the analysis, I identify three features that make generative AI qualitatively different from prior cognitive technologies as a modifier of epistemic niches: representational opacity, generativity, and substrate asymmetry. Each produces distinctive feedback dynamics that can either enrich or erode human epistemic niches. Drawing on emerging literature across contexts, I present different enrichment and erosion mechanisms and discuss implications for designing epistemic environments for learning and knowledge work in the AI era.