The Generative Capacity Erosion Hypothesis: Does Generative AI Assistance Degrade Human Ideation Capacity?

Abstract

The rapid integration of generative artificial intelligence into creative workflows raises a critical but underexplored question: does habitual AI assistance in creative tasks lead to measurable atrophy in human generative capacities? Drawing from established research on cognitive offloading—including the “Google effect” on memory and GPS-induced degradation of spatial cognition—this paper introduces the Generative Capacity Erosion Hypothesis (GCEH): sustained reliance on AI co-creation tools may systematically weaken human ideation, divergent thinking, and creative self-efficacy. The hypothesis is situated within five decades of research on technology-mediated cognitive change, tracing parallels from calculators to search engines to satellite navigation. A theoretical framework distinguishes between augmentation effects (additive benefits) and atrophy effects (subtractive costs), arguing that current discourse overemphasizes the former while neglecting the latter. The paper develops operationalizable constructs for creative atrophy, outlines methodological approaches for empirical investigation, and discusses implications for the design of “creativity-preserving” AI systems. The Well-Being AI paradigm must account not only for what AI enables humans to produce, but for what sustained AI assistance may prevent humans from becoming.