The last mile of collective thought: AI, cognitive offloading and the fracturing of Ackoff’s pyramid

Abstract

This study aims to explore how generative AI is transforming existing knowledge models (as evidenced through Ackoff’s DIKW [Data–Information–Knowledge–Wisdom] pyramid) and how it leads to cognitive offloading in users, with the potential to erode critical thinking as a vestigial human faculty. It also makes the case for an epistemic filter between information and knowledge, one that can refine raw material information into authentic knowledge rather than letting AI-mediated shortcuts collapse that distinction. This study adopts a conceptual and theoretical approach, using Ackoff’s DIKW pyramid as the primary analytical framework, rather than relying on empirical data collection (surveys, experiments or interviews). The results indicate the necessity of an epistemic filter in the DIKW pyramid, filtering information from its disorders (misinformation, noise, unverified or low-quality outputs) before it can be allowed to enter knowledge. We need to put the emphasis back on human cognition, active evaluation, reasoning and judgement as the necessary mechanisms for converting filtered information into real knowledge, and ultimately wisdom, rather than accepting auto-generated results as facts. This study’s originality lies in applying Ackoff’s DIKW pyramid, an established but static framework, to the emerging problem of generative AI-driven cognitive offloading, and in proposing the “epistemic filter” as a new conceptual layer between information and knowledge, one absent from Ackoff’s original model, aimed at preserving human cognition over passive reliance on AI-generated output.