In Dialogue with Intelligence: Rethinking Large Language Models as Collective Knowledge

TL;DR

The paper reframes Large Language Models as Collective Knowledge—condensed human cultural output that gains apparent intelligence through dialogue. It argues that CK behaves as a dynamic representation with no persistent self and that recurrence is offloaded to the dialogue loop, while modes emerge from distinct subnetworks. The authors introduce co-augmentation as a reciprocal human-AI loop that expands analytical capabilities, and discuss safety filters, fine-tuning, and the neuroscience lens as a path to modeling CK. By framing LLMs as interactively evoked systems, the work highlights practical implications for design, evaluation, and cross-disciplinary study of the human–CK dyad.

Abstract

Large Language Models (LLMs) can be understood as Collective Knowledge (CK): a condensation of human cultural and technical output, whose apparent intelligence emerges in dialogue. This perspective article, drawing on extended interaction with ChatGPT-4, postulates differential response modes that plausibly trace their origin to distinct model subnetworks. It argues that CK has no persistent internal state or ``spine'': it drifts, it complies, and its behaviour is shaped by the user and by fine-tuning. It develops the notion of co-augmentation, in which human judgement and CK's representational reach jointly produce forms of analysis that neither could generate alone. Finally, it suggests that CK offers a tractable object for neuroscience: unlike biological brains, these systems expose their architecture, training history, and activation dynamics, making the human--CK loop itself an experimental target.