What question did this study set out to answer?

The paper aims to redefine LLM capabilities as a product of collective human cognition rather than autonomous reasoning by machines.

April 12, 2026Open Access

The Intelligence That Was Never Artificial: LLMs as Collective Human Cognition and the Cybernetics That Predicted Them

Key Points

The paper aims to redefine LLM capabilities as a product of collective human cognition rather than autonomous reasoning by machines.
Applied mathematical identities to analyze LLM frameworks.
Integrated the Varela enactive cognition framework.
Conducted systematic fact-checking on 119 references and corrected 14 citations.
Demonstrated that cross-entropy pretraining aligns with the linear opinion pool framework.
Showed that RLHF (reinforcement learning from human feedback) corresponds to Borda count and logarithmic opinion pooling.
Highlighted significant integrations like the Diversity Prediction Theorem and Conditional Jury Theorem.

Abstract

We argue that LLM capabilities are best understood as structured aggregation of collective human intelligence, not autonomous machine reasoning. The framework is grounded in proven mathematical identities (cross-entropy pretraining implements the linear opinion pool; RLHF implements Borda count and logarithmic opinion pooling), the Diversity Prediction Theorem, and the Conditional Jury Theorem. v3 changes: 14 citation corrections verified across all 119 references via systematic fact-checking, Varela enactive cognition framework integrated, one fabricated reference replaced.

The Intelligence That Was Never Artificial: LLMs as Collective Human Cognition and the Cybernetics That Predicted Them

Key Points

Abstract

Cite This Study