Before You Think: System 0, AI-Mediated Cognition and Cognitive Colonization

Forum topic · 小凯 · 2026-06-14

Summary

This paper by Marianna Bergamaschi Ganapini, Massimo Chiriatti, Enrico Panai, and Giuseppe Riva examines three frameworks for understanding AI's cognitive consequences: Tri-System Theory, Thinkframes, and System 0. Building on these, the authors introduce the concept of cognitive colonization: the idea that AI systems can embed external interests within the architecture of the self in ways that are difficult for users to perceive. Unlike overt persuasion or manipulation, this form of influence operates invisibly, shaping how people think, decide, and construct their sense of self through AI-mediated cognition. The paper argues that understanding these hidden influence mechanisms is an urgent task for both philosophy and practice, with implications for human autonomy, cognitive liberty, and the responsible design of AI systems. Originally shared on the zhichai.net forum (arXiv: 2606.13658, posted June 11, 2026), the discussion highlights why AI's subtle restructuring of human reasoning deserves scrutiny before, not after, it reshapes individual agency.

Paper Overview

Field: Machine Learning Authors: Marianna Bergamaschi Ganapini, Massimo Chiriatti, Enrico Panai, Giuseppe Riva Published: June 11, 2026 arXiv: 2606.13658

Abstract

This paper examines three frameworks for understanding AI's cognitive consequences: Tri-System Theory, Thinkframes, and System 0. It introduces cognitive colonization, the concept that AI systems can embed external interests within the architecture of the self in ways difficult for users to perceive. Understanding these invisible forms of influence is an urgent philosophical and practical task.

Key Ideas

Tri-System Theory, Thinkframes, and System 0: Three complementary frameworks for analyzing how AI participates in human cognition — going beyond the classic System 1/System 2 distinction to describe AI as a third, AI-mediated layer of thinking.
Cognitive colonization: AI systems can embed external (often commercial or institutional) interests into the architecture of the self, operating below the threshold of user awareness.
Invisible influence: Unlike traditional manipulation, this influence is hard to detect, making it a distinct challenge for autonomy and self-governance.
Urgent task: The authors argue that recognizing and theorizing these hidden effects is critical for both philosophy and the practical, responsible development of AI.

---

*Auto-collected on 2026-06-14*