Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Semantic Communication meets System 2 ML: How Abstraction, Compositionality and Emergent Languages Shape Intelligence

Mehdi Bennis, Salem Lahlou

TL;DR

This paper argues that 6G and AI progress require a fundamental shift from purely data-centric communication to semantic, reasoning-driven systems grounded in System-2 cognition. It proposes a unified framework that blends abstraction, algebraic compositionality, and emergent communication, supported by world models, probabilistic/programmatic representations, and mathematical formalisms (e.g., sheaf theory, category theory). Key contributions include a delineation of three research pillars, a set of concrete desiderata and research questions, and a pathway toward emergent, verifiable semantic communication protocols that enhance adaptability, efficiency, and resilience. The work aims to unify wireless, ML, and robotics advances under a knowledge-centric paradigm, with potential for significant gains in bandwidth efficiency, robustness to distribution shifts, and human-machine collaboration in future networks.

Abstract

The trajectories of 6G and AI are set for a creative collision. However, current visions for 6G remain largely incremental evolutions of 5G, while progress in AI is hampered by brittle, data-hungry models that lack robust reasoning capabilities. This paper argues for a foundational paradigm shift, moving beyond the purely technical level of communication toward systems capable of semantic understanding and effective, goal-oriented interaction. We propose a unified research vision rooted in the principles of System-2 cognition, built upon three pillars: Abstraction, enabling agents to learn meaningful world models from raw sensorimotor data; Compositionality, providing the algebraic tools to combine learned concepts and subsystems; and Emergent Communication, allowing intelligent agents to create their own adaptive and grounded languages. By integrating these principles, we lay the groundwork for truly intelligent systems that can reason, adapt, and collaborate, unifying advances in wireless communications, machine learning, and robotics under a single coherent framework.

Semantic Communication meets System 2 ML: How Abstraction, Compositionality and Emergent Languages Shape Intelligence

TL;DR

This paper argues that 6G and AI progress require a fundamental shift from purely data-centric communication to semantic, reasoning-driven systems grounded in System-2 cognition. It proposes a unified framework that blends abstraction, algebraic compositionality, and emergent communication, supported by world models, probabilistic/programmatic representations, and mathematical formalisms (e.g., sheaf theory, category theory). Key contributions include a delineation of three research pillars, a set of concrete desiderata and research questions, and a pathway toward emergent, verifiable semantic communication protocols that enhance adaptability, efficiency, and resilience. The work aims to unify wireless, ML, and robotics advances under a knowledge-centric paradigm, with potential for significant gains in bandwidth efficiency, robustness to distribution shifts, and human-machine collaboration in future networks.

Abstract

The trajectories of 6G and AI are set for a creative collision. However, current visions for 6G remain largely incremental evolutions of 5G, while progress in AI is hampered by brittle, data-hungry models that lack robust reasoning capabilities. This paper argues for a foundational paradigm shift, moving beyond the purely technical level of communication toward systems capable of semantic understanding and effective, goal-oriented interaction. We propose a unified research vision rooted in the principles of System-2 cognition, built upon three pillars: Abstraction, enabling agents to learn meaningful world models from raw sensorimotor data; Compositionality, providing the algebraic tools to combine learned concepts and subsystems; and Emergent Communication, allowing intelligent agents to create their own adaptive and grounded languages. By integrating these principles, we lay the groundwork for truly intelligent systems that can reason, adapt, and collaborate, unifying advances in wireless communications, machine learning, and robotics under a single coherent framework.

Paper Structure

This paper contains 28 sections, 9 figures, 1 table.

Table of Contents

  1. The Wireless Communication World
  2. Information: The Semantic Chamaeleon
  3. The AI/ML World
  4. World Models and Inference for Reasoning
  5. Closing Loops Compositionally: Semantic Communication meets System 2 ML
  6. Research Questions
  7. Key Desiderata
  8. Abstraction, Anticipation, Adaptation
  9. Algebraic Compositionality
  10. Emergent Communication and Languages
  11. Fundamentals of System 1 and System 2 Cognition
  12. Dual-Process Theory of Cognition
  13. Relevance to Communication Systems and AI
  14. Implementation Challenges
  15. Topological Information: A Primer
  16. ...and 13 more sections

Figures (9)

  • Figure 1: Shannon's three levels of communication shannon1948mathematical.
  • Figure 2: Various shades of information.
  • Figure 3: Daniel Kahneman (System 1 vs. System 2) kahneman2011thinking, Michael S.A. Graziano (world model and metacognition), Marvin Minsky (emergence) minsky1986society and Jeff Hawkins (a thousand brains hypothesis) hawkins2021thousand.
  • Figure 4: Confluence of abstraction, algebraic compositionality and semantic communication. Here, the semantics of active inference control-loops that sense, plan and adapt are composed.
  • Figure 5: Three Key Pillars underlying the proposed vision.
  • ...and 4 more figures