Physicist Ido Kanter Reveals Why AI Follows "More is Different" Rule While Traditional Physics Remains "More is the Same"

Bar-Ilan University research shows that AI intelligence emerges from specialized nodal cooperation, proving that more is truly different in machine learning.

By: AXL Media

Published: Apr 3, 2026, 10:43 AM EDT

Source: Information for this report was sourced from Bar-Ilan University

Revisiting a Nobel-Winning Philosophy

In 1972, Nobel Prize winning physicist Philip W. Anderson introduced the concept "More is Different," a philosophical argument against pure reductionism. Anderson posited that as systems grow in complexity, they develop emergent properties that cannot be predicted by simply studying their individual elementary particles. While this viewpoint has influenced chemistry and biology for decades, it was conceived long before the era of deep learning. Now, Professor Ido Kanter of Bar-Ilan University has published a study in Physica A that quantitatively examines this philosophy in the context of Artificial Intelligence, revealing a fundamental divergence between the laws of physics and the architecture of machine learning.

Physics as "More is the Same"

Professor Kanter’s research identifies a key information-theoretic distinction: most physical systems actually embody the principle of "More is the Same." In many physical contexts, individual components reflect the same general information about the state of the system as the whole. Adding more components to a measurement often results in redundant data rather than an increase in the total unique information available. From an information viewpoint, the system’s state is distributed such that larger scales do not necessarily manifest the "different" emergent behaviors seen in biological or social structures, at least not in the same way intelligent systems do.

The Emergence of Nodal Specialization in AI

In contrast, the study finds that AI models represent a true embodiment of "More is Different." As an AI model undergoes training, its internal units—known as nodes—do not remain identical or redundant. Instead, they undergo a process of intense specialization. One node might become an expert in recognizing specific visual patterns, while another focuses on nuanced linguistic features. This division of labor means that the capability of the collective system far exceeds the sum of its parts. Because each node contains meaningful, unique information about the overall task, the interaction between these specialized components creates the "emergent intelligence" that defines modern AI.