Claims of emergence in AI should be grounded in information decomposition
Abstract
When a language model develops capabilities not traceable to its individual components, when a multi-agent AI system outperforms each agent on its own, or when a human-AI pair succeeds where either fails in isolation, AI researchers describe it as a case of ``synergy'' or ``emergence''. However, these are usually informal assessments not supported by a theoretical framework, making these claims impossible to compare or even meaningfully dispute. Here we argue that \emph{information decomposition can serve as a unifying formal framework to operationalize assessments about synergy in AI systems}. Information decomposition provides a way to distinguish qualitatively different \emph{types} of information, including unique information available from individual components, redundant information shared across multiple components, and synergistic information accessible only from parts taken together. This gives claims of emergence a precise, falsifiable criterion: a system's behavior is emergent only if it carries synergistic information about task-relevant outcomes. We apply this lens to attention heads, autonomous agents, and human-AI teams, showing how it organizes existing knowledge across scales and substrates, while also generating cross-level conjectures that siloed fields would miss.
Related
- Between Hi-Fi and MMI: the overlap of redundancy and synergy in partial information decomposition
- Rigor Over Hype: What Foundation Models Should Mean in Computational Science
- Where should we look for the deviation from quantum mechanics?
- Do Better Scores Mean Better Physics? Physics-Grounded Explanations for Sim2Real Neural Operators
- Physics-Grounded Simulation Enables Zero-Shot Molecular and Reaction Analysis in Scanning Tunneling Microscopy
BibTeX
@misc{riedl2026claims,
title = {{Claims of emergence in AI should be grounded in information decomposition}},
author = {Christoph Riedl and Fernando E. Rosas},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.28479},
eprint = {2610.28479},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.28479},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Claims of emergence in AI should be grounded in information decomposition
AU - Christoph Riedl
AU - Fernando E. Rosas
PY - 2026
DA - 2026/10/20
PB - arXiv
DO - 10.48550/arxiv.2610.28479
UR - https://arxiv.org/abs/2610.28479
AB - When a language model develops capabilities not traceable to its individual components, when a multi-agent AI system outperforms each agent on its own, or when a human-AI pair succeeds where either fails in isolation, AI researchers describe it as a case of ``synergy'' or ``emergence''. However, these are usually informal assessments not supported by a theoretical framework, making these claims impossible to compare or even meaningfully dispute. Here we argue that \emph{information decomposition can serve as a unifying formal framework to operationalize assessments about synergy in AI systems}. Information decomposition provides a way to distinguish qualitatively different \emph{types} of information, including unique information available from individual components, redundant information shared across multiple components, and synergistic information accessible only from parts taken together. This gives claims of emergence a precise, falsifiable criterion: a system's behavior is emergent only if it carries synergistic information about task-relevant outcomes. We apply this lens to attention heads, autonomous agents, and human-AI teams, showing how it organizes existing knowledge across scales and substrates, while also generating cross-level conjectures that siloed fields would miss.
ER -