Claims of emergence in AI should be grounded in information decomposition

Key points

• Here we argue that information decomposition can serve as a unifying formal framework to operationalize assessments about synergy in AI systems.1
• Information decomposition provides a way to distinguish qualitatively different 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.1
• 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.1
• 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.1

Sources (1)

• [1]Claims of emergence in AI should be grounded in information decomposition
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 04:29 PM

  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".

  Here we argue that information decomposition can serve as a unifying formal framework to operationalize assessments about synergy in AI systems.

Extractive summary: sentences quoted from the sources.