Saved in:
Bibliographic Details
Main Authors: Parfenova, Angelina, Denzler, Alexander, Pfeffer, Juergen
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.00047
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914174445027328
author Parfenova, Angelina
Denzler, Alexander
Pfeffer, Juergen
author_facet Parfenova, Angelina
Denzler, Alexander
Pfeffer, Juergen
contents Large language models (LLMs) are increasingly deployed in collaborative settings, yet little is known about how they coordinate when treated as black-box agents. We simulate 7500 multi-agent, multi-round discussions in an inductive coding task, generating over 125000 utterances that capture both final annotations and their interactional histories. We introduce process-level metrics: code stability, semantic self-consistency, and lexical confidence alongside sentiment and convergence measures, to track coordination dynamics. To probe deeper alignment signals, we analyze the evolving geometry of output embeddings, showing that intrinsic dimensionality declines over rounds, suggesting semantic compression. The results reveal that LLM groups converge lexically and semantically, develop asymmetric influence patterns, and exhibit negotiation-like behaviors despite the absence of explicit role prompting. This work demonstrates how black-box interaction analysis can surface emergent coordination strategies, offering a scalable complement to internal probe-based interpretability methods.
format Preprint
id arxiv_https___arxiv_org_abs_2512_00047
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Emergent Convergence in Multi-Agent LLM Annotation
Parfenova, Angelina
Denzler, Alexander
Pfeffer, Juergen
Computation and Language
Artificial Intelligence
Large language models (LLMs) are increasingly deployed in collaborative settings, yet little is known about how they coordinate when treated as black-box agents. We simulate 7500 multi-agent, multi-round discussions in an inductive coding task, generating over 125000 utterances that capture both final annotations and their interactional histories. We introduce process-level metrics: code stability, semantic self-consistency, and lexical confidence alongside sentiment and convergence measures, to track coordination dynamics. To probe deeper alignment signals, we analyze the evolving geometry of output embeddings, showing that intrinsic dimensionality declines over rounds, suggesting semantic compression. The results reveal that LLM groups converge lexically and semantically, develop asymmetric influence patterns, and exhibit negotiation-like behaviors despite the absence of explicit role prompting. This work demonstrates how black-box interaction analysis can surface emergent coordination strategies, offering a scalable complement to internal probe-based interpretability methods.
title Emergent Convergence in Multi-Agent LLM Annotation
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2512.00047