Communication Dans Un Congrès Année : 2025

MOSAIC: Multiple Observers Spotting AI Content

Résumé

The dissemination of Large Language Models (LLMs), trained at scale, and endowed with powerful text-generating abilities, has made it easier for all to produce harmful, toxic, faked or forged content. In response, various proposals have been made to automatically discriminate artificially generated from human-written texts, typically framing the problem as a binary classification problem. Early approaches evaluate an input document with a well-chosen detector LLM, assuming that low-perplexity scores reliably signal machine-made content. More recent systems instead consider two LLMs and compare their probability distributions over the document to further discriminate when perplexity alone cannot. However, using a fixed pair of models can induce brittleness in performance. We extend these approaches to the ensembling of several LLMs and derive a new, theoretically grounded approach to combine their respective strengths. Our experiments, using a variety of generator LLMs, suggest that this approach effectively harnesses each model’s capabilities, leading to strong detection performance on a variety of domains.

Fichier principal
Vignette du fichier
2025.findings-acl.1244.pdf (624.43 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05207044 , version 1 (11-08-2025)

Licence

Identifiants

Citer

Matthieu Dubois, François Yvon, Pablo Piantanida. MOSAIC: Multiple Observers Spotting AI Content. Findings of the Association for Computational Linguistics: ACL 2025, Association for Computational Linguistics, Jul 2025, Vienna, Austria. pp.24230-24247, ⟨10.18653/v1/2025.findings-acl.1244⟩. ⟨hal-05207044⟩
200 Consultations
290 Téléchargements

Altmetric

Partager

  • More