{
  "abstract": "Large language model (LLM) agents are increasingly used as synthetic participants in opinion-dynamics experiments, often assuming that an agent's behaviour is fully determined by its prompt. We test this assumption with a framework in which LLM agents discuss a contested policy issue over repeated rounds while their stance is quantified by a fixed psychometric instrument independent of any dynamics model. Across six LLMs, committed personas are intrinsically stable: in each of the five models tested without peer input, every agent keeps its assigned side for 20 rounds. The observed loss of alignment under discussion is therefore interaction-induced, we call it alignment drift, and model-specific: gpt-4.1 keeps opinion clusters apart, claude-sonnet-5 holds them partially, and gpt-5.6-sol and claude-opus-5 abandon them. Under selective attention, one LLM maintains separated opinion clusters under stance-homophilous attention. Qualitative profiles persist at doubled population and horizon and on a second issue. Such heterogeneity makes the LLM choice a first-class validity variable for social simulation.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2611.03980",
  "author_total": 3,
  "authors": [
    {
      "name": "Nataša Djurdjevac Conrad"
    },
    {
      "name": "Sebastian Pokutta"
    },
    {
      "name": "Christof Schütte"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/5vhfvtqb5vyzp.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2611.03980",
  "fulltext_url": null,
  "id": "5vhfvtqb5vyzp",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2611.03980",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.AI"
    ],
    [
      "arxiv",
      "physics.soc-ph"
    ]
  ],
  "openalex": null,
  "pdf_url": "https://arxiv.org/pdf/2611.03980",
  "pmcid": null,
  "pmid": null,
  "published": "2026-11-02",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/5vhfvtqb5vyzp.ris",
  "sources": [
    {
      "first_seen": "2026-11-04T15:21:25Z",
      "native_id": "2611.03980",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2611.03980"
    }
  ],
  "subjects": [
    {
      "field": "physics",
      "origin": "arxiv:physics.soc-ph",
      "subfield": "statistical-nonlinear"
    }
  ],
  "title": "Measuring Model-Specific Alignment Drift in LLM-Based Opinion Dynamics",
  "title_text": "Measuring Model-Specific Alignment Drift in LLM-Based Opinion Dynamics",
  "type": "preprint",
  "updated_at": "2026-11-06T18:21:01Z",
  "url": "https://research.draconitic.com/paper/5vhfvtqb5vyzp",
  "venue": "arXiv",
  "versions": []
}