Metraipolis › archive › dc_fcbc0e2fba60

Toward a Graded Measure of Belief Stability in Large Language Models — arXiv:2609.34158

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.34158
License (as archived): CC0-1.0
Archived: 2026-10-07T21:02:11.967Z
SHA-256 of the archived text: 714bd67a9011f60586339ea6a99682986e0be4abc69ed0bedf4b813db3611d30
Stable URL of this page: https://metraipolis.com/documento/dc_fcbc0e2fba60

How to cite

Toward a Graded Measure of Belief Stability in Large Language Models — arXiv:2609.34158 — arXiv (en), https://arxiv.org/abs/2609.34158 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fcbc0e2fba60 (Metraipolis archive dc_fcbc0e2fba60, sha256 714bd67a9011f60586339ea6a99682986e0be4abc69ed0bedf4b813db3611d30)

BibTeX

@misc{dcfcbc0e2fba60,
  title = {Toward a Graded Measure of Belief Stability in Large Language Models — arXiv:2609.34158},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34158},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fcbc0e2fba60; sha256 714bd67a9011f60586339ea6a99682986e0be4abc69ed0bedf4b813db3611d30},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34158. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models (LLMs) increasingly mediate how people access and reason with information, yet factual reliability is usually evaluated one judgment at a time. We introduce graded belief stability, a relational measure of how well a belief persists within an LLM's broader belief system. Unlike individual belief probability, it asks whether support for a claim persists when that claim is considered alongside the model's other epistemic commitments. We operationalize this idea with a Direct Conditional estimator that uses internal model representations to estimate conditional belief probabilities. Across 12 LLMs and three domains, lower-stability beliefs exhibit greater mean behavioral movement under conversational challenge in 83.3% of model-domain settings after matching on individual belief probability. Graded belief stability therefore extends reliability assessment beyond how strongly an LLM supports a claim to how robustly that belief is supported within its broader system of beliefs.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.34158 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T20:29:02.878Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).