Metraipolis › archive › dc_b557fe6bcd98

Do large language models keep the hedge? Uncertainty preservation when large language models summarize aviation and maritime incident reports (2026) · DOI 10.3389/flang.2026.1953328

Source
Metraipolis archive
Original
https://doi.org/10.3389/flang.2026.1953328
License (as archived)
CC-BY-4.0
Archived
2026-10-17T23:09:30.407Z
SHA-256 of the archived text
cbf47559b00f9ade1450730d9bebe1fca77e67d314d0d5cf30d98c0601df59da
Stable URL of this page
https://metraipolis.com/documento/dc_b557fe6bcd98

How to cite

Do large language models keep the hedge? Uncertainty preservation when large language models summarize aviation and maritime incident reports (2026) · DOI 10.3389/flang.2026.1953328 — Metraipolis archive, https://doi.org/10.3389/flang.2026.1953328 (CC-BY-4.0); archived copy: https://metraipolis.com/documento/dc_b557fe6bcd98 (Metraipolis archive dc_b557fe6bcd98, sha256 cbf47559b00f9ade1450730d9bebe1fca77e67d314d0d5cf30d98c0601df59da)
BibTeX
@misc{dcb557fe6bcd98,
  title = {Do large language models keep the hedge? Uncertainty preservation when large language models summarize aviation and maritime incident reports (2026) · DOI 10.3389/flang.2026.1953328},
  howpublished = {Metraipolis archive},
  url = {https://doi.org/10.3389/flang.2026.1953328},
  note = {License CC-BY-4.0; archived copy https://metraipolis.com/documento/dc_b557fe6bcd98; sha256 cbf47559b00f9ade1450730d9bebe1fca77e67d314d0d5cf30d98c0601df59da},
  year = {2026}
}

Archived text

Reproduced under CC-BY-4.0. Attribution: Metraipolis archive, https://doi.org/10.3389/flang.2026.1953328. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Articolo recente con licenza aperta: scheda dai metadati Crossref (journal-article).
Titolo: Do large language models keep the hedge? Uncertainty preservation when large language models summarize aviation and maritime incident reports
Autori: Simon Baradziej
Pubblicato: 2026-10-12 · in: Frontiers in Language Sciences · editore: Frontiers Media SA
DOI: https://doi.org/10.3389/flang.2026.1953328
Licenza dell'articolo: CC-BY-4.0

Abstract (testo degli autori, CC-BY-4.0):
Safety-incident reports are written with deliberate epistemic caution: A reporter marks what was seen, what was inferred, and what remains unsure, and a dropped hedge can turn a tentative observation into an apparent cause. As large language models take over summarizing such reports for triage, indexing, and search, whether they preserve that caution becomes a property of the safety record itself. This study asks whether models keep a reporter's expressed uncertainty when they summarize, testing two hypotheses fixed before any summary was generated: that models systematically harden uncertainty by default and that an explicit instruction to preserve it overcorrects into softening. Six open-weight models summarized 42 real incident reports, 30 from the United States Aviation Safety Reporting System and 12 from the United Kingdom Marine Accident Investigation Branch, under a plain prompt, a preserve prompt, and a paraphrase of it. Uncertainty change was measured two ways: at the passage level, by a deterministic epistemic-marker lexicon, and at the claim level, by aligning those markers between source and summary, each validated against manual annotation and cross-checked against two independent language-model judges. Under plain summarization, all six models hardened the reporter's uncertainty (claim-level hardening 0.59–0.85), dropping explicit hedges almost entirely (retention about 0.08) while epistemic modal verbs survived far better. The preserve instruction reversed the pattern into over-softening: In a cluster-robust logistic regression, it carried an odds ratio of 0.12 (95% CI [0.03, 0.49], p = 0.003) and held once summary length was controlled, so the change reflects expressed certainty rather than compression. In practical terms, the instruction lowered the predicted probability of hardening from about 0.43 to 0.13 (Cohen's h = 0.69), and the effect held in mixed-effects models and under adverse imputation of the missing outputs. The pattern was directionally consistent across the tested models and both instruction wordings. Automated safety triage that relies on model summaries should measure epistemic fidelity directly and not treat a single preserve instruction as a solution.

Fonte: https://api.crossref.org/works/10.3389%2Fflang.2026.1953328 — metadati Crossref, consultati il 2026-10-17. Interrogazione: «evaluation of large language models».