Metraipolis › archive › dc_2b2b573e1098

Whose Assessment of Distress? Community Perspectives and LLM Alignment on Well-Being Posts — arXiv:2608.29446

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.29446
License (as archived)
CC0-1.0
Archived
2026-10-17T08:06:59.176Z
SHA-256 of the archived text
3676a7f11cd8d47e7a5d00cb8ca75483696d4b00f8c1061a1f442432e5d9b2c3
Stable URL of this page
https://metraipolis.com/documento/dc_2b2b573e1098

How to cite

Whose Assessment of Distress? Community Perspectives and LLM Alignment on Well-Being Posts — arXiv:2608.29446 — arXiv (en), https://arxiv.org/abs/2608.29446 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2b2b573e1098 (Metraipolis archive dc_2b2b573e1098, sha256 3676a7f11cd8d47e7a5d00cb8ca75483696d4b00f8c1061a1f442432e5d9b2c3)
BibTeX
@misc{dc2b2b573e1098,
  title = {Whose Assessment of Distress? Community Perspectives and LLM Alignment on Well-Being Posts — arXiv:2608.29446},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.29446},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2b2b573e1098; sha256 3676a7f11cd8d47e7a5d00cb8ca75483696d4b00f8c1061a1f442432e5d9b2c3},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.29446. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Judgments about psychological distress are socially situated: what counts as concerning hinges on community norms around emotional expression, vulnerability, and help-seeking. Yet large language models (LLMs) used for distress detection are typically aligned to a single, undifferentiated standard. How well do these models capture the perspectives of the communities whose language they assess? We address this question through a perspectivist annotation study in which 321 participants provided 9,587 judgments on 1,198 Reddit posts spanning six identity-based communities, yielding community-specific labels. Raters in the contextualized in-group condition show a modest tendency to agree more with their community than uncontextualized out-group raters (OR = 1.18), an effect varying significantly across communities. We then evaluate nine open-weight LLM configurations and four frontier configurations against these labels. Open-weight LLMs systematically over-estimate distress: when communities perceive none-to-mild distress, these models achieve only 31-44% accuracy, predominantly producing false positives. GPT-5 and Gemini 2.5 Pro show the same none-to-mild inflation even when their full-sample over/under rates are mixed, while Claude Opus 4 is more conservative. This pattern does not simply mirror an outsider reading position: uncontextualized out-group human aggregates were nearly symmetric, with 18% over-estimation versus 19% under-estimation. Instead, the models that inflate none-to-mild cases exhibit a distress prior that exceeds both contextualized in-group and uncontextualized out-group human judgments. These findings have implications for equitable AI deployment in mental health contexts, where miscalibrated distress detection may unevenly affect the communities being assessed.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.29446 — 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 2026-10-17T07:47:52.698Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).