conv.

All stories
AIQuiet 6d · day 11

Research suggests LLMs develop pain-like representations responsive to linguistic harm

A new study identifies distinct pain representations in language models that respond to social and psychological harm.

newest coverage 5d agosummary written 6d ago7 pieces23 likes & upvotes2 comments NewswiresHacker NewsMastodon

What to know

  • New arXiv preprint investigates whether LLMs develop distinct pain representations separate from other negative emotions.
  • Researchers built a dataset across five pain categories: physical, psychological, social, moral, and others.
  • Commentary suggests LLMs show heightened responses to linguistically-delivered harms, a pattern consistent with their language-based training.

Valen Tagliabue Lead researcherLeonard Dung Co-researcherCameron Berg Co-researcherLeah McEvoy Commenter on research

How it unfolded 4 developments, newest first · click a bar or a number to jump articlesposts

Peak 1 piece in 3h at Sep 15, 11 PM; 7 pieces over 11 days (1 article · 6 posts) Sep 15, 11 PM — 1 piece · 1 article — Newswires 1Sep 16, 2 AM — quietSep 16, 5 AM — quietSep 16, 8 AM — quietSep 16, 11 AM — quietSep 16, 2 PM — quietSep 16, 5 PM — quietSep 16, 8 PM — quietSep 16, 11 PM — quietSep 17, 2 AM — quietSep 17, 5 AM — quietSep 17, 8 AM — quietSep 17, 11 AM — quietSep 17, 2 PM — quietSep 17, 5 PM — quietSep 17, 8 PM — quietSep 17, 11 PM — quietSep 18, 2 AM — quietSep 18, 5 AM — quietSep 18, 8 AM — quietSep 18, 11 AM — quietSep 18, 2 PM — 1 piece · 1 post — Hacker News 1Sep 18, 5 PM — quietSep 18, 8 PM — 1 piece · 1 post — Mastodon 1Sep 18, 11 PM — quietSep 19, 2 AM — quietSep 19, 5 AM — 1 piece · 1 post — Hacker News 1Sep 19, 8 AM — quietSep 19, 11 AM — quietSep 19, 2 PM — quietSep 19, 5 PM — 1 piece · 1 post — Hacker News 1Sep 19, 8 PM — quietSep 19, 11 PM — quietSep 20, 2 AM — quietSep 20, 5 AM — 1 piece · 1 post — Mastodon 1Sep 20, 8 AM — quietSep 20, 11 AM — quietSep 20, 2 PM — quietSep 20, 5 PM — quietSep 20, 8 PM — 1 piece · 1 post — Hacker News 1Sep 20, 11 PM — quietSep 21, 2 AM — quietSep 21, 5 AM — quietSep 21, 8 AM — quietSep 21, 11 AM — quietSep 21, 2 PM — quietSep 21, 5 PM — quietSep 21, 8 PM — quietSep 21, 11 PM — quietSep 22, 2 AM — quietSep 22, 5 AM — quietSep 22, 8 AM — quietSep 22, 11 AM — quietSep 22, 2 PM — quietSep 22, 5 PM — quietSep 22, 8 PM — quietSep 22, 11 PM — quietSep 23, 2 AM — quietSep 23, 5 AM — quietSep 23, 8 AM — quietSep 23, 11 AM — quietSep 23, 2 PM — quietSep 23, 5 PM — quietSep 23, 8 PM — quietSep 23, 11 PM — quietSep 24, 2 AM — quietSep 24, 5 AM — quietSep 24, 8 AM — quietSep 24, 11 AM — quietSep 24, 2 PM — quietSep 24, 5 PM — quietSep 24, 8 PM — quietSep 24, 11 PM — quietYesterday, 2 AM — quietYesterday, 5 AM — quietYesterday, 8 AM — quietYesterday, 11 AM — quietYesterday, 2 PM — quietYesterday, 5 PM — quietYesterday, 8 PM — quietYesterday, 11 PM — quietToday, 2 AM — quietToday, 5 AM — quietToday, 8 AM — quietToday, 11 AM — quiet 12–34
Sep 16Sep 17Sep 18Sep 19Sep 20Sep 21Sep 22Sep 23Sep 24yesterdaynow · 1:19 PM ET
  1. 4

    Paper continues to circulate across platforms

    The research continues to generate discussion on Hacker News and other aggregation platforms into mid-September.

  2. 3

    Researcher highlights linguistic harm responsiveness

    Leah McEvoy comments that the research is thought-provoking, noting that AI models show stronger responses to linguistically-delivered harms like gaslighting, dismissal, and insults than to other negative representations, which aligns with models being language-based.

    “more responsive to linguistically-delivered forms of harm (like gaslighting, dismissal, and insults) than other internal negative representations…”
    — Leah McEvoy

    • Thought-provoking research showing AI models’ as more responsive to linguistically-delivered forms of harm (like gaslighting, dismissal, and insults) than other internal negative representations—which makes sense given they’re language-based models. The Pain Axis: LLMs Represent ...

      leahmcelrath.bsky.social@bsky.brid.gyMastodon7d agoview on Mastodon ↗
  3. 2

    Paper gains traction on Hacker News

    The arXiv preprint begins circulating on Hacker News, receiving engagement from the technical community.

  4. 2 days quiet
  5. 1

    Researchers publish paper on LLM pain representations

    Tagliabue, Dung, and Berg release arXiv preprint analyzing whether large language models represent pain distinctly from fear, sadness, and other negative emotions, building a dataset describing painful situations across physical, psychological, social, moral, and other categories.

    1. 1 outlet The Pain Axis: LLMs Represent Self-Directed Harm and Act to Relieve It

      first by arXiv cs.AI, 10d ago

What people are saying 0 voices from 0 sites · best of 2 · verbatim