Metraipolis › archive › dc_143635cb33ea

Latent space bias directions in LLMs capture confidence, not fairness — arXiv:2610.08559

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.08559
License (as archived)
CC0-1.0
Archived
2026-10-07T11:21:57.478Z
SHA-256 of the archived text
4c0ecfedd320e4bd7354fc3679ae485940f1f0197e38c0a9f7e769e835aa06e8
Stable URL of this page
https://metraipolis.com/documento/dc_143635cb33ea

How to cite

Latent space bias directions in LLMs capture confidence, not fairness — arXiv:2610.08559 — arXiv (en), https://arxiv.org/abs/2610.08559 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_143635cb33ea (Metraipolis archive dc_143635cb33ea, sha256 4c0ecfedd320e4bd7354fc3679ae485940f1f0197e38c0a9f7e769e835aa06e8)
BibTeX
@misc{dc143635cb33ea,
  title = {Latent space bias directions in LLMs capture confidence, not fairness — arXiv:2610.08559},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.08559},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_143635cb33ea; sha256 4c0ecfedd320e4bd7354fc3679ae485940f1f0197e38c0a9f7e769e835aa06e8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.08559. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Activation steering has gained popularity as a lightweight inference-time debiasing technique for large language models. However, prior work reports that steering vectors generalise poorly, with unintended effects on model performance and limited transfer to new datasets. Our work analyses what the debiasing direction used for activation steering actually encodes, in order to shed light on its inconsistent performance. We study the linear debiasing direction obtained by contrasting the activations of anti-biased and biased prompts, and evaluate it as a steering intervention across bias and general knowledge benchmarks. We find that this direction is dominated by model confidence, pointing from regions of high to low-probability tokens in activation space rather than encoding a meaningful representation of model bias. Steering along it does reduce measured bias, but this is a consequence of reducing model confidence: on QA benchmarks we find that this steering drives the model to abstain from answering, with a side effect of improving fairness metrics. Our experiments show that model confidence is the dominant separating factor between biased and anti-biased prompts in hidden space, indicating that isolating a linear representation of bias which is disentangled from model confidence is difficult and steering-based debiasing results should be interpreted with care. In short, steering appears to reduce bias, not by correcting the model's underlying preferences, but by making it less confident, even on tasks unrelated to bias.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.08559 — 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 2026-10-07T11:15:24.844Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).