Latent space bias directions in LLMs capture confidence, not fairness

A paper submitted to arXiv on 6 October 2026 finds that the linear debiasing direction used in activation steering for large language models is dominated by model confidence rather than encoding a meaningful representation of bias. The authors report that steering along this direction reduces measured bias only because it lowers model confidence, driving the model to abstain from answering on QA benchmarks, which has the side effect of improving fairness metrics. The paper concludes that isolating a linear representation of bias disentangled from model confidence is difficult and that steering-based debiasing results should be interpreted with care.

Computer Science Computation and Language Submitted on 6 Oct 2026 Title:Latent space bias directions in LLMs capture confidence, not fairness View PDF https://arxiv.org/pdf/2610.08559 HTML experimental https://arxiv.org/html/2610.08559v1 Abstract: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. Additional Features Current browse context: cs.CL References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .