English

Selection Collider Bias in Large Language Models

Computation and Language 2022-09-14 v2 Artificial Intelligence

Abstract

In this paper we motivate the causal mechanisms behind sample selection induced collider bias (selection collider bias) that can cause Large Language Models (LLMs) to learn unconditional dependence between entities that are unconditionally independent in the real world. We show that selection collider bias can become amplified in underspecified learning tasks, and although difficult to overcome, we describe a method to exploit the resulting spurious correlations for determination of when a model may be uncertain about its prediction. We demonstrate an uncertainty metric that matches human uncertainty in tasks with gender pronoun underspecification on an extended version of the Winogender Schemas evaluation set, and we provide an online demo where users can apply our uncertainty metric to their own texts and models.

Keywords

Cite

@article{arxiv.2208.10063,
  title  = {Selection Collider Bias in Large Language Models},
  author = {Emily McMilin},
  journal= {arXiv preprint arXiv:2208.10063},
  year   = {2022}
}

Comments

12 pages, 16 figures, UAI 2022 Causal Representation Learning Workshop

R2 v1 2026-06-25T01:51:35.059Z