The Confidence Trap: Gender Bias and Predictive Certainty in LLMs

Ahmed Sabir, Markus Kängsepp, Rajesh Sharma
1/12/2026
cs.CLcs.LG

No official code found. Best implementation: ahmedssabir/GECE (1 star, last commit 2026-01-15).

Abstract

The increased use of Large Language Models (LLMs) in sensitive domains leads to growing interest in how their confidence scores correspond to fairness and bias. This study examines the alignment between LLM-predicted confidence and human-annotated bias judgments. Focusing on gender bias, the research investigates probability confidence calibration in contexts involving gendered pronoun resolution. The goal is to evaluate if calibration metrics based on predicted confidence scores effectively capture fairness-related disparities in LLMs. The results show that, among the six state-of-the-art models, Gemma-2 demonstrates the worst calibration according to the gender bias benchmark. The primary contribution of this work is a fairness-aware evaluation of LLMs' confidence calibration, offering guidance for ethical deployment. In addition, we introduce a new calibration metric, Gender-ECE, designed to measure gender disparities in resolution tasks.

View on arXivView PDF

Code Implementations(1)

ahmedssabir/GECECommunity70%

The Confidence Trap: Gender Bias and Predictive Certainty in LLMs. AAAI 2026

12Python, Jupyter NotebookNov 7, 20259 months ago

Similar Papers

Assessing Gender Bias in LLMs: Comparing LLM Outputs with Human Perceptions and Official Statistics

88% match1 repo12024

GenderCARE: A Comprehensive Framework for Assessing and Reducing Gender Bias in Large Language Models

86% match1 repo272024

Bridging the Fairness Gap: Enhancing Pre-trained Models with LLM-Generated Sentences

86% match2025

Decoding Biases: Automated Methods and LLM Judges for Gender Bias Detection in Language Models

86% match2024

Towards Fair Rankings: Leveraging LLMs for Gender Bias Detection and Measurement

86% match2025

Is Your Model Fairly Certain? Uncertainty-Aware Fairness Evaluation for LLMs

86% match1 repo82025

Cite this paper

@article{sabir2026confidence,
  title  = {The Confidence Trap: Gender Bias and Predictive Certainty in LLMs},
  author = {Ahmed Sabir and Markus Kängsepp and Rajesh Sharma},
  year   = {2026},
  eprint = {2601.07806},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2601.07806v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2601.07806.svg)](https://codeofpaper.com/paper/2601.07806)

Discussion