AI Reveals Two Centuries of Sexism in British Parliament

Mohammad Omar Khursheed, Mandira Sawkar, Ashiqur R. KhudaBukhsh· September 3, 2026 View original

Key takeaways

  • AI can effectively analyze vast historical texts to uncover subtle and overt biases.
  • Sexism persisted in British parliamentary debates for two centuries, even in pro-women's rights speeches.
  • Different types of sexism (hostile vs. benevolent) were used in distinct rhetorical contexts.
  • Computational analysis provides empirical evidence for theories like Ambivalent Sexism.

Who benefits

Social SciencesPublic PolicyHuman ResourcesMedia Analysis

Summary

A computational analysis of 200 years of British parliamentary debates (Hansard corpus) using large language models reveals persistent patterns of sexism. The study found that both pro- and anti-women's rights speeches contained sexism, with different types of sexism used in each context.

This research delves into two centuries of British parliamentary debates, utilizing large language models to computationally analyze the Hansard corpus from 1803 to 2005. The study aimed to classify speakers' perspectives on women's suffrage and political representation, as well as to identify and analyze sexist language through the lens of the Ambivalent Sexism Inventory. A significant dataset of 6.7 million speeches was curated and released for computational social science research. The findings indicate that sexism was prevalent across both pro- and anti-suffrage rhetoric. Specifically, 54% of speeches opposing women's representation contained sexist content, compared to 21% of speeches supporting the cause. The analysis further revealed that anti-suffrage arguments often combined hostile and benevolent sexism, while pro-suffrage arguments predominantly featured benevolent sexism. Female Members of Parliament showed significantly higher support for women's political rights than their male counterparts, a gap that only narrowed after women gained the right to vote.

Why it matters

This study demonstrates the power of AI and computational linguistics to uncover deep-seated societal biases and historical patterns in large text datasets, offering valuable insights for social science and policy analysis.

How to implement this in your domain

  1. 1Apply similar computational linguistic techniques to analyze historical or contemporary corporate communications for bias detection.
  2. 2Utilize large language models to audit internal documents, meeting transcripts, or public statements for subtle forms of bias.
  3. 3Develop tools based on this research to monitor and flag potentially biased language in organizational discourse.
  4. 4Inform diversity, equity, and inclusion (DEI) initiatives with data-driven insights into language patterns.

Original post by Mohammad Omar Khursheed, Mandira Sawkar, Ashiqur R. KhudaBukhsh

"arXiv:2608.30485v1 Announce Type: cross Abstract: The language a legislature uses to debate women's rights, even in favour of them, encodes systematic patterns of sexism that persist across two centuries. In this work, we analyse 6,531 speeches over 200 years of UK parliamentary…"

View on X

Originally posted by Mohammad Omar Khursheed, Mandira Sawkar, Ashiqur R. KhudaBukhsh on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses