When Stereotypes GTG: The Impact of Predictive Text Suggestions on Gender Bias in Human-AI Co-Writing

Synopsis

Tests whether anti-stereotypical text suggestions debias users' writing in 414 participants; finds the effect is unreliable while stereotypical suggestions do bias output.

Keywords

·debiasing ·predictive text ·gender bias ·co-writing

Related entries

• Does Writing with Language Models Reduce Content Diversity?
  September 11, 2023 · ICLR 2024
• 'It was 80% me, 20% AI': Seeking Authenticity in Co-Writing with Large Language Models
  November 20, 2024 · CSCW 2025
• Wordcraft: Story Writing With Large Language Models
  2022 · IUI 2022
• CoAuthor: Designing a Human-AI Collaborative Writing Dataset for Exploring Language Model Capabilities
  January 18, 2022 · CHI 2022
• Strategic Algorithmic Monoculture: Experimental Evidence from Coordination Games
  April 10, 2026 · arXiv
• Modeling Distinct Human Interaction in Web Agents
  February 19, 2026 · arXiv