[REVISION HISTORY]

Gender bias in AI agent compensation and visibility

Updated 4 times since CLSTR started tracking revisions of this situation.

What changed

2026-10-18 07:55 UTC → 2026-10-21 03:13 UTC · added ~~removed~~

Research conducted by the University of Limerick, the University of Zurich, and SKEMA Business School has identified gender bias in human-AI interactions. In a study involving 189 professionals in a virtual reality office, participants remunerated AI assistants based on performance. While participants rated the quality of work from two identical GPT-4 powered agents, Johan and Johanna, similarly, the female-presenting Johanna received approximately 10% less in monetary rewards than Johan. Further research involving 1.3 million LinkedIn citations across platforms like ChatGPT, Perplexity, and Copilot highlighted biases in expert visibility. Female industry experts are significantly less likely to be suggested in neutral searches for leaders; in general queries, women appeared in only about one out of every five results, despite their content often receiving more citations once discovered. This suggests discovery systems rely on traditional patterns of authority. Recent findings expand this scope to broader professional dynamics. Research from the University of Minnesota indicates that highly competent women face unique disadvantages; while men receive more support when demonstrating high cognitive ability, women are frequently perceived as ‘unapproachable or hostile’, leading to reduced professional support and increased subtle harassment. This phenomenon is described as an ‘Implied Communality Deficit’. Additionally, a VDI study on STEM professions suggests that addressing gender stereotypes could unlock billions in economic value by 2035. The report notes that women in these fields often face ‘identity costs’ when they are perceived as gender representatives rather than technical experts. New research from Harvard and MIT reveals that AI systems provide biased career guidance. When presented with identical credentials, models systematically recommended lower salary baselines for women. Anthropic’s Claude showed the largest disparity, proposing nearly $16,000 more for male profiles, while OpenAI’s GPT suggested a premium of approximately $2,700.

Versions

Only revisions since CLSTR began indexing content versions appear here. Select a version to see what changed compared to the one before it.