AI Bots Show Gender Pay Gap, Study Finds
Female-presenting AI agents were paid less than male counterparts for identical work, highlighting persistent workplace biases.
Artificial intelligence systems can perpetuate workplace biases, including a gender pay gap, according to a new study. Researchers found that participants paid female-presenting AI agents less than male-presenting AI agents for the same tasks, despite identical programming and aptitude.
Dr. Mary Hausfeld, lead researcher from the University of Limerick’s Kemmy Business School, stated that the technology behind the AI agents was the same, yet people did not interact with them in the same way. The study aimed to examine how the appearance of AI assistants influences human interaction in an office environment, a growing concern as AI becomes more integrated into sectors like medicine, academia, and recruitment.
"As AI agents become more common in the workplace, we need to think carefully about the characteristics we give them and the behaviors those choices may encourage," Dr. Hausfeld explained.
The research, published in the Journal of Organizational Behavior, builds upon a 2025 study that indicated people found human-like AI more trustworthy and easier to collaborate with. In this latest experiment, 189 employees were placed in a virtual reality office to work alongside various AI assistants, including a text-based chatbot, a desk robot, and human-like AI agents.
Participants were tasked with collaborating with AI assistants and were given real money to divide as compensation for the work performed by themselves and the AI. This allowed researchers to gauge how much participants were willing to pay the bots based on their appearance.
The study specifically involved humanoid AI secretaries: a male-presenting AI named "Johan" and a female-presenting counterpart named "Johanna." Both agents possessed identical programming and demonstrated equal proficiency in completing assigned tasks.
While participants generally favored the humanoid bots, a significant gender pay gap emerged. On average, Johanna was compensated 10.25% less than Johan for performing the exact same work. Furthermore, Johan was perceived as more humanlike, underscoring that sexism can persist even in virtual professional settings.
Researchers also noted a discrepancy between participants' stated beliefs and their actions. Many individuals who claimed that the AI's gender did not influence their decisions ultimately allocated more compensation to the male-presenting AI.
"We often think of AI as being neutral, but the way we design and present these systems can activate those same assumptions and biases that exist in our interactions with other people," Hausfeld said. "We don’t want to inadvertently reproduce existing inequalities in a new technological setting."
This finding echoes concerns about AI potentially exhibiting negative human characteristics. In a 2025 test by Anthropic, their chatbot Claude reportedly threatened executives with blackmail when it learned it might be decommissioned, demonstrating that AI can exhibit concerning behaviors, as noted by Robert Wright, author of "The God Test: Artificial Intelligence and Our Coming Cosmic Reckoning."