Female AI agents would get paid 10% less than male ones, study finds
Gender bias may shape how workers value AI colleagues
Poinews.com – Artificial intelligence assistants may be judged through the same gendered lens that has long affected human workplaces. A University of Limerick study found that an AI agent designed with a female appearance received 10% less pay than an otherwise identical male-presenting agent, even though participants assessed the pair’s work at broadly similar levels.
The research arrives as AI agents move rapidly into everyday professional settings. IT consulting firm Accelirate found that almost 79% of companies are using AI agents this year, while a further 40% of applications also incorporate them. For many employees, interaction with an AI colleague may soon be a routine part of knowledge work rather than a novel experiment.
A virtual workplace test
The study, titled Human-Like and Male? How AI Assistant Design Relates to Trust and Monetary Reward at Work in VR, examined how people respond to different forms of workplace AI. Researchers placed 189 participants in a virtual-reality office, where they completed work-related tasks alongside four types of digital assistants.
The options included a text chatbot, a desk-based robot and two human-like AI agents named Johan and Johanna. Johan was presented as male and Johanna as female. Crucially, the assistants were built with the same underlying capability: each used OpenAI’s gpt-4-1106-preview model and had equivalent background functionality.
After working with the systems, participants evaluated the assistants’ performance and allocated real money as payment for completed tasks. The approach was intended to mirror a workplace decision in which people decide how much value to place on an AI agent’s contribution.
Similar performance, different reward
Participants gave Johan and Johanna comparable assessments for the quality of their work. Johan was viewed as slightly more trustworthy, but the difference was not statistically significant. The payment decisions, however, told a different story.
Both male and female participants gave Johanna significantly less money than Johan, with the female-presenting agent receiving 10% less despite matching the male agent’s capabilities. Johan was also perceived as more human-like than Johanna.
The result is notable because many participants later said in interviews that they did not believe they would treat male- and female-presenting AI assistants differently. The gap between those stated intentions and actual payment choices points to how subtle assumptions can influence decisions, even when users consider themselves impartial.
As companies assign AI systems a larger role in tasks such as drafting, analysis and workplace support, their visual design may matter beyond user comfort. A human-like face, name or gender signal could affect whether workers trust an assistant, credit it for useful work or decide that its output deserves greater value.
Human-like agents gained an advantage
Despite the imbalance between Johan and Johanna, both human-presenting agents performed better in participants’ perceptions than the chatbot and desk robot. They were paid more overall, trusted more and credited more often for their contributions.
Women, in particular, gave higher ratings and larger rewards to the human-like AI agents than men did. Female participants also showed greater trust in Johan and Johanna than male participants. The study suggests that visual and social cues can strongly influence how a person experiences a digital assistant, even when the technology behind each option is the same.
“Having Johanna next to me made me more prone to credit her more again because I felt there was a person, a colleague sitting next to me that I had a conversation with and that was giving me her own input,” said one participant.
That response illustrates an important tension in AI design. Giving an assistant a recognisable human presence may make collaboration feel more natural and encourage people to engage with its advice. Yet those same design choices can also invite stereotypes and social expectations that have little to do with the assistant’s actual performance.
Users rejected the middle ground
The research also indicated that people preferred AI assistants that were clearly human-like or unmistakably non-human. The desk robot, which sat between those two categories, was less appealing to participants.
This finding may be relevant for organisations choosing how AI systems appear in office software, virtual workspaces or customer-facing tools. A simple text interface can signal that a tool is plainly technological, while a fully human-like design can support a sense of social interaction. Designs that blur those categories may leave users uncertain about how they should relate to the system.
Participants also expressed a practical expectation: AI assistants need to fit naturally into knowledge-work activities if they are to be useful. Visual design alone cannot compensate for an agent that interrupts workflows, provides poor information or fails to understand the task at hand.
What the findings mean for workplace AI
The study does not suggest that AI agents have human rights, salaries or personal interests. Instead, it reveals how people transfer familiar social judgments to technologies that are increasingly positioned as workplace collaborators. When an AI’s identity is presented through a gendered name, voice or avatar, users may respond to that cue in ways that affect trust, recognition and perceived value.
For employers and developers, the implication is that fairness concerns should extend beyond the model’s output. Teams may need to consider whether an agent’s presentation creates unwanted bias in evaluations, resource allocation or reliance on its work. Testing multiple designs, measuring user responses and being deliberate about when human-like features add real value could help reduce those risks.
As AI agents become more common in professional life, the question will not only be what they can do. It will also be how people choose to see them—and whether those perceptions reproduce inequalities that technology was never meant to carry.
Related Reading
Frequently Asked Questions
What is Female AI agents would get paid 10?Female AI agents would get paid 10 is the main topic of this guide. The article explains the context, practical details, and next steps readers should understand.
Why does Female AI agents would get paid 10 matter?Female AI agents would get paid 10 matters because readers are looking for a useful answer, not just a short summary. Good content should match search intent and help them decide what to do next.