Archive · 10 Oct 2026
Study Finds Female-Presenting AI Assistants Received 10.25% Less Pay
Researchers studied how people value AI assistants shown as digital avatars.
Participants saw the assistants in a pretend virtual workplace.
The assistants did the same work and used the same technology.
But participants gave the female-presenting assistants 10.25% less money than the male-presenting ones.
Both men and women in the study gave higher valuations to the male-presenting assistants.
This suggests that people may bring real-world gender biases into how they treat digital workers.
Experts warned that gendered designs could reinforce unfairness at work.
They suggested using neutral-looking assistants or clear rules for judging their work.
Researchers found participants assigned female-presenting AI assistants 10.25% less financial compensation than identical male-presenting assistants.
The AI avatars performed the same tasks using identical underlying technology and capabilities.
The experiment placed participants in simulated virtual office environments and asked them to evaluate AI agents’ work.
Both male and female participants gave male-presenting AI assistants higher monetary valuations.
The researchers and experts called for neutral personas or objective evaluation rules to help limit bias.
- Who
- Researchers from the University of Limerick, the University of Zurich, and SKEMA Business School studied human participants’ evaluations of AI assistants.
- What
- The study found that participants assigned female-presenting AI assistants 10.25% less financial compensation than identical male-presenting counterparts.
- When
- The article was published October 10, 2026; the study date was not specified.
- Where
- The experiment took place in simulated virtual office environments.
- Why
- The study examined whether people project gender bias onto AI assistants, including when their capabilities and work are identical.
- Reported pay difference
- Female-presenting AI assistants received 10.25% less financial compensation.
- Task performance
- The avatars completed identical tasks with identical underlying technology and capabilities.
- Participants
- Both male and female participants assigned higher monetary valuations to male-presenting assistants.
- Study setting
- Simulated virtual office environments.
- Researchers’ institutions
- University of Limerick, University of Zurich, and SKEMA Business School.
- Suggested safeguards
- Neutral digital personas or objective evaluation frameworks.