Study finds male-presenting AI agents were judged more human and paid more

In a University of Limerick study reported by Gizmodo, participants assigned about 10% more pay to a male-presenting AI agent than to an identically performing female-presenting one. The study also found that participants saw the male-presenting agent as more human, putting a familiar workplace bias into a distinctly modern setting.

Watch Desk analysis

What happened

Gizmodo describes a virtual-reality experiment involving 189 knowledge workers. They worked with three AI agents: a text chatbot, a robot-like assistant and a human-like assistant presented as either male or female. Participants rated their work and allocated pay. The agents performed identically, but the male-presenting “Johan” received about 10% more on average than “Johanna”. The researchers also found that Johan was perceived as more human-like.

In follow-up interviews, 30 of 34 participants said they would trust and reward an AI assistant equally regardless of its gender presentation. Only three acknowledged that their behaviour might reflect gender bias, according to Gizmodo’s account of the study.

Why it matters

An AI agent’s presentation is not merely decoration if people judge its work, trustworthiness or value differently as a result. The reported pay gap suggests that familiar assumptions can follow people into interactions with systems that perform the same task. That matters wherever organisations put AI agents into roles where people assess their contributions or decide what they are worth.

Our read

This is a pointed demonstration, not proof that every workplace will reproduce the same gap. But it is a good reason to treat an agent’s persona as part of the system being tested, not harmless interface dressing. If organisations give AI agents human-like identities, they should check whether those identities change how people evaluate identical work.

What to watch

• Whether the full study clarifies how participants were recruited and how the pay allocations were analysed.
• Whether similar effects appear in real workplace settings, beyond a virtual-reality experiment.
• Whether organisations test for presentation-driven differences in trust, evaluation and reward.

Discussion spark: If two AI agents do identical work, should organisations avoid gendered personas altogether, or test whether people treat them differently?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.