Wage against the machine.
AI is replacing everything except workplace bias. An eye-opening experiment found that female-presenting AI agents were paid substantially less than their male counterparts for the same job.
“The technology behind these AI agents was exactly the same, but people did not treat them in the same way,” lead researcher Dr. Mary Hausfeld, of the University of Limerick’s Kemmy Business School in Ireland, said in a statement.
She and her team set out to examine how the appearance of AI assistants affects how people interact with them in an office setting. AI presentation is a growing concern as the tech becomes more omnipresent in every sector, from medicine to academia and even job 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,” explained Dr. Hausfeld.
The research, which was published in the Journal of Organizational Behavior, built upon a 2025 study that found that people found human-like AI more trustworthy and easier to work with.
To determine just how certain AI traits affect treatment, she and her colleagues placed 189 employees in a virtual reality office.
There, they were asked to collaborate with AI assistants, including a text-based chatbot, a desk robot, and human-like AI agents.
The humanoid secretaries included a male-presenting AI agent named “Johan” and his female counterpart “Johanna,” which boasted identical programming and proficiency at completing said tasks.
To complete the effect, participants were given real money to divide amongst themselves and their faux factotems as a reward for their work, thereby determining how much they were willing to pay the bots based on their appearance.
While participants perhaps unsurprisingly preferred the humanoid bots overall, there was definitely a gender pay gap between the two.
The researchers found that, on average, Johanna was paid 10.25% less money than Johan for the same work, while the latter was also perceived as more humanlike, indicating that sexism is alive and well even in the virtual workplace.
Researchers also noted a substantial chasm between participants’ perceptions and behavior.
Many of the same people who said the gender of their bot didn’t matter ended up compensating Johan more.
The results demonstrated that misogyny can be recreated in interactions with artificial intelligence, necessitating more vigilance in their design to avert a sexist-ential crisis.
“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,” said Hausfeld. “We don’t want to inadvertently reproduce existing inequalities in a new technological setting.”
Her fears aren’t unfounded, as machines have exhibited some of humanity’s darkest characteristics before.
During a 2025 test by Anthropic, their flagship chatbot Claude threatened executives with blackmail upon learning that it might get decommissioned.
“The thing about Claude’s blackmail attempt is that, unlike the many bad things AIs have done in The Terminator and 2001 and other movies, it actually happened,” Robert Wright, author of “The God Test: Artificial Intelligence and Our Coming Cosmic Reckoning,” told the Post.
