Female AI agents are getting paid less than male ones, research finds
A shocking new experiment revealed that female-presenting AI tools aren't free from discriminatory pay practices, as despite completing the exact same work, these tools were paid less than their male-presenting counterparts.
It came to light as part of a new research project from experts at the University of Limerick, with findings of the study published in the journal, Proceedings of the ACM on Human-Computer Interaction, as per Futurism.
189 participants were placed into a virtual reality office filled with both human workers and AI agents, with the former instructed to complete tasks with a budget of money that they could allocate to different employees.
Included within these employees were Johan and Johanna, male- and female-presenting AI tools who completed the exact same work to the exact same standard, yet the former was designed to appear as a man with the latter appearing as a woman.
Female-presenting AI agent receives less money than male counterpart
Throughout the experiment, it was revealed that Johanna received 10.25 percent less money for completing the same tasks compared to Johan, with participants also perceiving it as 'less human-like' than the male version.
Advert
It is an important revelation in the wake of the growing importance of AI tools and 'assistants' in the workplace, dispelling the notion that artificial intelligence is a neutral agent that is separate from the conscious or unconscious bias applied to humans.
Whether participants were aware that they were paying Johanna less than Johan during the experiment is unclear, but the notion that it was an unconscious decision is arguably worse as it shows how in-built these perceptions are within so many people.
What have researchers said about these findings
Responding to the revelation, coauthor Mary Hausfeld called the findings 'striking', indicating that "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.
"We don't want to inadvertently reproduce existing inequalities in a new technological setting," the researcher added.
Hausfeld also emphasized the importance of monitoring and managing how AI tools are presented going forward, with their appearance and demeanor emerging as far from solely an aesthetic choice.
"Such choices may affect how much people trust AI," she noted, "how they judge its contribution, and even how they financially reward it."
Whether we like it or not, an alarming number of people are unable to shake the unconscious view they have of certain characteristics, and even having them present in something that you know isn't real doesn't make it go away.