AI Screening Tools Being Used By US Employers Are Showing Racial Bias Toward Black And Asian Job Applicants – AfroTech
A new study reveals that racial bias arises when employers rely on AI screening tools to evaluate job applicants.
The study, “Algorithmic Monocultures in Hiring,” conducted by researchers from Stanford University, Chapman University, and Northeastern University, found that employers are viewing nearly three times as many applications for entry-level positions as in 2022. Additionally, AI has disrupted the job-hiring process. In fact, more than 90% of employers in the U.S. rely on AI screening tools from third-party vendors to organize and rank applications, the study notes.
“They almost have to use an algorithm in some way to be able to look through the number of applications that are coming in,” said Sarah Bana, an assistant professor at Chapman University and one of the co-authors on the Stanford study, according to Marketplace.
Stanford University researcher Rishi Bommasani said the team examined how vendors use AI to make hiring recommendations. The study analyzed the algorithmic pooling of data from 3.4 million real job applicants by submitting 4 million applications to 156 employers across 11 market sectors. The results revealed racial bias in the AI screening tools from a single vendor being used by U.S. employers.
“We are the first to demonstrate large-scale evidence of racial disparities and homogeneous outcomes in high-stakes hiring decisions,” a blog post from the study read.
Results
The study found evidence that the AI system favored white applicants over Black and Asian applicants. Specifically, 26% of Black applicants and 15% of Asian applicants applied for positions where the AI system exhibited bias against their racial group. If these applicants had advanced through the hiring process at the same rate as their white counterparts, an estimated 40,000 additional candidates would have advanced to the next step.
“Today when you apply to jobs, you might submit multiple job applications to different employers. And you might think that each of those employers is gonna make a different decision about you,” Bommasani explained. “But what we find in our recent research is that if one of those employers rejects you, they all do. And that’s because today we find firms are using the same hiring AI tools to inform their decisions in hiring practices.”
“Our first major finding relates to racial discrimination. What we found is adverse impact for Black and Asian applicants in particular. What that means is that for some positions, the hiring AI tool was less likely to recommend Black and Asian applicants to certain positions,” he later added.
Bommasani also said the study cautions against a wider adoption of AI in hiring. Applicants who applied to multiple jobs were more likely to be rejected by all of them than would be expected by chance. This outcome likely would have been different if companies had made hiring decisions on their own, rather than relying on a vendor that leverages AI.
“As more and more companies adopt AI in their hiring practices, policymakers are actively grappling with the questions of how to govern. Hiring AI is high stakes and incredibly prevalent, yet there’s very little external scrutiny into these tools. And so it’s important that we can build the evidence base and understand how hiring AI is affecting the practices of firms and the outcomes applicants see in the labor economy,” Bommasani said, per the study.
Aalok Mehta, director of the Wadhwani AI Center at the Center for Strategic and International Studies, warned that if companies do not incorporate human review and better manage vendors using AI screening tools, the study’s findings will “replicate patterns all across industry,” according to Marketplace.
