Is AI in hiring actually making things fairer?

A new multidisciplinary review published in Human Relations by Alberta School of Business scholar Karen D. Hughes , alongside co-authors Alla Konnikov (Concordia University), Nicole Denier (Colby College) and Yang Hu (University College London), dives deep into this question. The research problematizes the role of algorithmic hiring, revealing that instead of eliminating prejudice, AI may actually reshape and obscure the very mechanisms that sustain workplace inequality.

“AI is often presented as a neutral tool that can make hiring fairer and faster,” says Hughes. “But when hiring processes become embedded in complex algorithms, it can become harder to see how decisions are made, and how inequalities may be reproduced.”

In their comprehensive review spanning computing science, human resources, social sciences and legal scholarship, the research team introduces the concept of “algorithmically mediated inequality regimes,” extending existing theories of organizational inequality to the digital age. This framework highlights how AI’s opaque “black box” nature can make discrimination harder to spot and even harder to challenge, all while maintaining a false illusion of technological objectivity.

In practice, AI systems increasingly mediate multiple stages of hiring — from shaping who sees job ads, to collecting digital “behavioural traces” from online applications, to ranking and filtering candidates — often in ways that are difficult for applicants to see or challenge.
The researchers conclude that we cannot rely on just one group to fix this. To make sure AI doesn’t recreate old inequalities, they call for a “chain of responsibility.” This means computer scientists, HR professionals, lawyers and policymakers must work together to build, use and regulate AI hiring tools fairly.

“It is only through a multidisciplinary lens that these patterns, tensions and blind spots become visible,” notes Hughes, stressing the importance of looking beyond just technical code to understand AI’s broader impact on job seekers.

Key takeaways

• AI operates as a “black box”: AI in hiring often functions invisibly, making it difficult for applicants to understand how and why decisions about their careers are made. This invisibility can reproduce and hide systemic inequalities within the system.
• Technical fixes aren’t enough: While many developers view hiring bias as a data flaw that can be “de-biased” with better algorithms, the research emphasizes that technology alone cannot solve deeply rooted social and structural inequities.
• Shifting power dynamics: AI tools bundle vast amounts of data — from digital resumés to behavioural traces generated through everyday online activity, such as the device used to apply — granting employers increased power and surveillance capabilities while limiting the agency and privacy of job seekers.

• Regulation is lagging behind: There is significant regulatory lag and legal contestation when it comes to governing AI in hiring. This creates a need for an interdisciplinary approach and a multi-stakeholder “chain of responsibility,” a concept developed in the research to highlight how accountability is shared across those who design, deploy and regulate AI systems.
  

Read the full article in Human Relations at DOI:10.1177/00187267251403902 . The article was one of four selected from 82 original submissions for an inaugural special issue of reviews.

U of A research. Real-World impact.

Closing the gap means tackling the inequality head on. By championing inclusivity and justice, the U of A is striving to achieve the UN's goal of a more equitable world.

See The Fight For Fairness