Job applicants can be wary of AI in hiring, but perceptions improve when they know the algorithm ignores gender, race, or age, says Yakov Bart, a marketing professor at Northeastern University.
Bart, along with researchers Lily Morse and Mike Teodorescu, conducted a behavioral experiment to study fairness perceptions based on algorithm transparency. They tested three scenarios:
- Fairness-through-unawareness: The algorithm ignores protected characteristics.
- Demographic parity: The algorithm ensures equal outcomes regardless of demographics.
- Equality of opportunity: The algorithm ensures equally qualified candidates have the same chances, regardless of demographics.
The fairness-through-unawareness scenario was the most positively received. Other scenarios did not significantly impact perceptions or were unpopular. This was especially true for male participants, though reasons remain unclear.
“Our findings indicate that people perceive hiring algorithms as procedurally fairest when companies adopt a ‘fairness through unawareness’ approach to mitigating bias,” Bart said. “They are also likely to view companies who use this approach more positively and are more motivated to apply for open positions.”
Bart suggested companies consider the fairness-through-unawareness approach to improve candidate perceptions and protect their reputation.
Bart also explained that people often act based on their perceptions rather than objective truths. He stressed that the algorithms do not consider race, gender, age, or other protected characteristics, making them “blind” to these factors to ensure equal treatment for all applicants. He noted that while companies might believe that implementing demographic parity is fair, average job applicants might disagree. The study found that women and non-binary individuals responded most positively to the fairness-through-unawareness approach. Bart suggested that companies adopt this approach, as it has the highest potential to attract applicants and should be consistently applied, not just used as an explanation.
