Employees’ perceptions of the fairness of AI-based performance prediction features
Predictive Artificial Intelligence (AI) algorithms are being increasingly adopted by organizations to inform workforce-related decisions. However, their effectiveness could be undermined by a lack of ground-truth data, particularly concerning human perceptions of fairness in predictive factors. This...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor & Francis Group
2025-12-01
|
| Series: | Cogent Business & Management |
| Subjects: | |
| Online Access: | https://www.tandfonline.com/doi/10.1080/23311975.2025.2456111 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Predictive Artificial Intelligence (AI) algorithms are being increasingly adopted by organizations to inform workforce-related decisions. However, their effectiveness could be undermined by a lack of ground-truth data, particularly concerning human perceptions of fairness in predictive factors. This study examined 306 employees’ perspectives on the fairness of 31 predictive features in five categories: demographic, academic, work-related, cognitive abilities and personality traits, and professional development and certification. Our findings revealed that employees generally regard features as fair when they align closely with job performance and contribute to prediction accuracy. Features related to technology usage attitude, workplace comfort, job rank satisfaction, alignment between one’s field of expertise and work, years of service in the same field, and total work experience were the highest perceived fair predictors. Conversely, more participants tended to perceive features such as gender, marital status, and number of children as unfair due to concerns about their actual contribution to prediction accuracy, privacy implications, and the risk of bias and discrimination. Features pertaining to professional development and certification were considerably viewed as fair predictors. These findings are valuable for the development of fairer AI algorithms for predicting employee performance by ensuring that predictive features align with employee perceptions of fairness. |
|---|---|
| ISSN: | 2331-1975 |