Management Accounting Section Midyear Meeting

Differing Perceptions of and Preferences for Human-Driven and Artificial Intelligence-Driven Evaluation Systems

• In Event: Performance Measurement and Evaluation (Experimental)

Abstract

Many employees and managers report dissatisfaction with existing performance evaluation
systems. New developments in artificial intelligence (AI) have made AI-driven performance
evaluation systems more accessible to more firms than ever before. Using an experiment, we
examine three factors that influence whether employees prefer performance evaluation systems
that are driven by AI or by human managers: employee disposition towards AI, the stability of
the firm’s operating environment, and the format of the data used in the evaluation. We find that
employees who have a more positive disposition towards AI are more likely to prefer an AIdriven system than a human-driven one. We also find that the preference for human-driven
systems is higher in unstable environments than in stable environments. This effect is driven by
employees’ belief that human managers are better able to adapt to new situations than AI is.
Humans are perceived to have a greater ability to be fair to employees when the reference point
for good performance in unclear. Despite this, we find that employees perceive AI-driven
systems to be better than human-driven systems at being free from bias against women and
minorities and evaluating quantitative data.

Author

• Jasmijn Bol, Tulane University