HuangChaochao / University of Science and Technology of China
ZhengQuan / University of Science and Technology of China
An increasing number of firms are incorporating artificial intelligence (AI) technology into the interview process to assess applicants more precisely. However, the pre-programmed nature of AI systems also makes them susceptible to manipulation by applicants, thereby diminishing their predictive accuracy. In contrast, human resource (HR) professionals can rationally infer the manipulation strategies of applicants. Motivated by the gap between the widespread use of AI interviews and the lack of practical guidelines, we develop a game-theoretic model to examine the algorithm design strategy and the implications of AI interviews for firms. We refer to the competency framework of a job as the extent to which it requires personality and expertise, with personality being manipulable by applicants at a cost. Our analysis shows that an HR professional may resort to extreme hiring strategies (either aggressive or conservative), while the AI system can strategically limit the algorithmic precision of either personality or expertise to mitigate manipulation behavior. The firm always prefers AI interviews for a lopsided job where the competency strongly depends on one feature relative to the other. However, for a balanced job where the competency depends moderately on both features, the firm leans towards HR interviews if HR's ability to assess expertise is high and the cost of manipulation is low; otherwise, the firm prefers AI interviews. Our findings align well with real-world observations.