Social Robots and Digital Humans as Job Interviewers: A Study of Human Reactions Towards a More Naturalistic Interaction

More and more companies have started to use nonhuman agents for employment interviews, making the selection process easier, faster, and unbiased. To assess the effectiveness of the above, in this paper, we systematically analyzed, reviewed, and compared human interaction with a social robot, a digital human, and another human under the same scenario simulating the first phase of a job interview. Our purpose is to allow the understanding of human reactions, concluding to a disclosure of the human needs towards human – nonhuman interaction. We also explored how the appearance and the physical presence of an agent can affect human perception, expectations, and emotions. To support our research, we used time-related and acoustic features of audio data, as well as psychometric data. Statistically significant differences were found for almost all extracted features and especially for intensity, speech rate, frequency, and response time. We also developed a Machine Learning model that can recognize the nature of the interlocutor a human interacts with. Although human was generally preferred, the interest level was higher and the shyness level was lower during human-robot interaction. Thus, we believe that, following some improvements, social robots, compared to digital humans, have the potential to act effectively as job interviewers.