Can a machine hire better than a human?
In a recruitment experiment involving more than 70,000 job applications, candidates who were allowed to choose between a human interviewer and an artificial intelligence system made an unexpected…
In a recruitment experiment involving more than 70,000 job applications, candidates who were allowed to choose between a human interviewer and an artificial intelligence system made an unexpected decision: 78% preferred to speak with the machine (Jabarian & Henkel, 2026).
A few years ago, this would have sounded like a scene from a science-fiction film: an artificial voice asks questions, analyses the answers and prepares an assessment for the company. Today, this is already possible. Artificial intelligence can review résumés, interview candidates and organise information before a person makes the final decision.
The idea can be unsettling. A machine assessing our abilities seems cold, mechanical and impersonal. But before rejecting it, we should ask an uncomfortable question: are recruitment processes led by humans really that human?
Anyone who has looked for a job knows some of their common problems: repetitive interviews, tests that have little to do with the position, weeks without a response and interviewers who do not always understand the role they are trying to fill.
Selecting a strong professional requires more than knowledge of human resources techniques. It also requires an understanding of the industry, the business model and the capabilities missing from the team. In sport, we would hardly trust the selection of a striker to someone who does not understand the game. In business, however, it is common for the first assessment of a specialist to be conducted by someone with limited technical knowledge of that profession.
When this knowledge is missing, recruitment tends to rely on signals that are easy to compare: years of experience, academic qualifications, previous employers, keywords and salary expectations. The selected candidate is not always the person who can create the most value, but the one who performs best in a process that sometimes measures interview skills more than the ability to do the job.
This is where the machine enters as a competitor.
An AI system can interview hundreds of people without becoming tired, ask similar questions and organise information consistently. It is not in a hurry to finish the working day, and its criteria do not change according to its mood. For a candidate who is used to sending applications that seem to disappear into a void, receiving a fast and structured interview may even feel like a form of respect.
The study by Jabarian and Henkel (2026) helps us understand this possibility. The researchers analysed 70,884 applications for entry-level customer service positions. Candidates were assigned to interviews conducted by human recruiters, AI voice agents or a format in which they could choose between the two. In every case, human recruiters remained responsible for evaluating the information and making the final hiring decision.
The results were striking. Candidates interviewed by AI were 12% more likely to receive a job offer and showed higher rates of onboarding and subsequent retention, without any reduction in the productivity of those who were hired. The authors suggest that part of this difference came from the fact that automated interviews were more structured and consistent, allowing recruiters to collect more comparable information from each candidate.
The candidate experience was not clearly worse either. The average willingness to recommend the company was 8.97 out of 10 after an AI interview and 8.84 after a human interview. The difference was not statistically significant. Participants also gave similar ratings for comfort, stress and perceived fairness. They even considered the AI’s questions slightly more relevant, although they found the conversation less natural (Jabarian & Henkel, 2026).
This does not mean that the machine has won.
The study focused on structured, entry-level customer service jobs in the Philippines. In addition, 5% of candidates abandoned the interview because they did not want to speak with an AI system, and technical difficulties appeared in 7% of the interviews. These findings cannot automatically be applied to the recruitment of doctors, researchers, engineers, executives or professionals whose abilities are more difficult to measure.
Other studies have found less favourable reactions. Langer, König and Papathanasiou (2019) reported that highly automated interviews received lower levels of acceptance than interviews conducted through videoconference with another person. Participants perceived less social presence, less control and a lower sense of fairness, particularly when the interview could have real consequences for their professional future.
Deriu, Pozharliev and De Angelis (2024) reached a similar conclusion through three experiments. Candidates tended to view an artificial evaluator as less trustworthy, and this lack of trust reduced their willingness to accept a potential job offer. One of their main concerns was whether a machine could recognise the qualities that made each candidate unique.
The evidence, therefore, does not produce an absolute winner. Artificial intelligence may outperform a person when the human-led process is slow, improvised and superficial. A good interviewer, however, still has an important advantage: the ability to listen to a story, interpret an unconventional career path and recognise qualities that do not fit easily into a rigid classification.
Machines will probably not completely replace the people who lead recruitment processes. They will, however, take over many of the initial tasks: reviewing applications, verifying requirements, asking structured questions and summarising answers.
This will force human resources professionals to demonstrate the value they truly provide. Those who merely filter résumés and repeat standard questions have reasons to be concerned. Those who understand the business, know the industry and know how to listen will remain necessary.
Ultimately, a recruitment process is not human simply because there is a person on the other side of the screen. It is human when it respects our time, evaluates abilities that are relevant to the job and gives us an opportunity to explain who we are.
Perhaps machines are not coming to remove humanity from recruitment. Perhaps they are coming to show us how much humanity had already been lost.
References
Deriu, V., Pozharliev, R., & De Angelis, M. (2024). How trust and attachment styles jointly shape job candidates’ AI receptivity. Journal of Business Research, 179, Article 114717. https://doi.org/10.1016/j.jbusres.2024.114717
Jabarian, B., & Henkel, L. (2026). Voice AI in firms: A natural field experiment on automated job interviews [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2607.28222
Langer, M., König, C. J., & Papathanasiou, M. (2019). Highly automated job interviews: Acceptance under the influence of stakes. International Journal of Selection and Assessment, 27(3), 217–234. https://doi.org/10.1111/ijsa.12246