By Nader Sleiman, Analyst
‘Could AI decide your job fate?’ recently led my LinkedIn homepage in the News and Views section. The storyline included a link to an article in The Telegraph entitled ‘AI used for first time in job interviews in UK to find best applicants’. As the article went into detail on how AI would be introduced into recruitment practices, and the positive and negative impacts it might have, the ethics behind introducing this technology in selection were put in question. Despite our fascination with introducing technology to the workplace, there are points when one must pause and reflect on when and how technology is used and whether this use is adding value. I hope that what follows serves as an eye-opener regarding what AI means for recruitment today, and why it is still too early for this tool to be adopted for the purpose of assessing people’s candidacy.
Is AI in recruitment simply ‘good’ or ‘bad’?
The Telegraph article introduced both sides of the argument: the side presenting a positive view of AI in recruitment, and the side that saw the flaws AI demonstrated in this area only in 2018 when Amazon shut down their recruitment AI project for racial and gender bias. Technology has grown to play a crucial role for HR processes. Information systems such as Workday, Oracle’s Taleo, and SAP’s SuccessFactors have facilitated the automated side of HR, paving the way for a focus on process improvement. Even AI offers recruiters great benefits, such as facial recognition that detects candidates’ emotional state and body language during recorded interviews to identify personal characteristics and quality of information delivery. Similarly to its simpler predecessors, AI reduces time spent on HR procedures and allows recruitment professionals more time to focus on developing the process. This technology also could contribute to reducing subjectivity, as well, but it is yet to achieve this level of advancement.
Adopting AI remains a questionable approach to recruitment, largely because of its history of bias. Dr. Muneera Bano of Swinburne University of Technology points out that AI’s gender bias is the result of historical gender discrepancy in cyber content. As AI gathers information, the biases expressed by human beings are integrated into AI’s assessments, threatening the chances of women being fairly assessed. Joy Buolamwini’s research found that IBM, Microsoft, and Amazon’s AI recruitment systems failed to even identify the gender of famous African American women, such as Michelle Obama, Serena Williams, and Oprah Winfrey. The result, therefore, is an AI preference for male candidates over female candidates in the assessment, which could lead to dismissing qualified female candidates of colour solely on the basis of race or gender.
Reflecting on Interviews as a Selection Tool
Although automation is without doubt a key part of the future of work, the automation of HR professionals’ selection practices, particularly the practice of interviewing itself, must be re-evaluated.
The predictive validity of interviews, in their various types, has been discussed and questioned by HR researchers for decades. This is only partly because of the possible subjectivity of the interviewer, but it is also because of the differences that may appear in performance from the same person under varying circumstances. Numerous circumstantial variables can affect a candidate’s performance in interviews, making this selection technique less representative of candidates’ skills. For that reason, businesses need to focus on more than just the technological aspect of the future of work, as the process itself could be flawed to begin with. In other words, introducing technology to a flawed process, whose flaw has little to do with its use of technology or lack thereof, cannot improve it.
As automation and technological changes make their way further into the way we operate as organisations, we must never forget to question and improve the core of how we do things. The most important element will always be the human resource. If you would like to discuss recruitment practices and how technology can currently contribute to selection practices, please feel free to send me an email.
 Dastin, Jeffrey (Oct 10, 2018). Amazon scraps secret AI recruiting tool that showed bias against women. Reuters.
 Burbeck, E. (1988). Predictive Validity of the Recruit Selection Interview. The Police Journal, 61(4), 304–311.