Are robots already bringing greater objectivity to recruitment? Attractive people are more likely to get hired and earn 10-20% more than those with asymmetric faces, a University of Buffalo study found.
Professor Anthony McDonnell of CUBS at UCC says HR specialists are putting artificial intelligence to good use. He talks to Rita de Brún of the Irish Examiner.
Look in the mirror. If you’re going for a job anytime soon, let’s hope you’re looking good.
No, make that great. Better still, attractive. Why? Because attractive people are more likely to get hired and paid more. That’s why.
This uncomfortable, but far from surprising, fact was confirmed recently in a University of Buffalo study. Worse, or better — in accordance with that reflection you saw in the mirror — the venus and adonis set can expect to earn 10-20% more than their asymmetric-faced counterparts.
While chewing on these inner confidence gnawing facts, you might ponder who the beauty-worshipping hiring set account to. As in: Who’s watching them watching us? And if the answer is robots, is it the programmers behind their antics that are to blame?
Throwing light on these sorts of conundra and indeed the mystery that is bias in an increasingly AI-led recruitment process, Anthony McDonnell, professor of human resource management, at UCC’s Cork University Business School, is a font of knowledge.
On some of the merits he says: “They speed up the hiring process, and provide for fairer and more objective decision making.”
Speedy, fair and objective. This sounds good. But as we’re talking AI, algorithms come to mind. While that’s a word that’s constantly bandied about, it’s one we might all benefit from better understanding, given the power they yield over our lives
Expounding on the topic, Prof McDonnell says: “An algorithm is basically a very precise recipe that sets out the sequence of specific steps required to solve a problem or provide a solution. Therefore, the design is fundamental and who is or is not involved in such development is crucially important.
“If appropriate expertise from different perspectives is not involved, the dangers and drawbacks may vastly outweigh the possibilities of AI in hiring.” Explaining that it is in this context, that concerns are mounting over the often-assumed bias-free nature of AI-enabled hiring, he says that at the heart of the disquiet, is how the algorithms are designed in the first place and by whom.
“Simply put,” he says, “if built into the design of the algorithm, it is likely that bias could be magnified rather than being eradicated or reduced, given how such technology works.”
Confirming that discrimination and bias are matters with which HR professionals have learned much over the decades, he continues: ”They have become attuned to this and have been progressively implementing better structures and processes, to reduce bias in hiring decisions and to improve diversity of applicant pools and hires.”
Acknowledging that it is "now accepted good practice to have diverse interview panels rather than a single interviewer or a homogenous interview panel," he continues: “Similarly, it is good practice that all interviewers are trained."
Musing then, on the question of how to ensure these good practices are similarly incorporated into algorithmic evaluations, he says: “This may be aided by those involved in the design and the regular evaluation of how the algorithms are performing.”
Emphasising the importance of human expertise in the increasingly AI-driven hiring process, Prof McDonnell says: “If HR algorithms are developed without the input of key stakeholders, such as HR professionals, we might find that human biases are inscribed and more deeply embedded, and thus increase inequality.”
On the evidence of gender bias in the use of AI in early hiring assessment decisions, he says: “This has been ascribed to how algorithms using historical employment data may evaluate men as primarily holding managerial and leadership roles, and women less so.
This may therefore lead to the increased exclusion of women in the screening process.
“Similarly, if the algorithm is set to evaluate employment gaps, this may also negatively impact women who take leave, in greater numbers than men, for caring responsibilities. Often, the challenge is around the use of proxy measures that can have unintended consequences.”
Prof McDonnell says that "if recruitment algorithms are focused on attracting potential applicants with particular interests and backgrounds, there may be a limiting of under-represented groups".
He notes: “This will be because they are less likely to be made aware of those employment opportunities, given those interests and backgrounds, and how the algorithms — that bring them to one's attention or not— work.
“If such algorithms had a match between application and qualifications based on past organisational hiring, what you may see is a reproduction of the past individual which impedes more diverse talent pools. In effect, we know that these people have worked before for us and the technology facilitates getting someone along highly similar lines.”
While AI offers much potential in improving several aspects related to the world of work, Prof McDonnell says technology in itself does not lead to better, fairer outcomes: “The need for diverse, multi-stakeholder input doesn’t decrease due to the capability of AI. If anything, it makes the need for this — especially at design stage — more important, given the way this technology works.
“So, the use of AI in hiring does not automatically mean better and more objective decision-making and outcomes; it could be the opposite.”
Referring then, to the evidence around attractive-looking people being more likely to be positively evaluated in a traditional hiring process, Prof McDonnell continues: “In the context of AI and the use of facial factors or variables, some research indicates that in terms of interview ratings by human managers, an applicant’s attractiveness and first impression confer a similar advantage, whether video interviews are live or asynchronous.
“When compared with the synchronous mode, the asynchronous mode may decrease the initial impression primacy effect, when evaluating competencies via interview. As a result, misinterpretations may increase when using AI for evaluation. However, the evidence is still emerging on the validity and robustness of facial recognition algorithms.”
For some reason, that sets me thinking of the words ‘…serenity to accept the things I cannot change, the courage to change the things I can…’ and remember that prep is something we can do prior to a video interview.
Noting that "issues with audio and video signals can significantly compromise the interview process, as well as increase anxiety for those involved," Prof McDonnell say this can lead to negative perceptions for both individuals and organisational decision-makers, "depending on the attribution of blame for such issues".
He adds: “They may also have wider implications, in terms of how individuals feel they’ve been treated and how they feel about the employer brand.”
He’s right. With the volume of digital voices through social media and review websites growing louder by the day, the employer brand is more vulnerable than ever.
Given the convolutedness of the evidence relating to fairness and bias in an AI-impacted hiring market, it’s little wonder that that complexity is mirrored in human attitudes to digital interviews.
Throwing light on the matter, Prof McDonnell says: “Emerging evidence indicates that while individuals may view algorithmic evaluation as more objective, there seems to be a desire for ensuring that the human element, or touch, remains very much involved in evaluation decisions. In addition, applicant perceptions appear less favourable to asynchronous methods.”
Perhaps we shouldn’t be surprised that candidates might have little appetite for pre-recorded interviews, they being one-way, filmed recordings of them answering pre-set questions. That’s like talking into a void. Something that still feels a little alien to us. For now.
Tips for video interviews
- Know everything there is to know about the hiring organisation.
- Know precisely why you have what it takes for the job.
- Believe in yourself.
- Exude confidence.
- Do a trial run to practise speaking to camera.
- Ensure your pace of speaking is not too fast.
- Check your mic and webcam are working.
- Ensure your WiFi signal is strong and stable.
- Make sure your battery is sufficiently charged.
- Look into the webcam not at the screen.
- Keep your notes out of view.
- Join the interview early.