HR2D2 and CVPO: AI in Recruitment

HR2D2: A droid that works in HR. I know that’s lame but it’s extremely difficult to be funny about HR. Try it; all attempts welcome in the comments.

You’re Not What These Droids Are Looking For

Are you worried that Artificial Intelligence will take your job one day? If so, are you angry about that? Well, it may just be best to keep your complaints to yourself because, while AI may not be able to do your next job just yet, it has been bestowed the power to decide if you can – so better not upset it. AI Recruitment is here.

Unilever, an organisation so enormous as to be invisible to the naked eye, recognisable only by brands you neither love nor hate but without which modern civilisation would collapse tomorrow, is taking the term ‘faceless corporation’ to a whole new level. The first interview in Unilever’s recruitment process is now carried out, not by a person, but by AI in the form of the HireVue platform. They’re not alone either. If you apply for a position at Vodafone or Urban Outfitters, you’ll encounter the software there as well.

How Does AI Recruitment Work?

It would be easy to get carried away and hysterical and start referencing dystopian sci-fi movies (and if anyone is developing an AI for firing people, please, please call it The Terminator) but, really, this development is both inevitable and sensible. Yes, it will have an impact on hiring practices but some of this could be positive and, for those who are willing to do a little research, may even give them an advantage. Consider this post the beginning of that research. 

Currently, the most widely-used program used for AI recruitment is HireVue and it’s used to screen applicants near the the beginning of the process. When you apply for a job online, if you make it past the first stage, you are sent a link and asked to download an app in order to conduct the video interview in your own time. Interviews can last between fifteen and thirty minutes but they suggest allowing forty-five. If you’re successful here, you make it to the next stage, which is usually a face-face-interview or selection day. The main thing to remember is that the AI is there to help large organisations sift through huge numbers of applicants; this has some implications for your strategy, as we’ll look at later but, first…

Haven’t Systems Like This Been Shown to be Prejudiced?

Not exactly, although there certainly have been some problems. The most famous case, which made the news back in 2018, was when Amazon started to use their own in-house system for selecting new recruits. The problem was that the system was mainly selecting men because that’s what it had been told to do. OK, it hadn’t actually been told to penalise women applicants; the problem was that the program was looking for candidates who were similar to successful employees already working at Amazon. Most of those employees were men and so the system ‘thought’ that being a man was an indicator of strong performance at Amazon. The interesting question here, of course, is, ‘why did Amazon mainly employ men under the old system?’ Well, that’s for a whole post entirely and the situation is similar throughout the tech industry, but it does behoove us to mention that…

Human Recruiters Are Already Horribly Prejudiced

If you think that, as things are, we more or less have equality now, then this might come as a shock: In the US, white, male, convicted criminals are more likely to be selected for an interview than black males who have never committed a crime. No, I didn’t write that sentence in the 60’s, I wrote it in 2019 and I’m going to pop it here again: In the US, white, male, convicted criminals are more likely to be selected for an interview than black males who have never committed a crime. Things look pretty similar in the UK as well, so it’s not an isolated problem.

This means that, at this point, an AI system which is not actively rejecting applicants solely because they are black is a massive improvement. The other benefit with AI here is that if it’s learning prejudices from its human predecessors in recruitment, it can be programmed to unlearn them; changing ingrained human beliefs is not so simple. It should also be mentioned that, certainly with HireVue at least, it isn’t judging you on your appearance but on your words and facial expressions. There are still question marks hanging over the use of AI recruitment (and AI in general), though, and one of the more important of those doubts is about transparency.

How Does an AI Recruiter Account for it’s Decisions?

The short answer is that it can’t. First of all, it doesn’t actually make decisions in the sense that a  human would understand them, it just looks for traits and mannerisms which resemble those of its  ‘ideal candidate’ – a composite of all the strongest performers in any given role. People who demonstrate those characteristics move on to the next stage of the process. Secondly, it doesn’t actually know why those traits and mannerisms are shared by strong performers in that particular role. To understand that, we still need good, old-fashioned human intelligence.

There are concerns that the software will only reinforce current biases as, no matter how sophisticated its algorithms get, it bases them on people already in the workplace. While this should preclude selecting on the basis of gender, race or any other identity, there is undoubtedly a cultural element to tone of voice and mannerisms; if people from a particular cultural background tend to share a ‘favourable’ mannerism, they may be at an advantage. Again, though, the AI will only be learning from the data that it’s given and at least the resulting algorithms can be scrubbed clean of any identity biases learned during the training.

As things currently stand, then, the jury is still out on whether AI will reduce or increase discrimination in the workplace. Like with all implementations of AI, there are still many unknowns.

OK, But Can I Game The System?

If you work out a way to do it, let us know how; we’ll employ you on the spot. The fact is, unless you know exactly what the AI is looking for, you can’t trick it. You can plan though.

As far as candidates go, the best way to prepare is the same as with any other interview. Have a good idea of the questions you are likely to be asked and make sure you answer them in a sincere, convincing and memorable way. Practise with someone you trust or hire a professional to make sure you are operating at your full potential. The competition will probably be very stiff as AI recruitment allows companies to field a lot more candidates but, then again, the convenience of applying online and interviewing from home means you can apply for more positions. 

Finally, check out HireVue’s website for a detailed primer about what to expect. If you do have an automated interview soon, please let us know how you got on in the comments below.

Ciao for now!

Why not connect with me on LinkedIn here?

Also, don’t forget to check out our Interview Tips: The Top Ten

Finally, if you want me in your corner, landing you interviews and then helping you crush them, check out our services here.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: