Valentine’s Day: Can Artificial Intelligence Predict Office Romances?

How these algorithms work and what they are capable of predicting

As Valentine’s Day approaches, singles across the globe will be eagerly scrolling through dating apps, hoping to secure a date for the big day. But could the algorithms that match budding lovers predict who amongst our colleagues will become love-struck – and with who?  

Office romance is still taboo - 84% of employers are not happy with their employees dating. Yet, it is nearly twice as likely you will meet your partner at work than via online dating, according to Totaljobs.

When delivering our artificial intelligence (AI) recruitment platform to a new client, the conversation often turns to how our algorithms can predict the affinity between teams. And, from time to time, a client will ask about how narrow or broad that affinity can go – forty people teams? four? What about two?  
 

Now, since it’s approaching Valentine’s Day, I thought it would be a perfect time to clarify two questions that sometimes occur when we discuss AI and predictions of ‘chemistry’: no, AI recruitment algorithms should not be used to predict office romance. But, technically, could they? The short answer is, yes. 

There are a few reasons I’ve decided to explore this topic further, rather than dismissing it outright on the grounds of poor ethics. Firstly, it’s not a facetious matter – some 48% of us will data a co-worker at one point or another and 59% of work relationships lead to resignations. Perhaps you remember the highly controversial decision taken by McDonald’s to fire its chief executive for being in a consensual relationship with a colleague: “poor judgement” that had “violated company policy”, according to the fast-food giant.  

Then there is the ethical question, which can either be ignored or confronted and discussed maturely. Since AI is an increasing presence in the workplace, and, given that we know some companies have strict policies regarding employee relationships, it is worth exploring how these algorithms work – and what they are capable of predicting.  

Dating Data

By 2008, I already had around five years’ experience of developing data that could assess behaviour in the workplace. I had begun research and work at the dating site Meetic, working on an affinity algorithm that could match couples based on their psychological and behavioural compatibility. This was four years before anybody had heard of Tinder.  

A year later, I created an online orientation service for Studyrama, used by more than 2,000,000 students. Then, in 2012, I released predictive algorithms that enabled AssessFirst to help recruiters find candidates that will succeed, thrive, and be motivated to develop in their work. 

So, when I claim that AI is capable of accurately predicting the outcomes of relationships in teams and between individuals, I don’t make the statement lightly. Nearly 10 million people in more than 40 countries have benefited from the predictions and systems I helped develop. 

And, be in no doubt, the algorithms and data that found your partner on your favourite dating app is a close relative of the data currently revolutionising recruitment. Should that worry or deter the implementation of AI recruitment technology? No, it should not. Here’s why.   

High Performers  

While the principles of affinity prediction were born from data app algorithms, there are crucial differences. An AI recruitment platform is not just Bumble transported wholesale to an HR department. Rather, the algorithms seek to match work-oriented compatibility.  

There are two things that we do not do:  

1. Hire clones - individuals can have vastly different personalities and backgrounds yet have a small set of common traits that correlate to - and predict – an efficient approach to their anticipated job roles. This approach ensures specific personalities are not considered – just specific traits – and it fosters greater diversity than more traditional recruitment methods. 

2. Consider physical appearance and preferences - with this crucial element of ‘romance’ prediction removed, a company is never able to accept or reject a candidate based on whether a high workplace affinity with another employee would translate to a romantic affinity. 

AI-powered recruitment is a much ‘blinder’ process than seeing a CV and prioritising a person’s stated experience and references. Theoretically, sound hires can be confidently made without the need for an employer to subjectively validate whether a new candidate will ‘fit the culture’ or ‘get along with their boss.’ 

While we can predict the affinity of two people to be compatible, the algorithmic match is not about matching soulmates – but about bringing together high performing, high potential people that keep your business growing. And we’d all swipe right to that. 

David Bernard is the Founder of behavioural assessment and AI recruitment firm, AssessFirst. Visit AssessFirst Connect David Bernard