The artificial recruiter

Seedlink’s Rina Joosten-Rabou says the power of AI can create happier workers.

It doesn’t take much courage or brilliance to predict that artificial intelligence will play a large role in the future of recruiting.

Already, about a third of companies are using some element of AI in their hiring processes, and popping up to serve that market is a growing field of startups such as Hirevue, Mya and Seedlink Tech.

A better question is how dramatically a tight job market in the US and elsewhere will accelerate the growth of AI-enhanced recruiting. The internet has made it easier to apply for jobs, but that development has swamped recruiters with too many applications to process at human speed, especially when crucial openings need to be filled.

The result can be disgruntlement among job seekers whose applications never received a personal response, and wrong hires by recruiters who, unable to quickly distinguish the best candidates from among a vast field, may lean on biases toward, for instance, graduates from particular colleges.

What AI can do at nearly the speed of light is separate the most serious candidates from those with a fetish for job seeking, identify those with the most potential and further distinguish those with the particular skills, qualities and desires for the job at hand. It can also issue to each rejected candidate a response that includes a personal detail and, if appropriate, an encouragement to apply again for possible different openings.

Still to be determined is which approach will prove most effective and popular. Does the AI-driven video interview provide data and nuance that written programs can’t replicate? Or does video favor theatrical extroverts – who aren’t ideal for every job – while possibly perpetuating recruiters’ physical-appearance biases? Might programs that analyze applicants’ written responses prove adept at identifying solid candidates, while excluding possibly great ones that don’t fit a particular profile? If young swimmers with Olympic potential were selected based on the efficiency of their strokes, then Janet Evans – with her unorthodox style – might never have won four Olympic gold medals.

Proponents of AI-enhanced recruiting stress that computers will not replace in-person interviews, that hiring decisions will always be made by humans and that digital interviewing greatly expands the field of potential candidates. Perhaps most importantly, they say, AI-enhanced recruiting identifies candidates who would have been overlooked in the conventional process, and who turn out to be extraordinary employees.

Rina Joosten-Rabou, a co-founder and top executive of Seedlink, a Shanghai- and Amsterdam-based firm that is currently expanding to the US, spoke to us about the growing power of technology to create unforeseen opportunities for businesses and their prospective employees.

Though we’re talking by phone, I feel as though I know you because I watched your TedX talk.

Ha! But don’t let your brain fool you.

What do you mean?

Our language correlates with our personality traits, our personal preferences and our behaviors. So basically our language is a mirror of ourselves, right? And most of us are prone to affirmation bias, to preferring people who remind us of ourselves.

Our brains make these judgments very quickly. Men determine in six seconds, and women in 12 seconds, whether they like a person or not. During the hour of conversation that follows, the brain is only trying to confirm that first impression. As humans, we are very limited in how we look at the world.

AI can resolve that?

Computationally, you can build up large libraries of language data, far larger than any human brain, and with that data you can predict behavior and performance far more accurately than could any human brain. As I said in a PwC white paper last year, our application makes hiring 10 times faster, increases retention by 25 percent and 25 percent more applicants are interviewed.

Is PwC an investor in Seedlink?

No, it is a client of ours. And we are a PwC technology partner.

How does Seedlink work?

For example, L’Oréal, the French cosmetics company, has been with us since the beginning of our existence. If we want to understand what leads to great sales performance within L’Oréal, then we collect natural language data from people who have been successful over time in sales within L’Oréal.

Using the algorithms we’ve built, we can learn what it is that these successful people have in common, and build a model against which to compare recruits. Right now, we need around 50 employees to create a successful model for a client.

How does that work from the candidate’s perspective?

Once you apply, you would receive an invitation for what we call a digital interview, which is sent to you either through mobile or laptop. And you open that digital interview, and you’re asked, let’s say, three open-ended questions to which there is no right or wrong answer. For example, would you choose to live in the future or the past? We would need about a hundred words for your response. We’ve found this to be especially popular with Millennials. For them, a CV is just a piece of paper. But if they can actually share their thoughts through these questions, then they’ve contributed something significant.

This model can identify candidates who otherwise might have been overlooked?

Absolutely. One example is a young woman in China named Laurel Sun. Out of 33,000 applications, her answers to the questions scored sixth. Based on that score, she was interviewed, and in the end she got hired by L’Oréal. She’s now one of their highest-performing employees in e-commerce – with no e-commerce background at all. And L’Oréal has said that under its previous system, it probably wouldn’t have hired her. She didn’t come from a top-tier university. And she hadn’t done an internship with a L’Oréal competitor.

That PwC white paper talks about HR applications of AI far beyond recruiting.

We can very much help people within a company find the ideal role for themselves, where they will thrive beyond their own expectations. If we can help companies make the ideal use out of their existing employees, all the way up to the senior level, that will accelerate growth and performance.

We also have predictive models that can help with customer service. Whether we’re working with candidates or employees, we might pose the scenario where a customer comes in to shop and wants to buy a gift for a friend, but hasn’t decided yet. How do you go about helping this person so that he leaves the shop happy?

Is it true that AI-enhanced applications offer improved feedback to applicants?

Historically many applicants did not hear back. Or they would get a standard email that said, “Thank you very much for your application.” But that’s not what I want as a candidate. As a candidate, I would like to know, “OK, can you give me some insights on why I wasn’t the right fit?”

We can offer helpful feedback, and we can offer it more or less instantly, instead of in a few weeks. My goal really is to see if we can offer this technology to everyone out there, so that we can help them find the jobs where they would thrive, and where they would be a great fit.

Anne Bark is a Director in Brunswick’s Shanghai office. Additional reporting by Emma Cole, Recruitment Manager in Brunswick’s London office.