Listen to this article
When you apply for your next job, it could be a computer algorithm, rather than a human, that asks you the first round of screening questions.
That is the hope of Eyal Grayevsky, whose company Mya Systems uses artificial intelligence and a chatbot named Mya to help recruiters screen high volumes of applicants for hourly wage jobs.
“When the candidate applies for a job, Mya introduces herself and takes them through a conversation,” he explains. The chatbot will ask questions to see if they are qualified: “Can you work in the evenings?” “Are you comfortable with the pay range?” “Do you have forklift experience?” “How soon can you start?” If the applicant is a good fit, Mya will schedule their interview.
While being interrogated by a computer might sound degrading, Grayevsky points out that it is much better than never hearing back at all, which is the most common outcome.
The idea for the company was born out of his own job search experience after university. He applied to 40 positions and heard back from only two of them. He estimates that about 85 per cent of job applications in the US fall into a “black hole” and the applicants never get a response.
Recruiters are struggling to handle much higher volumes of curriculum vitaes than ever before. It is partly because people are changing jobs more often and partly because the internet has made it so much easier for applicants to apply to a lot of jobs at once.
The flood of CVs is particularly severe for hourly wage jobs, such as in warehouses or retail stores, where a company might need to hire hundreds or even thousands of workers ahead of a busy season. These types of roles will be the first where automation and algorithms will make a real difference. “Recruiters are so bogged down,” says Grayevsky. “As a result of the ‘spray and pray’ mentality that job seekers have, recruiters often deal with huge amounts of volume,” he says. “When it happens it creates a bottleneck and it becomes really hard to properly manage all these relationships.”
That is where AI comes in. A growing number of tech companies are working on perfecting algorithms that mimic recruiters’ simplest tasks — starting with finding CVs for a position.
“Today we are barely scratching the surface of what can be automated,” says Guillaume Champagne, president at SCGC executive search, a recruiting company. SCGC has just started working on its own bot, which will be used for more junior searches.
Champagne says that for years, the company has outsourced some routine tasks, such as searching for suitable candidate profiles on LinkedIn, to human workers in countries where labour is cheaper, for example India. “For that research piece . . . a bot could very easily replace that part of the job.”
Many recruiters say that automation can be helpful for handling the more annoying parts of the job.
“That initial searching and matching and talking to folks, that is what takes most of the time,” says Kelli Dragovich, who has been in recruiting for more than two decades and is now a senior vice-president at Hired. “That is the most painful part of the recruiting process. And that is the part where we want to use machine learning and algorithms to make searching as painless as possible.”
At LinkedIn, which has more than half a billion professional profiles, the company has invested heavily in improving the quality of its search results. “It is just more powerful today, it allows those recruiters to be more productive,” says John Jersin, who heads LinkedIn’s recruiting software tool.
Now the company is working on going one step further and using an algorithm to ask questions of people who are searching for jobs, such as the location where they would like to work. “We want to get those [initial questions] out of the way because that is not the most important part of what a recruiter does,” says Jersin.
At a time when AI is just beginning to have an impact on white collar work, recruiting could be one of the first sectors to be affected. Over time, this could eventually mean fewer human recruiting positions.
The human recruiters who remain will need a slightly different skill set — one that is more focused on the tasks that the algorithms cannot perform.
Steve Goodman, chief executive at Restless Bandit, a recruiting start-up that uses machine learning to help companies tap into their existing CV pool, says recruiters are not about to become extinct.
“You still need a human to evaluate whether someone is a good fit for a job,” says Goodman. “If you are evaluating people for the purpose of culture fit, I’m not sure a computer can do that.”
It will be difficult for machines to replicate the live interview, read a candidate’s body language, or judge their personality or their values. But more rote tasks are already being handled by algorithms.
“If you are a recruiter that does nothing but source and put people into piles, then yes your skill set is getting automated away,” Goodman adds.
Others share similar sentiments. “What I’m worried about is those volume searches and junior recruiters, I think those people are at very high risk of being replaced,” says Champagne.
For the human recruiters who remain, a slightly different skill set will be needed — one that is more geared toward the things that humans are good at and that are hard for computers.
“Artificial intelligence will increase the need for a sales-type personality,” says Champagne. “More time will be spent trying to ‘close’ the candidate, selling them on the company culture, etc,” he says, referring to a candidate who has already been selected but who has not yet accepted the job.
As algorithms handle more mundane tasks, Champagne also expects recruiters will need a higher emotional intelligence and will focus more on value-added chores such as crafting a job description. “Recruiters will spend more time on consultative type of work, sales type work,” he predicts.
Executive recruiting will be one of the last areas of the industry to be impacted by algorithms, however. Even the most tech-friendly recruiter admits that a senior-level candidate would probably not take kindly to having their credentials questioned by a bot. That is one arena where the old-fashioned approach, in which a recruiter’s main assets are their Rolodex and their personal relationships, will not be threatened by AI anytime soon.
AI interviewers put bias to the test
One of the challenges of being human is subconscious bias. So as algorithms start to play a bigger role in recruiting, will the job selection process become less biased? Some recruiters who work with machine learning job search start-ups say the answer is yes.
Kelli Dragovich, senior vice president at Hired, says that one focus of the company is “making sure the algorithms inherently remove any biases”.
This task is trickier than it sounds because in other fields some AI systems have ended up exaggerating human biases, rather than removing them. One of the most famous recent examples is Tay (above), a chatbot developed by Microsoft. It spewed racist and hate-filled phrases based on the real-world Twitter comments that Tay was using to train itself.
Eyal Grayevsky, founder of Mya Systems, explains that the chatbot his company is developing has been trained to reduce bias by ignoring factors such as race or socioeconomic status or gender when examining applicants.
As the algorithms train themselves using real-world data, one factor that also gets excluded is the actual hiring decision. “We take out the hiring decision because that is inherently biased,” Grayevsky explains. “Instead we use things like retention, or performance” to train the algorithm.