The Robots That Can Get You a Job
WHY YOU SHOULD CARE
Because we hated our second-grade career aptitude tests too.
By Anne Kadet
Over the past several weeks, I have learned the following fun facts about myself: My voice is engaging, but a little disturbing. I have a mediocre education and a terrible memory, but possess the focus of a boa stalking a mongoose. I’d make a great entrepreneur, but a lousy accountant.
All this insight and judgment have come courtesy of … robots. Clever recruiting algorithms, software programs and testing modules have analyzed every aspect of my workplace self, ranging from my risk tolerance to whether I’m likely to leave my current job. They’ve assessed my online presence, scored my gaming skills and weighed in on my vocal delivery like American Idol judges — all in an effort to determine whether I’d be a good fit at companies like Amazon or Revlon.
Yes, employers have finally admitted that for the last thousand years, they’ve been doing a lousy job of hiring. No matter how hard human recruiters try to be objective, they still bring their personal biases to the screening process. Some of these biases are pernicious, like racial prejudice. Some are simply goofy — I once got hired because I lived down the block from an editor. But they can all produce horrible hires. The robots have been called in to eliminate these biases. After all, what’s more dispassionate than a heartless computer?
Hence the arrival of new algorithm-based screening and sourcing technologies, which, a bevy of recruiting startups are betting, can do a far better job of reaching untapped labor markets and matching employers with candidates. These new firms have plenty of believers. Venture capital funding for recruiting technology startups has doubled since 2009, to $219 million last year, according to PitchBook, and more than 70 have launched since 2012. Never mind the fact that no one knows what results these technologies might produce. “There’s a general belief that if it’s automated, online and looks good, it will probably save us money,” says Steven Lindner, an executive partner at national recruiting and consultancy firm The Workplace Group.
But I’m not quite so concerned about whether these bots do a good job for recruiters as I am curious about what this brave new world means for me, a potential job hunter. Could the robots find me a great job? Or, for that matter, any job? I decided to let them have at it — I’d join them in their online games, and let them eavesdrop on my conversations and scrutinize my craziest workplace fantasies. I had nothing to lose, after all, but my time, my self-esteem and my most cherished delusions about my worth in the job market.
My foray into recruiting 2.0 started at the lower-wage end with Jobaline, a recruiting startup employed by companies like Hilton that pay $10 per prescreened candidate in their effort to source hourly wage workers. Jobaline’s claim to fame: It analyzes applicants’ voices to determine if they’d be good matches for customer service jobs. These days, you can’t just ask, “Want fries with that?” You need to ask with the customer-preferred vocal inflection.
Curious how I’d fare, I applied for a position through Jobaline as a customer service clerk. The site’s online form requested the usual info about my skills and work history. Then came the fun part. Immediately after I typed in my phone number, a friendly robot called to chat. “Hello,” said the lady robot. “This is a short phone interview. We will ask you a few questions. Relax and don’t rush.”
Which, of course, made me extremely nervous. Still, I managed to tell the bot about how I might placate angry customers by thanking them for their oh-so-helpful feedback. I hung up feeling optimistic. Surely an offer was headed my way. But after several weeks, I hadn’t heard back. This was alarming. As a freelance writer, I spend half my day on the phone, interviewing sources. Is my voice annoying?
Well, yes and no. When I asked Jobaline CEO Luis Salazar for an evaluation, he sent a bunch of complicated charts illustrating aspects of my voice. Jobaline technology analyzes more than 100 vocal attributes, including tone and inflection, then compares these characteristics to those of voices that customers find universally appealing, ignoring aspects like accents and gender. The good news, according to Salazar, is that the algorithm says my voice makes people feel “engaged, energized and wanting to hear more.” I’d make a fantastic salesperson or Walmart greeter. On the other hand, when it comes to sounding calm and soothing — key characteristics for calming cranky callers — I flunked out. No suicide hotline job for me!
What if robots could recommend the perfect employer, the way Netflix recommends movies?
I can take some comfort in the fact that the bot can’t judge the whole picture. Jobaline says its algorithm’s judgments coincide with those of the general population 77 percent of the time. “Generally, if something is valid three-quarters of the time, that’s not good enough,” says Lindner, the recruiter. “We want 90 percent or better.”
But for others, a successful test might be a life-changer. Salazar says Jobaline sometimes directs applicants to a more lucrative job than they might otherwise apply for. “If we see you applying for a hamburger-flipper job, you can make $8 an hour,” he says. “But what if you have a great voice and no one told you that? You could be making $14 an hour as a customer service rep.”
Thanks, robot overlords!
Vocal analysis is just the tip of the iceberg. What if robots could not only evaluate you for a given job, but also go one step further and recommend the perfect employer, the way Netflix recommends movies? Never mind the fact that product recommendations still generally disappoint — the race is on to do for hiring what Amazon does for books.
Introducing Pymetrics, a New York-based startup that says its online games can assess the key cognitive and emotional traits of any candidate in 25 minutes, then match her with the right company. There’s no such thing as a good or bad score on the games, says marketing head Alena Chiang. While impulsivity is a great trait for entrepreneurs, for example, “there are other industries that need a more measured, careful person.” So far, the model sounds benign. Rather than being used to screen applicants out, the tests are used to source folks who might otherwise never land on a hiring employers’ radar. Companies like Fidelity pay to access a database of 100,000 candidates who have so far taken the free tests on the Pymetrics website.
Putting aside the fact that your entire career may depending on the outcome of these simple games, the colorful tests are lots of fun. One had me deciding how much of a $10 windfall I’d share with a fellow player who would receive triple that sum and return as much or as little as he pleased. I offered the entire 10 bucks and received $12.30 back. “You’re trusting and it pays off,” Pymetrics reported at the test’s conclusion. I’m not sure, however, that I liked my overall profile. It says I’m a wild risk-taker, and totally naïve. I don’t get flustered by mistakes, but don’t learn from them, either. I’m extraordinarily focused, but have a paltry memory. I am, in short, a human fruit fly.
It would seem that for every problem, there’s a bot waiting to solve it.
Still, the Pymetrics robots say there are several occupations at which I’d excel, including corporate finance and product development. I’ve missed my calling, it seems, as a middle manager for P&G. Writer, however, did not make the list of appropriate occupations. Therein lies one of the fledgling technology’s shortfalls. I might be perfectly suited for journalism or bullfighting or firefighting, for that matter. But so far, the site has only gathered the data needed to assess fitness for a few dozen professions.
Then there’s the fact that while Pymetrics measures inherent traits, it doesn’t consider skills and experience. It’s a great opportunity for entry-level workers who have nothing to show on their résumé, says Lindner. But for an experienced worker, if your profile fails to match the company’s ideal, “it won’t matter what your resume looks like, you’ll most likely be eliminated.”
It would seem that for every problem, there’s a bot waiting to solve it. Meet Gild, a San Francisco-based recruiting technology startup that uses algorithms to assess a potential candidate’s skills, albeit in a narrow sector — software development. The outfit employs bots to crawl the Web, assessing coding projects developers have posted on open-source websites like GitHub and Bitbucket. It also incorporates peer evaluations to score each developer’s expertise. In the process, it assembles a database of several million techies that companies such as Sony use to source prospects. The idea is to “take bias and pedigree out of hiring,” says Robert Carroll, Gild’s chief marketing officer. “We’ve discovered people who didn’t graduate high school who are hugely talented.”
Can this approach be expanded to other professions? That’s what Gild is betting. It recently launched a new database of more than 100 million workers across every imaginable profession, from plumber to brain surgeon. While it can’t score every worker’s skills, it aims to compile sophisticated profiles by aggregating information pulled from sites like Klout (which rates a person’s social media influence), Twitter, Facebook, Reddit and Meetup. It can even detect when a person is ready to make a job switch based on factors like an updated LinkedIn profile, says Carroll.
Some might find this behind-the-scenes sleuthing a bit intrusive. Not me — it’s publicly available information, after all. But I was disappointed when the company let me peek at my Gild profile. So far, the bots aren’t scoring my online stories for grammar, or rating my tweets; it looked like a simple rehash of my LinkedIn page. It didn’t even report my 2012 arrest.
These recruiting startups offer useful sourcing solutions, but they have yet to revolutionize the hiring process, says Eliassen Group COO Tom Hart, who helps Fortune 500 companies select hiring technologies. While technology may aid the initial screening, “I don’t see it ever replacing humans,” he says. And so far, we have no idea how well the algorithmically sourced candidates perform after they’re hired. “I don’t have the data,” he says, “but, my God, I would love it.”
My favorite sourcing startup is Seattle-based Poachable, or as I’d like to call it, OkCupid for jobs. Folks looking for a new gig create an anonymous profile on the Poachable site detailing their work history and dream job. If the scoring algorithm finds a relevant position, it forwards you the hiring company’s anonymous profile, and vice versa. When both parties like what they see, their identities are disclosed: You’ve matched, and now you can chat! The hiring company pays Poachable $50 per “reveal.” Founder Tom Leung, like any good entrepreneur, says his service could “change how the whole labor market works.”
That change lies in getting job-seekers and companies to admit their secret wishes through Poachable’s anonymous shield, theoretically making for better matches. The problem with the wanted ads and résumés, says Leung, is that employees and companies alike are reluctant to list their actual preferences. For example: “You don’t want to be the guy insisting on vegan food at the office, but you can whisper that to us discreetly.” If Poachable finds a company offering vegan lunch, it’d rank the employer higher on your match list.
Candidates have nothing to lose by signing up for the service, says Lindner, as long as they update their profile and keep it authentic. And so far, more than 1,000 companies — including Amazon, Netflix and Facebook — have signed up for the service, while 35,000 job-seekers have created profiles. Make that 35,001. Creating my profile on Poachable’s sleekly designed site, I had an opportunity to include wishes I’d never reveal on a résumé, including my desired salary (six figures, minimum), dream employer (The New Yorker, of course) and fantasy job title (“Adventure Queen”). I even gave myself a great tagline: “Best Writer Ever.”
Alas, I’ve heard precisely nothing from Poachable’s clients. Leung says that’s typical. The algorithm is fussy, matching only the highest-scoring pairs. After that, a team of “human curators” reviews the suggestions, tossing obvious duds. Candidates typically don’t receive their first match for several weeks. On the other hand, Leung furnished me with my Poachable Score Report, an assessment of the demand for my services. It was flattering. Despite what it deemed as my mediocre education and dearth of specialized skills, I’m still an attractive candidate, with a Poachability Score of 88 — putting me in the 95th percentile for my field. The bad news? There’s not much demand for freelancers who specialize in writing about the new job application treadmill.
If I really wanted to maximize my desirability, said Leung, I should consider becoming a manager. Or would I consider a radical move? “If you transformed into a data scientist living in San Francisco,” he promises, “you’d have 100 matches.”
- Anne Kadet Contact Anne Kadet