How intelligent are hiring tech tools? Pretty dumb, if you ask me!
A few weeks ago I got some emails from recruiters. They had shortlisted my resume for jobs in the hospitality industry. Frankly, it looked like a bad joke. Not because it was from an industry where I haven’t worked before but because this is one sector about which I’ve not written a single paragraph in my 18-year writing career. Above all, I am not actively seeking a job and I’ve not posted my resume on any job portal. So, to get some emails in quick succession was puzzling.
Around the same time LinkedIn added to the conundrum by sending some pretty unrelated and unsolicited job offers (of material and computer science researchers) and endorsements (from stock market folks).
Just as I was recovering from that experience, bewildered that the explosion of Applicant Tracking Systems (nearly universal in recruitment) and human resource management tools in the last decade haven’t done much to match the job with the job seeker, an entrepreneur friend had a hilarious story to share.
Here’s his experience with a candidate who came for a j2ee developer’s job. Out of 8 resumes sent by the recruiter, this start up founder selected one and called the guy for an interview.
The candidate came, well-dressed and exuding enthusiasm for the position. As per the norm, he had to begin with a written test which this startup designs on the fly based on the initial conversation with every candidate and the resume.
The test was: To complete the high-level design and one component code. If he could do detail-design of a different component then it’d be a bonus.
The candidate starts his test and after an hour he had two simple drawings.
Is that all?, the founder asked.
Well, what else; I have drawn the high-level architecture of client-server and 3-tier architectures.
Please explain, pressed the founder.
For the client-server, I have a client and a server and for the web I have a thin client, a server and a database, the candidate pretended to explain.
So, what is the difference between the two clients?, the founder egged-on.
Well, one is thick [he gestures with his two hands stretched] and the other is thin [this time pinching the air].
The founder was flabbergasted: Sorry, I don’t get it.”
“It’s the Java Messaging Service (JMS) that makes up the difference between the two. You know JMS? he asked
Not that much, tell me something, the founder said.
Who doesn’t know about JMS? You must know otherwise you don’t know anything about web-apps, the candidate elucidated.
Really, I don’t. Tell me something about web-apps, the founder persisted.
You should find out. I cannot teach you web-apps now. JMS may be,” the candidate said.
“Anyway, tell me something about design pattern related to JMS, the founder said patiently.
Design is easy. I do that every day but why are you looking for patterns in it?
By then the startup folks had lost patience and thanked him for coming.
Those who recruit for IT industry will tell you how tough it is to filter such jerks through online or in-person screening processes.
The above two examples, one of a staunch writing experience (at least until now) and the other of a straight IT job description, are indicative of what’s wrong with the current recruitment technologies. Almost 90 percent of Fortune 500 companies use ATS, yet the time and cost per hire is increasing. As this May 2014 Wall Street Journal story shows, the average number of working days taken to fill a position has been steadily increasing since the economic recovery in late 2009. This is worth examining because a developed market like the United States uses a variety of technologies”from ATS to online assessment tools (which assess critical thinking as well as technical knowledge), recommending and ranking sites to video cover letters.
Practically, both resumes and ATS are dead. The latter scans paper resumes into a database, does some basic screening and tracks a candidates path through the hiring process; the former is a self-proclaimed piece of truth, half-truths, and aggrandizement. Not surprising why most recruitment companies say they are using social media to glean relevant information about the candidate. It is estimated that more than 90 percent of employers will use social media to hire this year.
Last month LinkedIn unveiled its sixth standalone app for job searches. Yet, if you’ve used LinkedIn for a while you’d know how off-the-mark its job suggestions are. At least I don’t know of anybody who’s found a job via LinkedIn. A few times that I’ve tried reaching out to some people (not so high, that the pecking order to be offended) for my journalistic work, I’ve hardly met with success.
The point I’m trying to make is that the herd mentality in tech development or adoption is once again at work. Most of these technologies are based on key-word searches and matches, which effectively exclude the weakest choices but don’t select the strongest talent. They look for the right buzzwords in the CV, a certain format in the cover letter but totally miss the idea of what makes a candidate right for a particular job. More so, exceptional talent doesn’t come in nifty packages. For that the automation, which today has led to industrialization of resume screening — several recruitment companies boasts of screening a million CVs per second– needs to acquire a human touch.
But it isn’t that easy. Text mining itself is tough, nothing like data mining we hear more often. In the latter the input is machine comprehensible and the output has to be human comprehensible, whereas in text mining the input stuff, that is the text, is already human comprehensible. It needs to be converted into something which is understood by the machine and then regurgitated in a form that the humans want. Then to have semantic matching to get the really right fit “ a combination of different expertise with XX number of years , preferably in XYZ places, etc. It’s like looking for a needle in a haystack.
In spite of a few decades of impressive work in artificial intelligence, machines don’t think and choose like humans. However, they can be made to learn to do so to some extent. I think automation will and should increase so that the mundane tasks in hiring are done by machines and algorithms, preferably self-learning ones which understand the user, freeing up hiring managers to spend more time with the candidate. They can then devise intelligent scenarios during the interview to test hidden skills, intuitive abilities and competencies. Even Google, famous for its brain-bending interview questions, has come to realize that hiring requires a more human touch than the interviewer coming overly prepared with questions which at best test a candidate’s presence of mind or smarts. (Google is now moving to testing behaviour in job interviews.)
Anything to do with human capital can’t be straight-jacketed. For every Bruce N Pfau and Ira Kay, authors of The Human Capital Edge who advise employers to go for people who’ve done exact job, in this exact industry…from a company with a very similar culture there’s a Lauren Rivera. A cultural sociologist at Kellogg School Management, Northwestern University, Rivera, who after studying recruitment practices at global law firms, investment banks and other places, has concluded that most often it’s the sharing of leisure interests with the hiring manager that clinched the job for employees.
Even though keyword searches and matching technologies have come a long way in screening talent, the future is about analytics which can study the past and present hires and predict who’s going to be successful new hires. In real world use, recruitment technology is a few years away from that. Employers will have to beef up their in-house analytics team or open up to third-party analytics provider and get their managers to work with the machines, be it in feeding their comments into the system or playing around with analytics to spot the right talent. One HR manager at a Fortune 50 company says while there’s a big difference between skill and competency, most interviewers in his organization lack the ability to distinguish between the two.
Knowing about the precise skills is a checklist, easy but essential to do. Knowing the experiential learning part is difficult to detect; it requires insight, listening skills, curiosity on behalf of the interviewer. Salary and bonus are used as the most important tool for retention, he admitted.
But as many of us know, even before reaching the self-actualization part in Maslow’s pyramid, people begin looking for subtle incentives and motivation to perform or hang in there. Today, the most advanced self-learning neural network can only aspire to perform human tasks like understanding true creativity and sensitivity. Still, we could look for technologies that allow time and confidence to the interviewer and interviewee to sit across the table and discuss, say, the defining Germany- Brazil match in FIFA 2014. Or, who knows your attitude to Uruguay striker Suarez biting Italy’s Chielleni on the field may just get you your dream job!