AI: putting the humanity back into the hiring of human ‘resources’

By 25 October 2018 2 Comments

By Emlyn Nardone, Senior Program Manager, K-12 Education

As companies in general and especially in the tech sector scramble to find the right talent in Canada, with all the requisite digital skills and experience, one barrier exists that hurts both job seekers and companies: The Automated Tracking System (ATS). If you haven’t heard of an ATS, it’s essentially a software that filters resumes for keywords and matches them with the job description in a posting. Most large corporations utilize applicant tracking systems. Jobscan research found that 98% of Fortune 500 companies use ATS while a Kelly OCG survey estimated 66% of large companies and 35% of small organizations rely on recruitment software.[1]  Essentially these systems digitize the early 20th century version of a job application process by breaking roles down into lists of tasks, duties, credentials and a history of where a candidate worked. From the employers’ perspective, they are generally satisfied with how this system functions. The automation delivers a quality, efficient, cost-effective outcome. From a candidate perspective, there is a very different story though. Could it be possible that many of the difficulties associated with finding the right candidates with the appropriate skills and experience could be partially explained by a broken application process? Let’s look at some of the facts:[2]

  • According to a Capterra report, 75% of recruiters and hiring professionals use a recruiting or applicant tracking system.[3]
  • The same report notes that 94% of recruiters and hiring professionals say their ATS or recruiting software has positively impacted their hiring process. Only 4% say it has had a negative effect.[4]
  • However, according to Jibe, job applicants disagree. Fully 80% of candidates described their online job search and online job applications as stressful.[5]
  • In the same Jibe report, another 60% of job candidates are unable to complete online applications due to encountering tech hurdles.[6]
  • Forbes mentions that as many as 75% of qualified job applicants are rejected by ATSs due to spurious reasons like incorrect resume formatting.[7]
  • A detailed 2017 report by CareerBuilder states that 60% of applicants quit an online application because it was too complex or too long.[8]
  • According to the same report, 70% of applicants feel an online application should be five steps or fewer, only 40% of employers think the same.[9]
  • And unsurprisingly the same report notes that almost 40% of recruiters and hiring managers have not gone through the job application process on their own website to test it.[10]
  • Bersin and Associates recount a test where one company created a perfect resume for an ideal candidate for a clinical scientist role, it scored a mere 43% relevance because the ATS it was submitted to misread it.[11]
  • Lastly, ERE Media discovered in another test that one large firm saw 3 out of 5 of their top engineers being screened out automatically by their ATS as not relevant.[12]

Clearly, there is a disconnect between what the employer is getting from the process in terms of outcomes and the experience of the candidates. ATS systems allow companies to filter through thousands of applications and isolate only the top 1-5% before any human even becomes involved. They assign rankings to job titles, work experience at renowned Fortune 500 companies and the quality of a school that awarded a candidate’s credentials. Aside from the multiple duplications for adding your personal details, education and work experience or having to tailor your resume and cover letter to align with job description keywords for every application you make, a candidate must also contend with postcodes and phone number prefixes being used as filter options by the system. If you live in the wrong city you can be kicked out of the running. If you worked in ‘Account Management’ but the job description asks for ‘Account Manager’ you’ll get kicked. If you use an acronym like MBA instead of the full definition it might not be counted. If you use an apostrophe it can appear as a code error. The Boolean logic used is ultimately flawed and the system can be gamed by keyword stuffing and using white text keywords. A clear case of automation requiring some human attention or maybe some AI enhanced automation?

There’s also the huge disconnect between what an ATS text friendly resume looks like versus one that may be read by a human at some point. It is very difficult to create a document that works as well for an algorithm as it does for an HR manager. According to Jobscan: “Many ATS parsing algorithms are outdated and unintelligent, causing your resume information to get distorted or lost. This means vital keywords or details might not be imported. Imagine your most important qualification slipping through the cracks!”[13] So for example, a candidate with the credential as a ‘Software Developer’ applying for a ‘Software Engineer’ position in an ATS world probably hasn’t a chance.

Consider candidates’ journeys for a moment: It can take perfectly qualified individuals many hours to simply submit their application. They must create a profile and register an account, usually on a desktop as many ATSs aren’t mobile friendly. They must change their resume each time they apply to a slightly different role with slightly different keywords, even within the same organization. They must recompose their cover letter each time, even though it rarely gets read by a human. They must create ATS and human-friendly versions of both documents. They must manually input numerous skills and micro credentials listing their number of years experience for each variable (everything from MS Office to Leadership Skills, from Python to Mandarin). They must input all their education and schooling, their professional certifications and accreditations. Sometimes they even need to manually input all their job history too and their responsibilities and achievements. They must go through a list of questions, some simply with yes/no answers but others can be in more of an essay format, such as the kind of answers you’d give in an actual job interview: Describe your experience analyzing user needs and software requirements to determine feasibility of design within time and cost constraints? Or tell us about 3 projects you worked on where you were supervising the work of programmers, technologists and technicians and other engineering and scientific personnel?  And all of this is done through a very outdated and clunky interface which regularly times out, has a time limit reminiscent of an online payment system, has an undisclosed word limit, doesn’t save or update and has no page back function. And all the while, it’s simply a relevant keyword count going on in the background. This type of automation lacking any sophisticated AI completely dehumanizes the applicant.

And what is the general outcome of the above for the candidate? Usually, an automated email thanking you for your application and stating that only qualified candidates will be contacted! The return on time invested for the candidate isn’t appealing, to say the least, and their perception of the company they applied to work for has most probably diminished enormously! And remember, this is all before you even get to interact with a human being and go through the 2-5 rounds of interviews. In fact, your resume (which by the way is a completely subjective document, ripe for exaggeration) was quietly filtered by a primordial form of AI and dumped in an electronic trash can! There’s also the strong possibility the job has already been filled internally or someone with a fast track to the hiring manager has already networked their way to the head of the pack. Sometimes the role doesn’t even exist and hiring managers to use the systems to get a read on what kind of candidates are on the market so they can try and request funds for a new role moving forward.

So, while many companies struggle to identify candidates with the requisite skills, is it possible that some great candidates are being unfairly overlooked and lost? While there are many candidates applying for roles they aren’t qualified for at all, which necessitates a measure of automation such as pre-application filtering questions, ultimately the whole process is built to fail both sides as currently conceived.

Before we start revolutionizing the education system to ensure today’s youth are ready for the jobs of tomorrow, let’s make sure we have a fair and transparent hiring process that gives candidates with the requisite experience and credentials a fair shot of being considered. Automation may be a challenge to the existence of many jobs from the past but let’s not make it a barrier to entry for the jobs of the future as well. Intelligently applied AI can be an evolution of automation and ironically help put the humanity back into the hiring of human ‘resources’ again.


[2] List based on statistics listed here:

About Emlyn: Originally from Ireland, Emlyn has been involved in educational programming for nearly two decades. Before moving to Canada, he had a career in research and lecturing at the National University of Ireland, Galway (NUIG), where he taught and published in the areas of economics, international political economy and globalization. Prior to joining ICTC, he managed the Adobe Design Achievement Awards (ADAA) and the Adobe ico-D Agency Mentorship program for Adobe Education through the International Council of Design (ico-D), based in Montreal. He also led ico-D’s policy and research workgroups in the areas of design research and education, understanding the value of design, professional design certification and accreditation as well as national design policy.

Join the discussion 2 Comments

  • Melissa Whitman says:

    As a Career Development Practitioner in Nova Scotia I also have a number of concerns about the use of ATS’s. I wonder if there are checks in place to ensure no bias (unconscious or unintended) or prejudice exists in the code? When humans screen there is usually a team which at least ensures there is some diversity in the screening, but with computers there is a single guard at the gate that has the perfect poker face. As a member of a couple of minority groups I wonder what purpose those self disclosing statements have. How difficult would it be to tweak the code somewhere to ensure those identifying with a certain group don’t get shortlisted? Who would know? Also, when it comes to questions relating to your “personality” or those checking to see if you are a “fit for the company culture” are these questions/answered screened for unintended bias? Another issue I have with these programs is how they look for “sameness”. There is a pass/fail. In an era that talks a lot about the value of diversity and inclusion, I wonder how diverse your workforce can really be when all candidates have to have the right check boxes and keywords. Is it all just lip service? It may make it easier to get along and move forward, it’s always been easier for managers to deal with people who think and act like themselves; But I wonder if a company doesn’t loose some ability to work with unique, interesting, and different types of talent. I’d like to see every company adopting the use of and ATP insist on all their existing employees to apply to their own company to see how many are successful. If you have a significant number that don’t make the cut, obviously that means you are screening out massive numbers of qualified applicants.

    Another thing I’ve noticed about the “personality” and “culture fit” type assessments is that the way people answer is often impacted by the job role they are currently working at. Someone currently working with people will sound very people oriented. Someone currently working with tech, may sound very people averse, but that doesn’t mean that they are always that way, just that it’s the perspective they’ve currently adopted to do well at their current job. I have worked with a manager of a company using an ATS before. I used their system to apply and barely scraped by with a yellow for a store clerk position. I’ve been a store clerk. I was quite good at it. I know I can run circles around many of the new hires I see (excellent math and social skills), but I haven’t been working in that field for some time. But sometimes people want a change. Sometime circumstances outside our control can force a need for a change. Some candidates are perfectly capable of change and transition, and those people often make very good employees, but their resume would never reach an employer unless they are comfortable submitting fake information which might be risky as often there are legal statements at submission stating that all the information is true to the best of your knowledge. Sometimes you are forced to fudge the truth when you education journey or work history is not a fit for the drop down lists. You lie to circumvent the system and get hung up at a later stage for being a liar. I’m really disappointed with the dehumanizing of “human” resources.

  • Good points.
    AI really helpful for minimize the hiring cost and time also.
    AI for recruiting is the application of artificial intelligence, such as the learning or problem-solving that a computer can do, to the recruitment function. This new technology is designed to streamline or automate some part of the recruiting workflow, especially repetitive, high-volume tasks.

Leave a Reply