Is Artificial Intelligence Bad for Job Seekers? Don’t Lose Hope Just Yet…

Typing on laptop

Written by Lawrence Dearth, originally published here on LinkedIn.

A lot has been made recently of the gap between the number of job seekers and the number of unfilled jobs. How can so many job openings be available when unemployment is still so high? One area where job seekers and analysts often blame is the current technology used in hiring systems. The WSJVOX, and dozens of outlets often surmise that artificial intelligence-based resume grading systems might be to blame.

Let’s look at how these systems work. In the 2020s, it is estimated that full-time jobs, on average at major corporations, receive over 250 applications per role, interview 5-10 of those candidates and hire only one person. Using natural language processing and machine learning, artificial intelligence-based tools take those resumes and compare them to the job descriptions. Based on the company’s logic that is hiring for those roles, it will automatically “grade” those resumes and remove them from the queue for the recruiting teams, and at scale, that happens a lot. In a joint study, Harvard Business School and global services firm Accenture estimate that over 10 million resumes a year never make it through this process. Those candidates never had a chance. That presents a moral question: is that fair?

Is that fair?
I want to ask another question: did they ever have a chance in the first place? Technically speaking, if you are one out of 250, you did have a chance; a .4% chance to be precise. That’s not a lot. When you think of the odds alone, did that AI system just send you an expedited rejection letter in first class fashion? Many experts say that it would have been hard to get a call and the application simply ignored you before the recruiter could.

Did they ever have a chance in the first place? .4%
Let’s look at how corporate hiring works a little closer, now from the recruiter’s perspective, the recruiter that AI is trying to impersonate and make more efficient.

The recruiter, at the company to which you’re applying, has 10 job openings. They have 250 applications per opening. Because of this, they are incentivized to focus on those 10 and only those 10. How could they do anything else? They have time to call 10 candidates a day, max. They spend their mornings sifting through resumes, identifying tomorrow’s subsequent 10 calls, and then the afternoon on the phone. They’re doing the same thing the AI system did. Because they have a req-first mentality, they just ask are these candidates good for these reqs? They just have to get those reqs filled.
Seems hopeless, right? Well, not necessarily. I’m here to tell you not to lose Hope.

The key here is combining the power of AI and machine learning with the business model of having a candidate-first mentality. Let me give you an example of what I’m talking about: Tommy is a technology recruiter at Insight Global. For the last six weeks Tommy has done nothing but respond to some of our 125,000 monthly job applicants (his team received about 800). Tommy doesn’t have AI to tell him if these candidates are good for the jobs to which they applied, but that’s not why he’s calling them.

He’s calling them to see if they are good for any of our 20,000 openings across North America. Weekly, Tommy talks to 25 of them and of those 25, is able to submit eight of those 25 to a total of 20 openings. He uses automated search tools to match good candidates with good jobs. At the end of the week, he gets four of them jobs for a placement rate of 25% (Insight Global’s ratio of candidates submitted to hires made).

Tommy is able to do this because he’s not comparing the candidate to the job for which they applied, he’s looking at them for who they are and if we can place them across all openings. As a matter of fact, of all candidates submitted and hires made, more than 60% of those hires were made for jobs other than the one to which the candidate applied.

What if AI could replicate that and tee up those connections for hiring teams? Have Hope job-seekers, it can.

That is where the real successful companies using algorithms in hiring have thrived. Job boards and soon to be staffing agencies, are using their own artificial intelligence algorithms not to exclude job applicants, but instantly match them with the additional jobs they could be a strong fit for. If this technology, putting Tommy’s brain into a recommendation system, is used to open doors it will create more assists than blocks for the job seeker. Technologically speaking it’s actually pretty simple.

Resumes can go through a technology process called “parsing,”which essentially looks at dozens of variables in the resume and gives it a value in a wide range of areas. Think of this as the “DNA” or “fingerprint” of a resume. From there, instead of comparing your “fingerprint” to one profile, it actually compares it to 20,000 jobs. Your recruiter is then able to start the conversation with an array of opportunities, and since he’s incentivized to make placements with a candidate first mentality, he immediately becomes your advocate. Through conversation, your requests and preferences alter the profile, and your matches grow, not shrink. This is the power of machine learning-based recommendation systems.

If you have AI in place, and you use it to be the air-traffic controller for opportunity, this is how the conversations will go with the right staffing agency:

“Hi, is this Taylor?”

“Yes this is Taylor, who is this?”

“This is Tommy from Insight Global, I’m giving you a call about an application I saw come through this morning. Do you have time to talk?”

“Oh, great. Yah, now is awesome. So you must think I’m a good fit for the role I applied for? It’s so nice to hear from someone.”

“Well, kind of. I don’t think you’re a good fit for one role, I actually think you’re a fit for five.”

In these scenarios the candidate gets the win, the recruiter gets the score, and the technology gets the assist. Let’s use this wonderful technology to open more doors and close a few less. Let’s have a little hope.