From the Desk of the Grumpy Data Dude – Straight Talk on All Things Data and Talent Management

We do a lot of writing at Talent Dimensions – blogs, books, articles.  All very professional, very proper.

This series?  Not so much.  No, this series from the Grumpy Data Dude is a little different.  Still informative, still pertinent!  Just…not quite so serious.  Our resident grumpy data dude has a lot to say about a lot of topics.  So why grumpy? Because so many of the things we worry about in talent management, we should know better, and we should DO better.  These blogs are an exploration of that through the lens of a data guy, the father of all INTJs, who believes if you’re not making decisions based on data, you’re just guessing.  And boy, is there a lot of guessing going on.  So, enjoy these, and we hope you learn something while having a little fun!

I grew up watching Star Trek. You know, the original, when you shot first and asked questions later and every single alien spoke English? Yeah, that one. What does that have to do with AI and recruiting? Well…

In one of the episodes called – appropriately enough – “The Ultimate Computer”, a new super-intelligent computer that has been made to think based on all the information it’s been given (sound familiar?) has been installed on the Enterprise. And it’s being tested to see how well it performs. As you might imagine, Captain Kirk and friends aren’t thrilled with this. Well, in one scene they visit a new planet, and Kirk makes recommendations as to who should go check it out.

He chooses the usual selection of people (including a disposable “red shirt” who only exists to show how the monster works). And one of those is a planetary geologist named Rawlins (and yes, I did actually remember that…data isn’t the only thing I geek out over!) But the computer picks another geologist named Carstairs. Kirk asks why, given that Rawlins is his most experienced guy. The computer dispassionately informs him that Carstairs grew up on a planet just like the one they plan to explore, so he has more expertise in this case. Kirk is not amused, but grudgingly accepts that the logic is sound. He’s also miffed that he didn’t know it himself.

In other words – and now we’re getting to the point – the computer made a talent selection decision based on information that it had that a person did not. AI at work.

So, what does this have to do with recruiting today? Well, believe it or not all the pieces are there to have AI do your recruiting for you in the very near future. And I don’t just mean what an applicant tracking system will tell you, though that’s where it starts.

Recruiting typically has these steps:

  1. Job description created and posted.
  2. Applications received.
  3. Applications evaluated against the job posting.
  4. Top candidates selected for interviews.
  5. Interviews conducted.
  6. The hiring manager and HR make a decision based on their interactions with the applicants.
  7. Offer is made.

Now. You might think that only step 3 is really where AI might have a role to play. It does; applicant tracking systems essentially do steps 2 and 3 for you, weeding out applicants who don’t meet your needs. But what about the rest of the process?

Well, if AI is trained appropriately, it could easily make suggestions about which applicants deserve an interview. This could be based on many factors you tell it to consider, such as experience, physical location, education, specific knowledge, skills or abilities they have, etc. And it could do this much faster than a person could…essentially it can do it instantly, even with a large number of applicants. Tools such as Seekout or Workable already do this, and take it even further, helping you refine your searches and requirements more efficiently. From writing job descriptions to summarizing an applicant’s resume and selecting top choices, these tools are already using AI to make more data-driven recruiting decisions.

But what about the interview? Surely you say, no computer can interview a person as well as another person can. Well, there already exist two types of software that could allow AI to do this for you. First, there are several providers in the market who have platforms for pre-recorded interviews. It’s straightforward stuff; you send a link to an applicant with a set of questions for them to answer and a deadline to complete it, and there you go; all your interviews are done for you. But here’s where AI could play a role: AI could interpret the interview for you. There are already platforms such as Orai, Speechify, Ummo and LikeSo that evaluate public speaking skills. How long until – if it isn’t happening already – AI can not only evaluate their speaking ability, but also evaluate the quality of their answers against a pre-determined benchmark? And after doing so, it could tell you who should advance to the next part of the process?

Sounds improbably futuristic. But it’s well within the grasp of current tools to do this work, and like self-driving cars or delivery drones, it won’t be long before it’s common practice. Imagine the day when you open a recruiting tool on your computer or even smartphone, ask it to generate a job description for a position based on what it finds out in the industry, and have it search for you. It then provides useful summaries of top candidates and recommends pre-recorded interview questions you should ask. Next it will email those interviews to the candidates and provide you with a detailed “score” of how they did based on answer rubrics that it designed for you. What’s next? Make a selection and greet them at the door when they show up the first day.

So, let’s get to the “Scarier than Friday the 13th” part.  You don’t have to look far to find an article or podcast that talks about the inherent bias that exists in the datasets and proxies AI uses in recruiting.  The algorithms AI uses run the risk of perpetuating these biases if not properly monitored.   A Recruitable newsletter cited a RAND report showed that between 10-50% of qualified candidates could be unfairly screened out before a human recruiter reviews their application.  This has a significant negative impact on improving workplace diversity.  That same RAND report estimated that AI bias costs US businesses around $100 – 300 billion in lost productivity annually by overlooking qualified diverse candidates.

With any great advance in technology that involves human beings, it’s critical that we understand and mitigate its downsides.  The good news is that there are strategies you can employ to ensure biases don’t unfairly influence your recruitment processes.  AI brings great promise and is influencing all aspects of talent management.   And having your trusty AI “bot” do your recruiting for you can’t be far away.