When it Comes to Hiring Algorithms, Be Sure to Check the Math

Share This
February 22, 2017 | Ray Bixler | HCI

From fake news and social media feeds, to the financial markets, and now hiring, the impact of algorithms is being talked about a lot lately. A recent column published by the Harvard Business Review, “Hiring Algorithms are Not Neutral”, highlighted some concerns. The article states that 72% of resumes are weeded out before a human ever sees them. My most recent column in U.S. News & World Report “3 Ways to Make Sure a Machine Doesn’t Judge Your Job Search” focuses on the impact that hiring algorithms can have on job seekers, and this blog post highlights why employers must also be diligent in looking at the solutions they may be using.

In fact, organizations who are using some algorithms may not only risk missing out on great candidates who are being filtered out through these formulas – they may also be facing major compliance risks.

The EEOC held a hearing last fall on the use of “big data” when it comes to hiring. Some of the discussion explored the fact that algorithms – if poorly designed or used – can reinforce unfairness rather than remedy the problem. Any HR manager using such a system needs to be aware of its limitations and have a plan for dealing with them.

With that in mind – here are some things you should consider:

  1. Do your homework and vet any hiring technologies you’re using – Understand if they are using any algorithms and how they work. Ask for validation that the solution is not showing any bias toward any candidates based on race, gender or any other protected classes.
  2. Evaluate the data being used in decision-making – it should be providing you with information to make your decisions, not steering or making your hiring decisions for you.
  3. Be sure the hiring criteria that is being used to make decisions is truly related to the job – Again, if your solution is screening out candidates based on faulty criteria you risk losing out on great candidates, and it is a major factor for EEOC claims.
  4. Rely on multiple data points – For example, many aspects of the hiring process rely solely on data that the candidate provides through a resume or a cover letter, tests or assessments. Expand your insights through in-depth reference checking and behavioral interviews.
  5. Audit your program on a regular basis – Look at your hiring outcomes, the diversity of your new hire pool and your turnover rates on a regular basis to ensure that you’re hiring effectively.

Another good read is the Inside Higher Ed article Academic Moneyball. It is about a new algorithm developed by management professors from the Massachusetts Institute of Technology who say it can predict faculty research success better than the traditional tenure committees. This quote from Brad Fenwick, senior vice president of global strategic alliances at Elsevier, sums up the limitations. “More and better data would be helpful.” Yet data “should inform, not make the decision … We are not at the point of using [artificial intelligence] to replace human judgment.”