Hiring bias: It's not just for humans anymore

In today’s ever-evolving world of technology, I sometimes find myself struggling to keep up with all the new tools available to small- and medium-practice owners. I’m always on the lookout for ways that I can leverage emerging resources to improve my business and my life. And, as someone who makes their living helping other entrepreneurs build stronger, more compliant businesses through HR, I’m also constantly searching for ways to guide others to do the same. In a recent episode of my podcast, What the hell just happened?!, I sat down with an expert HR advisor from my team to talk about a new tool that seems to be everywhere right now: artificial intelligence (AI).

What is AI, at its core? In a paper published by Stanford University’s John McCarthy, he defines artificial intelligence as the science and engineering of creating machines that harbor intelligence that mimics our own.¹ However, AI does not cage itself to the biologically observable ways of learning that human beings are adapted to. AI boasts exceptional pattern recognition. When paired with machine learning, AI has the ability to examine an immense array of human inputs and identify patterns that emerge from them, even in cases where we don’t realize we are providing patterns to follow.

Related reading:

As an HR specialist, this naturally got me thinking about the use of AI in HR. We are seeing more and more systems being offered to practices, such as Indeed, which utilize some form of AI to automate processes that would otherwise pull them away from things like patient care and employee management. These tasks include tools that help automatically determine who to interview and who to pass on. On the surface, this sounds like a dream come true for business owners.

AI can amplify human bias

Having AI present you with the top five applications out of a pool of 50—who wouldn’t want that? However, this poses a number of important questions. Namely, does AI have bias? And, if so, what could that bias be? And where did it come from?

As human beings, we all have some form of implicit bias. As employers, we work hard to extinguish those biases for the benefit of our businesses. But machine learning software is programmed to simply look for any patterns that emerge from human inputs and determine how those inputs correlate with successful outcomes. When used in hiring, that software can’t necessarily distinguish between helpful inputs and those which might be the product of personal bias.

Imagine this situation: your AI system presents you with five applicants who have all of the qualifications required for a position—specific certifications/licenses, years of experience, and all the other surface-level requirements. Out of those five, you hire two people; coincidentally, both are women. From there, your AI has learned which candidates best fit your needs and, unbeknownst to you, has now adapted its algorithm to favor this new gender parameter when searching for future applicants. Alarms should be going off in your head right about now.

AI is not a perfect technology. In theory, it may be able to help employers find better candidates by searching for patterns among previous hires. But it can also interpret and encode your personal biases into its algorithm, leading to hiring decisions that not only fail to be optimal or even beneficial for your practice, but that may, in fact, be illegal.

And this problem can become a compounding one. Just like our brains, the more unintentional parameters you feed into an AI system, the more likely you are to overlook candidates who could be difference makers for your practice in favor of those that might fit some arbitrary, machine-made profile of your “ideal” candidate. Those biases need not be related to illegal criteria like race and gender, either. Your AI could learn to favor other arbitrary criteria such as what they studied in college, where they live, and whether or not they own a car. This, in turn, could lead the system to favor candidates who you tend to like for reasons you may not even be aware of (e.g., those you might hire because you simply vibe with them better) over those who will actually provide the most benefit to your business.

Interestingly enough, Amazon dealt with this exact problem in 2018.² Their AI for hiring was unknowingly passing over applications that women submitted as it learned from previous hiring decisions that men were hired more. When this problem was discovered, Amazon took the system down after several adjustments failed to fix the problem (taking out the gender parameter didn’t solve the issue; the AI still looked for other indicators that an applicant was a woman to ensure they were cycled out). If it can happen to corporations as massive as Amazon, how can we be sure that this won’t also be a problem for smaller businesses, including private dental practices?

Humans still have the edge

Another place where AI might fall short in HR is in its evaluation of employee performance. AI can easily find patterns in performance; things like how often the employee is late, when they meet their deadlines (or don’t), and other metrics. However, considering that most work is heavily dependent on the performance of an employee’s coworkers (we all win together!), the AI may give an employee who has missed several deadlines a bad performance review, even when those missed deadlines are the result of other problems with a given process or workflow.

This employee could be exceptional, but the AI can’t see what happens outside the numbers that are fed to it. It will not be able to recognize that the employee is great at collaborating with other departments and bringing new ideas to the table when a problem arises. For AI, individual performance is difficult to distinguish from group performance.

AI still has potential

It’s not all bad when it comes to AI in HR, though! More and more businesses are turning to AI to help their current employees thrive in the workplace. There are great ways to get a system to recognize where an employee may be struggling and suggest training that might help them improve, for example. AI can also help greatly with randomized decision-making for things that don’t directly affect your employees, such as processes related to the organization of patient information, scheduling, and outreach.

AI can be a beautiful thing in helping a business reach its highest potential. However, you should keep in mind that no piece of technology can replace your analytical skills as a business professional. AI does not have your capacity for emotional intelligence, which is invaluable when it comes to running a successful business.

Rather than looking to AI to take over certain tasks, think of it as a tool to help you complete tasks more efficiently. Use it as a stepping stone to help unlock new possibilities for yourself and your team, pairing their individual skills with it to accomplish things you didn’t even know were possible. Additionally, when in doubt, contact an HR professional for guidance if you are ever unsure of the decisions you are making. The more prepared you are for the inevitable integration of intelligent systems, the more ready you will be to use them to your advantage.

Editor's note: This article appeared in the October 2022 print edition of Dental Economics magazine. Dentists in North America are eligible for a complimentary print subscription. Sign up here.

References

McCarthy J. What is artificial intelligence? November 12, 2007. Accessed July 29, 2022. http://www-formal.stanford.edu/jmc/whatisai.pdf
Cappelli P, Tambe P. Artificial intelligence in human resources management: challenges and a path forward. SSRN Electron J. January 2018. doi:10.2139/ssrn.3263878