.jpg){kind=avatar}
Implementing AI in talent acquisition has long proven to increase efficiency and speed. But, when designed correctly, it also has the potential to create a hiring process that is fair for all candidates.
Of course, as AI capabilities develop and expand, the risk of a biased hiring process at scale also increases. Organizations must ensure that as they implement AI into their hiring processes, they implement it in a way that guarantees zero discrimination.
Keith Sonderling, Commissioner of the United States Equal Employment Opportunity Commission (EEOC), recently discussed at a Paradox conference how AI can be used to reduce bias in the hiring process — and what smart companies need to ask themselves to ensure equitable practices.
Can you give a brief introduction of what the EEOC does?
Keith Sonderling: We are the federal agency responsible for enforcing all anti-discrimination laws and making sure employees have equal employment opportunities in the workplace. In other words, we are the regulator of HR. We are the government agency trying to help ensure that all employees and applicants get an equal opportunity to succeed in the workplace, regardless of their protected characteristics such as age, sex, religion, etc.
We're dealing with some of the most fundamental civil rights we have in the United States: the ability to enter and thrive in the workforce, and to provide for your family without being discriminated against. In addition to lawful hiring practices, we also deal with some of the biggest issues in HR, such as the #MeToo movement, DEI, pay equity, and workplace accommodations.
AI in the hiring process is becoming more common. How does the EEOC feel about that?
I believe employers are free to use technology in any way they see fit; it just must comply with long-standing laws. If a company wants to use AI in any part of HR, they certainly can, but must know that the EEOC will ensure the results of that employment decision occur without bias-just like an employment decision made without technology by a human.
The EEOC will ask: Did that tool actually make a lawful hiring decision unrelated to any protective characteristics such as age, race, religion, or national origin? Was bias somehow improperly injected? I personally believe that AI can give everyone a better candidate and workplace experience — but to the EEOC, that’s not as relevant as not being discriminated against. The EEOC wants to make sure that candidates are able to apply for the job and ensure that no unlawful factor — age, sex, national origin, et cetera — plays any role in that decision.
There are only a finite number of employment decisions out there, such as hiring, firing, promotion, wages, training, and benefits. That has long been HR's world, and all that AI tools do is increase the size, scalability and efficiency of making those established employment actions.
I haven't seen an AI technology in HR that introduces an employment decision that is yet to be regulated. Instead, it's taking all your existing processes and it's replacing or augmenting them. The lawful outcomes of all of these decisions, all of these practices have already been determined. With or without AI, companies just need to do what they know best and make employment decisions that comply with the same longstanding HR laws.
Do you think that AI can help make the hiring process less discriminatory?
AI tools, if they're carefully designed and properly used, can absolutely help reduce bias in the hiring process, by eliminating human bias that improperly factors into “human” employment decision making. If AI is unbiased and is designed to ignore such factors, it allows individuals who have been unlawfully prohibited from the application process to actually have a chance. Those candidates are now being judged on their ability to perform the job and nothing else. It’s finally and solely about their skills and abilities — not the color of their skin, their sex, if they’re pregnant or disabled.
Historically, those individuals have had to deal with human bias in the application process. A recruiter could think to themselves, “Okay, I know you’re disabled. I can see you’re religiously observant. I’m going to have to give you an accommodation. Even though you have the same skills and qualifications, I'm not going to move you forward because you're going to cost me more money by way of an accommodation request.” This is clearly discriminatory and no recruiter would ever admit to saying this — but it is the reality many applicants have historically faced.
With AI, when properly designed and carefully used, these candidates can move forward from this initial step, without those unlawful factors intentionally occurring. The algorithm doesn't see those characteristics like a human does. That’s a wonderful benefit that we didn’t have before technology.
Would you say that companies should be investing in AI or staying away?
It's no longer a question of “should you use AI in HR?” We're past that point, especially for larger organizations that need to hire at scale. If you want to hire for 50,000 seasonal roles, you don't have 300,000 talent acquisition people to schedule and do all the interviews. You’re going to use AI.
So, I am a fan of the technology. If it's properly designed and carefully used, then it can actually help employees take a more skills-based approach, while eliminating human bias.
AI tools, if they're carefully designed and properly used, can absolutely help reduce bias in the hiring process, by eliminating human bias that improperly factors into “human” employment decision making.
But, at the same time, just flip it. If it's not carefully designed or if it's improperly used, it can discriminate more than any individual possibly could — just because of the scale and volume of it all. For these technologies to work properly and in accordance with the law, organizations must take the time in advance to ensure that AI does exactly what they purchased it to do on their own workforce. Strong governance, including ethical use statements, training, policies, testing, and other procedures, will only help ensure the tools work and comply with the laws. Honestly, that advice applies whether you’re using AI or not.
So here's what I need everyone to focus on: What are you going to use? Why are you using it? How are you using it? And how are you going to ensure that the employees and applicants’ civil rights are not being violated within that process? How are you going to ensure fairness?
