Technology has been a great thing for recruiting, but there's a dark side. Namely, artificial intelligence. A.I. has made companies more efficient and effective, but it's also been a roadblock for job candidates and a hindrance for hidden workers to get ahead.
EEOC commissioner Keith Sonderling joins the boys to give a high-level education on the current state of how the government is dealing with the questions around tech and employment.
PODCAST TRANSCRIPTION sponsored by:
Hide your kids! Lock the doors! You're listening to HR’s most dangerous podcast. Chad Sowash and Joel Cheeseman are here to punch the recruiting industry, right where it hurts! Complete with breaking news, brash opinion and loads of snark, buckle up boys and girls, it's time for the Chad and Cheese podcast.
Oh yeah. We've got the EEOC in the house. Everybody what's up at your favorite podcast. This is cohost Joel Cheeseman as always joined by my co-host in chief Chad Sowash and dude, Chad, we're so excited by this guest. I'm going to hand it off to you to take it from here.
Nobody is excited when somebody says E E O C, but I'm going to tell you right now on this podcast, this episode, you're going to be excited. So welcome Keith Sonderling, Commissioner at the US Equal Employment Opportunity Commission. That's the EEOC. Yeah. You know, me three years prior at DOL at wage and hour division acting and deputy administrator there. So I should say the main reason why we wanted to get you on Keith is because you have recently written articles and done podcast on AI. We'll get there, but we're really excited because we have not, I repeat, we have not had the engagement that any of us, I think in our industry have wanted from government around technology.
Chad (1m 30s):
But before we get into that, that's just a little teaser. Give us a little background about you. Who is Keith? Are we talking about long walks on the beach?
Keith (1m 42s):
Well, thank you. And for those, when you say EEOC's in the house, most people would run away from that house or immediately turn off, but stay on please, because I'm very excited to be here. Joel and Chad, thank you so much for having me. I've listened to your podcast. You've been a tremendous resource for me as I dive into technology on AI for just the next 45 minutes let me tell you how great both of you are, but in all seriousness, we will get the technology. And I do appreciate what you both do and bringing to light all of these different various technologies that employers are using and employees are being subject to. But first a little bit about me.
Keith (2m 22s):
I'm a commissioner on the US EEOC. I was confirmed by the Senate last September, after going through the Senate confirmation process, which took around 14 months, which is his whole own separate podcast.
Chad (2m 36s):
That would suck.
Joel (2m 37s):
It sounds like a colonoscopy.
Keith (2m 41s):
You know, in a way you have to look at it and sort of pinch yourself that you are interacting with this process that you're dealing with the United States Senate, that you've been nominated by the president. So as much time as it took, it was just a very, very cool experience. But before he got to the EEOC and I've been here just around a year, I was at the Department of Labor, the Wage and Hour Division would you said, which does the minimum wage over time, the family medical leave act and then some of the immigration and agricultural laws. But what drew me to the Department of Labor is that I was at Labor and employment, where in Florida, that's all I've ever done my entire career. I was at a, a Florida based law firm doing labor and employment, defending corporations and labor and employment suits on the litigation side, but also working with HR and working with companies on best practices, policies, and procedures relating to labor and employment.
Keith (3m 34s):
So when the opportunity came up in 2017 to join the Department of Labor, to me, it was not only a once in a lifetime opportunity, but I looked at it as like I'm a labor and employment lawyer. I can go to the mothership, the Department of Labor and get, essentially get a PhD in labor and employment right when you're doing it at that level. So I left Florida. So I'm no longer technically a Florida man, although I will always be a Florida man.
Joel (4m 1s):
Keith (4m 2s):
That's right. Thank you very much. So I'm proud to be a Florida man in DC. So when I joined the Department of Labor, I really saw a national perspective on how these issues affect employees and employers, and not just from the Florida perspective that I was dealing with before. And I was able to do a lot of really very cool things that the Department of Labor, whether it was opinion letters, whether it was changing the overtime regulations. So that was a really great experience. And then I was nominated to the EEOC and for labor and employment lawyers, the EEOC is really the premier agency for civil rights.
Chad (4m 40s):
The premier league.
Keith (4m 44s):
That is football is for another, for you two to talk about.
Joel (4m 48s):
Keith (4m 50s):
So that's great show, I haven't seen season two yet only season one, but you know, the EEOC really, when you think about labor and employment and modern day issues, that affect employees on a daily basis, it's not the Department of Labor, which deals more with the health, safety and wages concern. It's not the National Labor Relations Board, although that's very well known and that deals more with union issues. The EEOC gets to the core of it. It's the agency that deals with all civil rights in the workplace. So think about pay equity, disability discrimination, the entire #MeToo movement, pregnancy discrimination, age discrimination, all the big ticket stuff. So when you say personally, what it means to me from having had that experience about Department of Labor and now at the EEOC, it's just more than I could ever ask for.
Chad (5m 38s):
So, so once again, you've written articles and you've been on podcasts and you've actually been talking about artificial intelligence, AKA AI, and you've been very vocal about the impact of AI or the prospective impact of AI and hiring firing and the managing process. So what incidents got the EEOC more importantly, you, cause it feels like you're leading this, I'm interested in AI in the first place.
Keith (6m 5s):
Well, let me answer that in a few different parts. Generally commissioners at the sea have their own projects or areas of law that they specialize in. You know, one commissioner really led the charge on LGBT discrimination. One commissioner led the charge on age discrimination. So it's not uncommon for commissioners to pick a specific topic and really champion it. And for me, that is, as you said, artificial intelligence in the workplace, for a whole host of different reasons. First and foremost, that is out there is being used. It's not one of these discussions about, let's talk about how robots are going to replace humans and there's going to be no more workers and we're just live as a society of robot workers, right?
Keith (6m 52s):
That's what people want to think about when they think about AI and as you know, that's not what it is. It's technology that's out there right now. So the conversation needs to happen now. And, you know, coming from practicing law and dealing with corporations who need to hire workers who want to genuinely diversify their workforce and take out some of the bias in recruiting, these issues need to be addressed right away. And there's been a lot of interest before I got here, there was interest from Capitol hill, senators wrote letters to the EEOC, demanding that the EEOC take up this issue. There's been a lot of advocacy groups asking the EEOC to look at this topic.
Keith (7m 34s):
So technology, first of all, really interests me, but more importantly, there's so many benefits to using technology in the workplace that I want to see it flourish and not get subject to certain government regulations that are not going to make it work because we're already too late, it's happening and it needs to be addressed now. So because there's no regulations out there because there's no guidance on it. Because as you know, technology generally gets very far ahead of the government. It's a time where we can all really work together. Everyone from employee groups, to employers buying and using the software to developers, to create a standard that actually allows these products in this AI to help diversify the workforce, help get the best candidates, but also not put on burdensome regulations that take it down or subject it to a massive federal investigations or class action lawsuits.
Joel (8m 27s):
Keith can you talk about sort of how we got here, because the phrase, you know, the road to hell is paved with good intentions. And I feel like we got to this tech heavy recruitment process sort of not on purpose, but it just sort of happened that way. And what's your perspective on just how we got here?
Keith (8m 47s):
I think obviously the default answer and the easy answer is the pandemic really pushed this forward, but it was being used before then. And I think for larger corporations, think fortune 500 company companies that need to hire hundreds of thousands of workers. How do you deal with that process? How do you deal with the amount of applicants and how do you have enough employees internally in HR that are actually going to be able to interview these people? So I really think that's where a lot of this came out from that you have thousands and thousands of resumes and a human just doesn't have the capacity, or you just need a lot of them to sift through them. So I think that's sort of the basis of where a lot of this came out from, but then when Silicon Valley and tech people started getting involved and sort of adding in the AI to a lot of that, that's when a lot of these decisions that I've been writing about and I've been talking about are really coming to forefront.
Keith (9m 43s):
And that's more recent. I think that's in the last three or four years. And I know on your podcast, you talk about how much money is going into these AI technologies. And for me, obviously, you know, that's a good thing. If we're getting funding, if we're getting better products and investors are looking in this, that's good. Let's just do it the right way so it all, doesn't go down in flames because of misuse, either by bad actors or by bad design. So that's really how I look at this, is that it's brimming with potential, but at the same time, if somebody in my position doesn't come out and say, here are the best rules of the road. Here are the best practices. Then it could be really subject to some very serious lawsuits.
Chad (10m 24s):
Are we missing the forest for the trees though? I mean, because I think most people understand that AI, they misunderstand that the decision AI is making doesn't stem from AI itself. Rather it stems from human decisions. Humans are biased, always have been, always will be. Although when you, the human being program that bias into systems, processes, and in this case, AI, you start to reach scale, meaning that your bias could impact thousands instead of dozens. So from my standpoint, as we talk about regulation, I almost want to say, look, you know, when we're talking about bias, the bias is there, it's programmed in, we already have regulations to enforce bias.
Chad (11m 8s):
And these decisions are just really being taken to another scale with AI. But these are human decisions. So are we pointing our finger at AI instead of Jeff Bezos? Are we pointing them at the wrong person or the wrong system?
Keith (11m 23s):
That's really a great point and what I talk about. This is it's not about the algorithm, although it could be about the algorithm because I'm a lawyer, I have to disclaim it, right? So there's two ways to look at this. And there's been so much of a focus on the algorithm and the secret computer coding that's discriminating, that even if the three of us saw, we would have no idea what it even said, because it's probably all math formulas. Right?
Chad (11m 47s):
Keith (11m 48s):
That's not what it's about. And you just summarized it perfectly. It's either number one, the data going into the algorithm. And you know, and there's the two classic examples about this, which I've written and spoke about. The first is that very public Amazon when they used AI and they wanted to, they gave the computer, their ideal candidate, which is based on their historical applicants and their workforce. And then, because that was mainly males, it started downgrading everyone who wasn't a male. So it automatically started downgrading you if you went to a woman's college or if you played a women's sports.
Chad (12m 24s):
Keith (12m 25s):
Right. And that, that wasn't because the AI has misogynistic intent, which a lot of people would just want to say, you know, the AI is discriminatory. It was simply because of the data fed to him. And another example, and I know you both will get a kick out of this one. One firms said, go find me the ideal applicant. I want to diversify my workforce. Here's my top performers. And the algorithm spits out "your ideal applicant is named Jared who played high school lacrosse." Thank you. I mean what does that say? What does that do? But whose fault is that? That the inputs that gave you the bias, the bias inputs give you the biased outputs. Now, what I did disclaim is that there are situations where it could be, I don't want to say a biased algorithm, but a bias tool, right?
Keith (13m 11s):
So if some of these programs are poorly designed and then, you know, carelessly implemented by the employer and they allow you to screen out certain race, gender, age, or do sort of brackets like you would target other advertising, then that could be a discriminatory tool within itself. It doesn't matter if the data is, whatever the data looks like, the data could be, you can have a completely diversified work force, potential workforce. And then if you have a tool that allows you to sift through it on protected characteristics, that's the scaling discrimination like we've never seen before, right?
Chad (13m 44s):
Keith (13m 44s):
Because now you have a tool to do it. So for the most part, I completely agree with you. It's not about the algorithm who cares, what the algorithm, how it's designed, what we're going to look at, and what employers should be looking at is what is the results of what you fed the algorithm? Because that will tell you a lot about the data you put into it.
Joel (14m 3s):
Should we be pointing the finger at the employers or the software developers or a little bit of both?
Keith (14m 9s):
Well, first of all, we don't point fingers. We assess situation under applicable law.
Chad (14m 17s):
Joel (14m 17s):
Sorry, Chad, and I point fingers.
Keith (14m 20s):
That's fine. And of course, we look at every investigation individually and would not point fingers at anyone till we have reasonable cause. But in all seriousness, the who, the employer, and now I have to do sound like a lawyer for a minute. The employer is liable. There's no question under our loss that the employer using these tools to make the decisions will be liable for the outcomes of the AI. Whether or not, whether they bought the AI to really diversify their workforce and eliminate bias or help employees up-skill and re-skill, if it has that discriminatory output, the employer's on the hook.
Chad (14m 57s):
We've talked about black box versus white box AI on the podcast for years. And for all those listeners who haven't heard it before, black box easily, it's just something that's not transparent so you can't see what the algorithm's doing versus white box you can actually see it, tweak it and explain it, which is the most important thing I believe for any employer, that's out there to be able to see how it's working and why it's doing what it's doing. My question to you, Keith. I mean, because an employer can choose whatever they want, but at the end of the day, the outcomes are where the EEOC comes into play. Right? Do you see black box AI actually surviving the next decade?
Keith (15m 38s):
You know, you may not like the answer to this, but I'm not advocating not to have very transparent algorithms, but at the end of the day, it's going to be the results that we're going to look at. But as far as the actual being able to see what's in the black box, you know, at this point, it's really up to individual legislators or legislation or regulators to make that determination and what you're seeing absent a federal standard right now, which is something else I've been talking about is you're going to get a lot of different laws and regulations on this. So in Illinois, they've passed a law, banning facial recognition technology.
Keith (16m 20s):
The city of New York now has a proposal that if you're being subject to AI, that not only are you being disclosed as an employee before you take the test or being assessed by it, you're being told, you're being AI is being used in this process. And here are your rights and remedies under the New York civil code. So absence national auditing standard, or you're going to see a lot of these individual states, local governments and state governments start to say, you know, here's what we actually want to see. Here's the auditing requirements we want to see, but right now it's sort of all over the place, not just here. And I know you have a lot of listeners in Europe and you deal with a lot of those issues.
Keith (17m 0s):
They're really also getting ahead of it. And I know from the data privacy side, that's one issue over there, but for them, they've already said that using AI in employment, in their proposed legislation is going to be in their highest risk category, the same as critical infrastructure and emergency services. They call it the high risk. So that's sort of a long-winded way of saying everything is all over the place right now. And that's why I'm here.
Joel (17m 26s):
And this is a low, a high, this reaches into a huge number of companies. I mean, talking about the Harvard study, I think from this year, last year, it said 99% of Fortune 500 use technology in their pre-screening 75% of all U S employers. In light of that I think one of the things, and to piggyback on my last question, was it showed that nine out of ten executives rely on software and believe that it is unfairly rejecting candidates. So you say that, you know, the employer is on the hook for this, and I think it's important to just sort of underscore that because so many things that it's messed up, and are they blaming vendors in that process. And I think that a lot of employers need to look in the mirror and figure out that, we're the ones that are liable to these and the tools that we use and let's make sure that we're dotting our I's and crossing our T's.
Joel (18m 17s):
Would that be correct?
Keith (18m 19s):
Yeah. And that is a great point. Just like they're liable for, if an HR personnel or a manager makes that decision, right. It's just, we're doing it much more high-tech on a much larger scale, but HR tech can be much more transparent than trusting a human brain, right. So when you talk about this whole black box and AI black box, we don't know what's going on. Well, do you know what's going on in your hiring managers head across the various divisions across the country?
Chad (18m 46s):
Keith (18m 48s):
No clue. You know, and you don't know what they've seen. You know, and one of the examples I like to give on the benefits of AI in recruiting in that, in those first steps is that, you know, an HR manager or a hiring manager, talent acquisitions, they can see that somebody is of a certain national origin. They can see that a person is disabled or pregnant, right. And let's say, use a disabled and pregnant example. You can't ever unsee that. And although it is totally illegal, highly, you know, unlawful to not give somebody a job because of that, right? In the back of their mind, no matter what they're thinking, how much is this going to cost me?
Keith (19m 28s):
If this person, if I have to make a reasonable accommodation for somebody who's disabled, you know, how much is that going to cost me? If somebody is pregnant, how much is it going to cost me in health care or leave? And then if you have 10 other candidates, it's easy just to say, okay, we're going to move with these same highly qualified candidates as well, where they're eliminated at that very early stage. Right?
Chad (19m 50s):