The Employment Law Counselor hosted by Jeff Stewart

In the newest episode of The Employment Law Counselor, in collaboration with the Professional Liability Underwriting Society, Laura Corvo, Counsel, joins host Jeff Stewart for a discussion of AI in the workplace and the potential liabilities employers face. Jeff and Laura cover AI in the hiring process, recent laws trying to address it and how employers are utilizing the tool. They also talk about the importance of employers having a policy to provide guidance for AI usage in the workplace. 

Special thanks goes to White and Williams, LLP for partnership in the production of this episode. 

For the best experience, it is recommended you update the “Auto-lock” setting on your mobile device to “Never” so that the podcast doesn’t pause while you are listening.

Transcript

PLUS Staff: [00:00:00] Welcome to this PLUS Podcast, the Employment Law Counselor hosted by Jeff Stewart. Before we get started, we'd like to remind everyone that the information and opinions expressed by our speakers today are their own, and do not necessarily represent the views of their employers, or of PLUS. The contents of these materials may not be relied upon as legal advice.

Jeff Stewart: Hello, everyone, and welcome to the Employment Law Counselor podcast. I'm your host, Jeff Stewart, and today we'll be talking about artificial intelligence in the employment realm. This podcast is a collaboration between White and Williams, LLP, and the Professional Liability Underwriting Society, commonly referred to as PLUS.

While our podcast is not legal advice, it is a practical discussion between two attorneys that deal with the maze and minefield of labor and employment laws on a daily basis. If you like what you hear, please give us a five star review and subscribe so you never miss an episode. Today, I'm joined by one of my colleagues at White and Williams, Laura Corvo, who practices out of our firm's Newark, New Jersey office.

How are you doing today, Laura? 

Laura Corvo: [00:01:00] I'm doing great. How are you, Jeff? 

Jeff Stewart: I'm doing great. I'm looking forward to this discussion on artificial intelligence, as it's certainly been a buzzword for a number of years now. 

Laura Corvo: It sure has. And, you know, it's funny, I'm not a technical person, Jeff, at all. And I want to start by saying that.

So when I think of artificial intelligence, I do two things. One is I go back to my childhood and I think of all the images of artificial intelligence that filled that, people like Rosie from the Jetsons, who was the famous robot housekeeper, who, you know, got the Jetsons out of many jams, and in my opinion, gave Alice from the Brady Bunch a run for her money, and I think of C3PO and R2D2, but I also tried to fast forward to the future and think about what AI is.

And I asked my son to help me with this new tool called ChatGPT, which some of you may have heard of. And I looked at that and I said, well, tell me what artificial intelligence is. And the answer I got was something like the simulation of the human intelligence [00:02:00] processes by machine. And includes things like learning, reasoning, problem solving, understanding natural language, perceiving and interacting with the environment.

Sounds a lot to me like Rosie, but I think it's a really long winded way of saying that, you know, artificial intelligence is this new concept we have where machines can start to take over some things that was once reserved solely for human intellect. 

Jeff Stewart: Absolutely, and I think we've gone from what was science fiction, when we were growing up to reality now and, and we have to find a way to navigate that new reality.

And, you know, I've had some discussions with people saying, "Oh, you know, why is everybody talking about AI and, and things like that? And why is it so important?" And I, I'd say, "well, think back 20, 25 years now when email was coming into the office. You wouldn't have email without having a policy. You wouldn't even dream of that today, but AI is kind of the new thing that people aren't sure how it's [00:03:00] going to be used in the office, but it certainly has the potential to become as common as email is today."

Laura Corvo: Absolutely. And I think it's a tool that business owners in particular are looking at and saying, this is something we want to incorporate in our businesses. This is something that's going to help us with efficiency. It's going to help us improve sales. It's going to help streamline our processes and they're ready to go for it.

And the tool is really cool and it's really impressive what a lot of these AI tools can do. But like you said, if you introduce it without putting proper parameters up, you're setting yourself up for, for problems. The analogy I like to use is a preschool teacher introducing a basket of scissors to a classroom.

On the one hand, there's a potential for great art to be created. But on the other hand, there's a potential for total disaster. If you don't show them what to do, tell them what to do and put the proper [00:04:00] parameters around it. And really that's what employers have to start thinking about when it comes to AI.

Because I think it will be part of our future, it's just how we implement it that's going to be the real difference. 

Jeff Stewart: I've never heard the basket of scissors. And now that's a, that's a nice one. I'll have to remember that one. So let's, let's kind of jump in and let's talk about, first off, you did a wonderful article recently on workplace AI risk that I found very insightful.

And that's why I thought we should. Dive into some of these things here on the podcast, but what are employers using AI for in the workplace the most currently that you see? 

Laura Corvo: Well, I think the number one thing we've seen is in hiring decisions, you know, We kind of had the great resignations where a lot of employers were frenzies trying to bring people into their workplace and trying to screen applicants and hiring, they turn to AI tools to sort out those [00:05:00] applications and figure out who to interview, when to interview, how to make those decisions instead of having a human go through the thousands of applications that may come in, you have an AI tool do it.

So I think that's the number one area we're seeing, but, but there are, there are some risks when you implement AI in hiring or other employment decision making processes. 

Jeff Stewart: And what are some of those risks or what are the biggest risks that you're seeing? 

Laura Corvo: So the biggest risk is concern over discrimination, right?

So we as humans are prone to bias and discrimination. And obviously there are statutes that protect against that. There's title VII, the federal statute that protects against employment discrimination based upon certain protected classes like race, gender, religion, et cetera. What has happened with the AI tools is.

There's been studies that shown that they are also prone to that same discrimination. [00:06:00] And if you use an AI tool and that AI tool eliminates a certain class of applicants or individuals from the hiring process or the promotion process, you're going to be faced with a discrimination claim. And I think there's two really important things that you have to think about.

Number one is the EEOC, that's the agency, the federal agency that governs discrimination claims has issued recent guidance that said, when you use an AI tool for hiring or other employment decision making processes, unless you audit that tool to make sure that it isn't prone for discrimination, you could be subject to a title VII claim.

And then we've also seen some state laws popping up in a couple of states, and particularly in New York City, where they say, if you're going to use an AI tool, you have to conduct a bias audit of the AI tool. And you not only have to conduct a bias audit, but you got to place it on your [00:07:00] website so your applicants can see what it's all about.

So it's really risky at this point for employers to engage in the use of AI tools when it comes to hiring, unless they're very carefully screening those AI tools and conducting bias audits to make sure that the AI tool is not prone to either eliminating or favoring a particular class of individuals.

Jeff Stewart: Absolutely, and I think I want to highlight a few of the things that you just said here for our listeners, because there was a lot of good information there, but first and foremost, employers can't use the defense in a claim like this of, "well, we asked the computer or the tool that we use to give us the five best candidates.

And we just took those five and the other 30 candidates that were disqualified. We never even saw them. So don't sue us. We didn't do anything. We just followed [00:08:00] our process." That's not a valid defense. Okay. I think first and foremost, not a valid defense. The second thing that is part of what you said here is the AI tool does not replace a human decision. Ultimately, there has to be a human decision involved in this, but I've had conversations with people about how they're utilizing AI. Many times they've told me AI deals with the transactional things, not the relationships. So the AI tool can see if a person meets the minimum qualifications. 

Do they have five years of experience? If that's what we're advertising for, do they have college degree from somewhere or in a related field and they can sort out those things, but any curve balls, the AI is not necessarily prepared to handle the curve balls. Would you agree with that? 

Laura Corvo: I would definitely agree [00:09:00] with that.

The AI tools have made tremendous strides, but they're not a substitute yet for human judgment. And you have to have a human oversight and not just human oversight, but a human oversight with somebody who's knowledgeable of the discrimination laws and the propensity for making sure protected classes are considered in the process.

You know, we call it human resources for a reason. There has to be that human oversight of this decision making process. And I just think when we're talking about hiring and any employment related decisions, you have to be so careful and so cautious that the AI tool is not engaging in bias or that you're not setting yourself up for a potential discrimination claim.

Jeff Stewart: Absolutely. And one of the classic examples of the bias in the AI tool is if you type in that you would like an image of a doctor you were going to [00:10:00] get first image, and I have watched this done, male middle aged white person in a lab coat If you say I want another picture of a doctor you may get a white middle aged female in a lab coat.

You do it again. You may get a male Asian middle aged lab coat. Okay. So that age is there in the bias. The race was there initially. So there's a number of things. Gender is there. What is your AI tool doing? How, what bias does it have? And how are you combating that is a challenge for employers. 

Laura Corvo: Precisely. And I think also the law in this regard is developing.

So, we know for example that the EEOC has put this on their radar, right? And they said, hey, you do this wrong, we're going to come after you for Title VII violations. President Biden [00:11:00] issued an executive order regarding AI broad executive order covering a number of different topics, but discrimination concerns and making sure that AI tools were used in a manner that was fair for workers was high on the list.

And then we're starting to see these state laws pop up. So at this point in time, you're concerned about not only eliminating the bias, making sure we're in compliance with what the EEOC has said, but you're also going to realize this is new and there's potential for other areas of regulation around potential discrimination claims.

So you're stepping into very hot water. 

Jeff Stewart: Absolutely. Do you think most employers fully understand the risks? And what regulations are out there?

Laura Corvo: I don't because it's so new. Obviously, we haven't seen big lawsuits hit yet, but that would scare employers. We don't have a lot of guidance yet. And I think employers, particularly, as I said earlier, are attracted by the [00:12:00] efficiency of these tools, and they're ready to go full steam ahead.

But you got to pull yourself back a little bit and say, okay, wait a minute. As much as this is going to improve my business, there's also potential pitfalls. So I've got to prepare for those pitfalls. 

Jeff Stewart: And in reading the tea leaves, it looks like to me, the EEOC is looking for a test case in this area. And they are looking for a way to make a statement to employers as to, this is what you can do.

This is what you can't do using a real life example. And usually they do that by trying to find a test case. 

Laura Corvo: Absolutely. And you don't want to be that test case if you're an employer. 

Jeff Stewart: Absolutely not. So in addition to the hiring area, what other concerns or what other areas do you see it being utilized and what other concerns do those raise?

Laura Corvo: So I think another area of concern for employers is privacy concerns. There are many [00:13:00] state laws out there that protect consumer privacy. That would include things like a customer's birth date, social security numbers, credit card information. If that gets out there and you have a data breach because you've used an AI tool, there's big risk and big damages that could come back to the employer for having done that.

So you have to consider. What information you're giving AI, are you giving AI your customer's credit card information? Are you giving protected health information? If you're putting it out there, where is it going? What are the parameters? A lot of people don't understand completely how it works. So you have to make sure the information you're inputting is not going to lead to a privacy breach.

And it's not just your consumers or your employee privacy that you have to be concerned about. It's also your own privacy. We as employers take great measures to protect our confidential and proprietary information. We have our employees sign confidentiality [00:14:00] agreements, non competes, non solicitation agreements.

If you then take that information that you've alleged is private. And you put it into an AI tool, which can then be publicly disseminated, there's a really good chance that you've blown your argument that that's now confidential information. So when you go to enforce those confidentiality agreements, you're going to push back to say it's really not confidential because you put it up on ChatGPT or whatever the AI tool is for everyone to see.

Jeff Stewart: That's a great point. Again, this is one of those areas that I don't think employers are thinking about utilizing AI will provide a benefit here, but what is the impact on some other things that we have done, like signing confidentiality agreements with employees, what impact is that going to have? And obviously that's where we, as the lawyers need to be looking and thinking proactively and having those conversations with our clients as to what impact can these things have. 

Laura Corvo: And I think it goes beyond [00:15:00] even the concern over our legal obligations, like in privacy or discrimination, but employers also have to think about their brand, right? Their brand integrity.

We do a lot as employers to make sure that what we're putting out there is quality product. AI has made tremendous strides and it's does amazing things, but it's not perfect. It's prone to things like hallucinations and don't ask me how it happens, but the AI tool produces something that doesn't exist in reality.

It's either a false positive, a false negative, or just some result that just doesn't exist in reality. We've heard a lot about deep fakes in the political world where there's somebody's image put in a situation that doesn't exist in reality in a fake situation. And there's just your general spelling, grammar errors, just quality that's out there in our profession.

I think we've seen some cases, Jeff, where lawyers have used AI tools to produce legal briefs, and [00:16:00] the result has been that the AI tool has produced a case that doesn't exist in reality. And we've seen lawyers brought up for sanctions or just sheer embarrassment. If you go before the judge and you cite a case for a legal proposition, then the judge's law court goes to find that case and it doesn't exist.

And suddenly there's a lot of egg on your face. So you want to think about protecting that brand integrity and making sure that the product that AI puts out is a quality product. 

Jeff Stewart: Absolutely. And I guess from there, what I want to get into is from a policy standpoint, as employment attorneys, we are very much believe in having clear policies so that employees know what the rules are, what the parameters are for, for certain things and utilizing those policies to frankly mitigate the risk for the employer. So Laura, I guess I'll, I'll [00:17:00] tee this up for you. What do you think needs to be in a policy on AI for an employer? 

Laura Corvo: So I think what employers have to figure out is the how, when, and why they're using AI, right?

How are you going to use this tool? Are you going to use it, for example, for hiring? If you're going to use it for hiring, are you doing the bias audits that are required by the EEOC's regulations and the state laws? Are you going to allow employees to use AI for certain functions and not others? And it's not a one size fits all policy.

Certain employers may decide not to use it at all. Certain employers may decide to use it for very specific functions, but you have to kind of figure out the parameters as to how and when the employees are going to use AI tools. And then, you have to prohibit the employees from using it in certain circumstances.

For example, you may say, we're not going to [00:18:00] use AI when it comes to personnel or employment related decisions. We want to stay away from that. So HR folks, you're not touching AI, but it's okay for employees who are doing some sort of processing or cataloging or something, maybe that will allow it to streamline.

And then you have to set up certain rules. You definitely want to prohibit your employees from using privacy information in an AI tool. You want to put in some sort of quality control checks so that you have human oversight of whatever the AI tool does. Are you checking the spelling? Are you checking for accuracy?

Because those hallucinations and deepfakes are out there. Is there a human being checking to see that that AI tool has done what you expect it to do? 

Jeff Stewart: And I've had discussions with some AI experts, which have told me AI is wonderful to create a starting point. For example, "give me a presentation on the FMLA" and AI can pull all [00:19:00] kinds of information about the FMLA and I can say, "I want to target it to new HR managers who don't have much of a background. Do that." 

All right. And then I can tail take that as the starting point and then tailor it and, and go through, but to just rely on it, would be stupid. That that would be not utilizing it the way it should be utilized, at least at this point in time. Would you agree with that? 

Laura Corvo: I would agree. It's not only stupid, it would be dangerous.

Jeff Stewart: Reckless. I think was the word that was used. 

Laura Corvo: Even a better word because you're setting yourself up for potential liability and potential embarrassment. Again, if you put something out there, forget the legal implications. You give your customer a product that's not quality because you relied solely on an AI tool, that's going to come back and hurt your bottom line. And that's, that's, I think probably the utmost concern. 

Jeff Stewart: Absolutely. I think very much like email policies [00:20:00] that if they're not enforced, you are asking for trouble. It's one thing to have a policy. But you need to actually monitor it, actually follow it and make sure that people aren't just completely ignoring it.

Because I think you and I know that we have a lot of clients that have an email policy that never look at email. And basically they might as well not have a policy anymore. But, this is an area where that policy needs to be enforced. Would you agree with that? 

Laura Corvo: Absolutely. I actually will take it a step further and say that you need the policy on paper.

You need to give it to your employees, but you also need to have some sort of training around that policy, right? You can't just throw a piece of paper at your employees and expect them to follow it. You want to show them, tell them what you expect from the policy, and then you want to enforce it and enforce it consistently.

A lot of employers will [00:21:00] ignore their policy for certain individuals or certain groups of people. But then when they have a problem employee, try to bring the policy out and say, "aha, they violated the policy. Now I can terminate them." But if you didn't come down on every other employee who violated the policy, that employee is going to argue, "well, you did it to me because I'm in a protected class and now you're facing a discrimination claim."

So it's really a four step process. Number one, you get the policy out there. Number two, you train them. Number three, you enforce it and then you enforce it consistently.

Jeff Stewart: Yeah. And I think the fifth piece, if I can add to your thing is you update it regularly. This is an area of not just the law, but an area of the world that is changing very rapidly. So the normal, we make sure our handbook is reviewed every two years, I think if you wait two years to review an AI policy, you're asking for trouble.

Laura Corvo: Absolutely. This thing is [00:22:00] exploding faster, um, than, than the regulators are able to keep up and you are going to see tons of regulation in this area in the next several months, several years, and it's going to change and it's going to change rapidly. And as the tools change, the regulations are going to change.

So you can't rely on a policy. The policy that you had yesterday may not be good tomorrow. So you, you've got to, you've got to stay abreast of those laws and, and be ready to update to, um, comply with them. 

Jeff Stewart: Absolutely. Well, Laura, I thank you for, for taking the time to have this discussion today. But as you know, as you've been a guest on my podcast here before, I like to give our listeners a few key takeaways.

So can you give a, a key takeaway for our listeners here on this AI topic? 

Laura Corvo: Sure. I think the key takeaway is that AI is here to stay. Employers have to get ready for it, but you have to get ready for it by understanding what the risks are. And, you know, [00:23:00] setting up some parameters for your employees so that, um, you're using it in a way that's going to benefit your business and bring the benefits and growth to your business that you're expecting without setting you back and hurting your business.

Jeff Stewart: Absolutely. And I, I guess I would add to that, that it is important to continually monitor these areas, not only in your policy, but what are your employees utilizing it for and does that align with your company's mission and vision? 

Laura Corvo: Absolutely. 

Jeff Stewart: All right. So with that, I want to thank you, Laura, for joining me and I want to thank all of you for joining us here on the Employment Law Counselor podcast, where we try to make sense of the world of labor and employment law.

On behalf of myself and Laura Corvo, we thank you for listening. If you enjoyed this episode, please leave us a five star review, tell your friends and subscribe to the podcast. For more information on [00:24:00] this and many other topics, please visit the White and Williams website at www.whiteandwilliams.com where you can visit our blog and learn more about the firm.

Until next time, stay safe and stay compliant. 

PLUS Staff: Thank you for listening to this episode of the Employment Law Counselor. If you haven't checked out the previous episodes, make sure to give those a listen and check back in the next few weeks for the newest episode. If you have an idea for a future PLUS podcast, you can visit the PLUS website and complete the Content Idea form.