This is being provided in a rough-draft format. Communication Access Realtime Translation (CART) is provided in order to facilitate communication accessibility and may not be a totally verbatim record of the proceedings
Anthony May: All right. Good morning everyone. How is everyone? Now you can hear me now. Good morning, good morning! Hello everyone! My name is Anthony May partner at Brown, Goldstein & Levy. I am a white male using he/him pronouns, short brown hair, blue glasses a gray suit with a checkered shirt and blue tie. I'm here to talk about AI and employment discrimination I want to give a huge shout out to our opening panel for setting this up very nicely. So you're probably going to hear a little overlap today we've got about five minutes to speed through a lot of updates how AI is used in employment discrimination law.
So, I'm going to go head and get started, I'm going try to save some time for questions at the end. But what I've already learned what about I love about this conference I get the opportunity to meet people who work in such a variety of different areas and spaces and know so much more about certain topics than I do. So I have done a lot of research on this topic I've learned a lot more, I'm really excited to have this conversation today.
As a bit of road map for where we're going to go this morning, I'm going start by defining what AI is, and how it is used particularly in the work place. Then we'll talk a little bit about how and what we've got a little preview already from this morning's conversation is how AI can discriminate. We're going talk about the legal efforts have been used to curb discrimination and then, finally I'll give some best practices both for employees or perspective employees and for employers to try to ameliorate issues with AI coming up in the work place. Let's get started, first we'll define what AI is and how it is used in a work place, I have a co-presenter is going to be joining me virtually here in a few moments anyone familiar with gentleman named John Oliver he does a great job being funnier and explaining the concepts much better than I am. There will be a couple of clips we'll have this morning, I'm let everyone know the PowerPoint should be able it should be accessible if it's not, please let me know because we need to fix that. I'm fairly confident that it is. So I'll set up this first clip here, I'm going introduce my co-presenter who will talk about artificial intelligence what it is and how you see it pop up in the work place.
All right that's a very general overview that's Mr. Oliver explaining the basics of AI so I'll go into a little bit of this I just gave this presentation at a assistive technology conference yesterday let me tell you, I realized how dumb I am, those people know a lot more about AI and technology, I'm dumbing it down for myself mostly. What we've on the seen here we're talking about AI what does it mean and what does it mean in the context of employment?
Artificial intelligence is a really broad term there's a graphic that is on the screen right now that is essentially a Venn Diagram one large circle that just says artificial intelligence.
That's a broadly used term to show that computer science can solve problems that typically humans solve, subset of AI is machine learning that's where we program machine it's can understand our reactions and understand the data input and make decisions based on that. One of the things that maybe everyone here or most people are familiar with is Spotify or Apple Music, something like that I like this song, this is a Genre of music I enjoy, it will bring you for recommendations for new artists, things like that.
That's examples machine learning we're telling the machine what's do and using a algorithm to decide. Subset of machine learning is deep learning. That is where we really get the nitty gritty with AI when it comes to AI with employment, what that is -- it is still terrifying to me, these machines are not using algorithm to make recommendations they're actually learning themselves, using kind of neural networks that mimic brain processes that days data pools and take it's the information and learns it on its own. We're going to talk broadly and narrow it down to the employment context. One definition that I particularly like comes from -- there's a image on the screen of a piece of legislation called the national artificial intelligence initiative act of 2022.
Congress has defined AI, as a machine based system that can for a given set of human defined objectives make predictions, recommendations or decisions influencing real or virtual environments. That's our kind of high level definition. And what was eluded to earlier, we were -- this morning's talk is the AI Bill of Rights, White House published the AI bill of rights in October 2022, very broad kind of road map for how AI is used in all facets including employment. It gives a road map how it can be used responsibly. The way the White House defined algorithmic discrimination specifically is that algorithmic discrimination occurs when AI contributes to the unjustified treatment or impacts favoring people based on -- on number of races, race, sex, religion and age and disability. All those protected classes are encompassed by the White House's blue print for an AI Bill of Rights.
So, when we're talking about employment, one of the best places that you can go for any sort of guidance on how AI is used in the employment discrimination context, in particular with disability is the employment opportunity commission EEOC. So with the EEOC has defined how it is defined AI in employment in the employment context, using AI is typically meant a developer relies partly on a computer's own analysis of data to determine which criteria to use when making employment decisions.
I always think of these decisions in two separate buckets, right. The first is kind of the hiring and getting a position. It is used in -- I'll show you in just a moment, there's that kind of piece. There's once you're hired are working the monitoring, promotional decisions, raises, those sorts of things, AI can be used to help provide data to reach decisions whether you should get promoted or whether you should get a raise. We will talk a lot more in-depth how that can be problematic in just a moment. So there was in February 2022, the society of human resource management, which is a national organization which is really pretty incredible organization that does a lot of research and does a lot of -- very interested in AI in employment right now. They conducted a survey of HR professionals, for those who participated found that 79 percent of those participated use AI in recruitment and hiring.
41% use AI in learning and development, 38% in performance management. 18% in productivity monitoring. 8% in succession planning and 4% in promotion decisions. I will say, I think we were all very familiar that the law is very slow to keep up with technology. And I would venture to guess that even similarly with this topic in particular, things change in a flash. So, when I put this PowerPoint together, probably within two weeks I could have added ten more slides at least on a bunch more information. I would venture to guess these, even these percentages are a little bit outdated today.
So as you heard from Mr. Oliver earlier, the CEO of zip recruiter has opined, three-quarters of the resumes are read by algorithms in the U.S. That is astonishing, virtually any place you're going submit a resume, algorithm is going to be the first kind of initial screening tool used to decide based on the words on a page, whether or not you fit the characteristics of the candidate that would be successful in a given position.
So what are ways that employers are also -- some other ways that employers are also using AI. For example, AI is being used to create job postings.
So employers will sometimes go in, type in I need -- what I usually do, I'm going save some time today, but what I usually do is go to chat GPT and I type in write me a job posting for an entry level disability rights associate for a law firm. And it will within seconds punch out a job description that is pages long. And when I do this for lawyers everyone is astonished when I did it for people who do assistive technology, they're like Uh-huh, this is pretty typical, right.
It is being used a lot, this is one way. As I mentioned scanning resumes for key words phrases, monitoring employee performance and we use to conduct video interviews, evaluating facial expressions and aptitude tests that's a variety of ways that the AI is being used by employers currently. So I'm going to turn this back over to Mr. Oliver what you're going hear in this next clip is both he and the CEO of Zip Recruiter that CEO he mentioned earlier is going to talk about how the algorithms are used in particularly to review resumes.
(captioned video)
>> Anthony May: The image Mr. Oliver is referring to is one I can only describe as a sexy computer tower with the back removed and the wires hanging out.
[laughter]
That is the days with Elwood with the pink scented resume is really out you just need to embed within your electronic version of your resume, certain key words that algorithm the pick onto decide if you're the best candidate for this job. So given that general overview, I think everyone in this room, can already start to imagine and speculate how these things can go wrong. So what happens when AI attacks and what happens when AI can discriminate? I'll start by is talking about the primary Federal lays in existence that can be used to combat AI employment discrimination I refer to these as the big three. Everyone is probably familiar with these but I'll talk a little bit about them how they're used.
But generally when we're talking about AI in employment discrimination, the three key Federal laws we want to pay attention to are Title VII of the civil rights act of 1924 the Americans with Disabilities Act and age discrimination or at ADEA. We'll start with Title VII refresher for everyone here, Title VII, states that if employers must fail -- if employers fail or refuse to hire or discharge or deprive any individual of employment opportunities or adversary effect status of employment because of -- a variety of protected classes including, race color, religion, sex, which includes sexual or generation.
And national origin. We've got the ADA obviously we're talking specifically about employment. We're talking about Title I of the ADA. But also, for government employees we're talking about also Title II of the ADA. The language that I have got quoted on the screen no covered entity shall discriminate against a qualified individual on the basis of disability, in regard to job application procedures, the hiring advancement or discharge of employees. Employee compensation and terms and privileges of the employment. I like this quote, it speaks to the breadth of the ADA as how broadly it is applied and how we as advocates can be in looking at applying the ADA which is has been around for 30 plus years, how we can apply it an ever changing evolving technical -- technical area of employment like using AI.
We also have the ADEA which makes it unlawful for an employer to fail or refuse to hire or discharge an individual or otherwise discriminate I love the otherwise discriminate it is such a good catch-all category against an individual with respect to compensation terms, conditions, privileges employment because of the individual's age. You also can't under the ADEA limit segregate or classify employees in a way which will deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect their status of employment because of the individual's age or reduce the wage of rate of any employee based upon age. These 3 laws I'll talk a little bit about current litigation in a little bit, in a couple of minutes.
We'll see how these 3 laws already started to be used. We've also got on the Federal level a lot of great technical guidance coming out from Federal agencies. One of those I think leading the charge is the EEOC we'll talk about it, EEO has issued guidance in the ADA act, in use of software algorithm and artificial intelligence to assess job applicants and employees I entire everyone to look at the technical guidance and it is real he great in explaining while people think AI is the emerging tool to use in standard employment. The standard rule as ply, the ADA applies this gives a basic overview of the ADA, it also gives specific applications to how the ADA relates to AI.
It also explains, I think what we were talking about earlier we're going to have some really great legal conversations over the next couple of decades about employers being responsible even if it's administered by a third party vendor. There are challenges to that, that were eluded to earlier, that I'll also talk about in a little bit. Also the EEO's guidance talks extensively about the duty to provide reasonable accommodations. Which I think is something that I think employers constantly need to be reminded of. When I was giving this talk yesterday, I was approached by some folks, who do -- basically the disability and accessibility Directors of a very, very large tech company who said that they have been coming into issues where people a ply and it says you have the right to request an accommodation.
Great! How do I do that?
And there's no explanation. There's no email this person, there's no fill out a form. So while that right, while they say they are naturally complying with Federal law, which I am sure we all heard earlier, that's not the case. They're not functionally required to comply with the Federal law. You should put an email there, give them contact information, give them someone to talk to. And then also this -- the EEO technical guidance provides specific examples how employment discrimination can occur with the use of AI. It is a great technical model someone of us mentioned earlier the DoJ has issued guidance both are really great models to take a look at.
So we heard earlier about the black box problem. I'm going talk a little bit more about that what that means -- so, what in terms of artificial intelligence what I have on the screen right now is a screen shot from an article written in the Harvard journal of law and technology called the artificial intelligence black box and failure of intent and causation.
What that essentially means is -- well you know what I'll turn it over to Mr. Oliver he will describe what the black box problem is.
(captioned video)
[laughter]
The graphic is we see on the blew labeled as black anuses pulling up to the plant, spitting out the angry meat twigs, what happens in that Slim Jim factory how do we get to the discrimination the answer is we don't know to me it is terrifying I'm not a tech expert to any stretch of the imagination. To have a piece of technology that creates the outputs, to reach the conclusion is a bit distressing. We'll talk in a few moments about what effort haves been made to try to figure out that problem. So as many of you have probably famously seen a lot of people in the tech industry have strong opinions about AI, the use -- rapid evolving use not just in employment but in all spheres of our lives.
With Elon Musk, famously reaching out and calling out for tech leaders to put a pause on the use of AI we need to understand it better we need to use this responsively. So some other examples we've seen in terms of AI, how it's been used I think Amazon is a really good example of a tool they developed they were using that they ultimately scrapped to their credit. This was screen tool used excluding women from the positions how does this come about? If we know there's a discriminatory output, where is it coming from? We know that working backwards we've got this discriminatory output and in between there's the black box but what is going in that is resulting there so I'll turn it back to John to set it up.
(captioned video)
[laughter]
That was Goldman Sachs by the way you can imagine the Goldman Sachs all must have a great lacrosse team. That piece is important, it is a lot to unpack there, what we've been learning is obviously we've traditional systemic issues with employment hiring we got a lot, we know there's a wage gender gap we know people with disabilities have historically had a lot of barriers to have gainful employment. So what we're looking at is this kind of historical view of oh our company, these are the people in our company who have done really well based on all these sort of biases that have already been baked into the human process. Then we're feeding that information into AI and it is saying oh yeah, of course these are the people you're going to want.
You're going to want white, straight men of certain age those are the successful candidates. Even when as Mr. Oliver explained, even when you try to train the AI not to pay attention to those things it will pick up on other miscellaneous factor that's are not at all related to the position itself. But were based upon resumes for people who have been successful in this position. Again, in this instance we're seeing males saying I was on a football team. Football team the AI will learn that encourages that is great team leadership, teamwork but football teams are generally made up of males.
So while that quality in and of itself has nothing to do with the position, it is going to be a correlation factor that the AI is going to decide this is relevant, this is who you want.
That's what then tends to screen out and prevents opportunities for people who don't fit that model. And when we're talking about AI for people with disabilities at work so again we talked a little bit about this earlier, 8 of 10 of U.S. private employers use AI to track productivity metric of individual workers this is huge, because if you're a person with disability that requires you to step away from your computer for any number of reasons, you're being monitored. This could ultimately impact whether or not you meet your goals for the year. Whether or not you get a raise. Or whether you get a promotion. I will share that I'm a type one diabetic I have to take moments out of the day, use the bathroom more than some folks, I have to test my blood sugar I have to take insulin.
And if I'm doing all of those things, and taking time out of my 8 hour workday the AI is saying, this guy is not there, he must not be productive, you're not getting a sort of reasonable accommodation or some sort of process to understand why that is, it can result in employment discrimination. Also, similarly we're talking about people who interact with sort of screens and cameras. So when we're talking about in the interview process, if you're being interviewed by an AI robot you have for example CP, you have these disability that is fan first in certain ways or people autistic you're not giving the facial expressions that the AI is looking for. You can be denied employment opportunities.
So you can imagine that there's so many different barriers there's already a lot of bearers with technology itself being accessible. So, again, going back to that point earlier about productivity, if you are a person -- if you're blind you're using technology at work. Let's say you work for a call center or something like that, the tech is as it has been for years not working correctly. You're not getting the prompts that you need, it is not working with your screen reading tools you're spending hours and hours of your day submitting tickets to the tech -- the IT people, they're taking time to work on this. You're taking time out of your day.
Are you being then penalized for not meeting your quotas? So I'm going to talk very broadly, so glad that Bryan and Megan and Maria touched open this earlier, some of the broad I will just categorize them as legal efforts to curb discrimination. So start first at the global level then I'll talk more about Federal level, state and local and I'll talk about active litigation that is going on. The first thing as Megan eluded to earlier we have the European AI effect, pasted a week ago, very early stages, is one of the first -- I won't say the first, there's a tension as to this or New York City's AI employment law are really the first to come out and have some sort of comprehensive guidance on the use of AI in particularly in employment. EU's AI's act has been in development for some time. It has been just released. Only time will tell how much impact it will have, it has a very broad I think very strong regulatory piece including outlawing and just completely prohibiting certain uses of AI.
Now I was reading an article the other day about experts in the field of opining on how the EU's AI impact will impact other nations and the -- just the globe in general. And I -- this might be my own personal view I tend to agree with one of the thoughts is that while the AI act will be a good guiding post, other countries particularly U.K. and the United States will probably not take as strong of an approach given our frankly capitalistic society and free market enterprise. Our thought on the acts. AI act gives us learning lessons and Congress will adopt something as similar and as strong, only time will tell. At the Federal level I mentioned White House blue print, which was issued in -- for AI Bill of Rights which was issued in October 2022, pretty comprehensive document it provides a road map talks about what we need to be thinking about in terms of responsible use of AI and again this is in a wide variety of context. There's a piece talking about employment, we're also talking about housing decisions made using AI, criminal justice system, when we talk about AI it's not a new phenomena, right. For years the criminal justice system has been using predictive policing models to look at we'll take data from police reports and figure out which areas are high crime areas in the city that we really need to increase police presence.
Where do you think? Those are areas where you're over policing traditionally over policing marginalized areas, bi-pop neighbors, those are the area that's are often overly pleased you're using data based upon the over pleased areas to say we need to put more police there. This is the vicious cycle and vicious circle. The AI bill of rights talks about a variety of ways that AI should be used responsibly in all of the contexts. I'll give another plug of the EEOC they have done an incredible job of being at the forefront of the issues. In fact, have made it a priority in the next five years to really focus and hone how is AI being used in employment? How are the laws we have currently in place going help impact that?
What guidance can we offer? I mentioned that earlier the DoJ issued guidance as well, there's a joint statement I think is really interesting if you're interested in looking from the DoJ and EEC the consumer finance protection bureau the FTC all on AI intelligence. Another area we overlook when we're talking about employment, I don't do labor law, is the labor relations board, they have issued internal memos and guidances, one particularly area that I think is really going to be, to me this kind of overlaps with the criminal justice issues we've talked about and heard earlier today, meet withing employees and when their conferring and if you're use to go spy on employees when they're gathering to talk about what they want as a union, that's unlawful. You can't do that.
So there's some good language from the NLRB as to where, they're focusing on AI if the employers are relying on artificial intelligence to screen applicants or issue discipline the employer as well as a third party software provider, may violate section 8A3 of the National Labor Relations Act. That something to be cognizant of and keep in mind, we're talking about theories of liability and third party vendors.
You heard a lot earlier and the often the screen is the President Biden's Executive Order, issued in November 2023, very sweeping Executive Order that calls for Federal government wide approach to responsible implementation of AI requires developers of the most power AI developer to share their results and other critical information. That is so key, as we learned earlier the only way to figure out where the problems are and how to solve them is by knowing what is going on. When companies are keeping that information private they're claiming it is confidential or related to a really interesting conversation with another lawyer that does class action work she has run up into instances where companies are saying oh, no we keep this data, it is attorney client privilege, attorney work product. You're preparing for litigation you know you're going to get sued that's the way you're shielding us from the information. Interesting debate there are going to be a lot of legal areas we'll have to litigate in the next couple of decades how AI is being used how we get access to the data.
One thing I think that is really great about the President Biden's Executive Order incorporates some standards from the national Institute of standards technology, NIST they have an artificial intelligence risk framework I think this is one of the really one of first in the United States at least, comprehensive guide on standards, tools and tests that can be used to ensure AI systems are safe before they're used in the public. And the Executive Order also calls on NIST to update that and to provide even more guidance which is really great I think, is a good model if you are particularly an employer or company developing this kind of technology to keep in mind because these are really best practices. What I hate to say is, we are living in a country in a world right now where best practices is what we've got to push there's the lack of laws right now really kind of not requiring anyone to do this. We're just hoping they will do this out of kindness of their hearts. Some companies are, some companies are not.
I'll talk a little bit more about that in a second. So, let's talk about the state and local levels I want to talk a bit about the New York City's automated employment decision tools law. So this really -- passed in 2021 it just took effect this year. There's been lots and lots of debate and lots of discussions about AI, AI's regulation in employment. So technically, I think this -- the passage of the New York's AI law preceded the EU's act it really was the first to kind of go out on a limb and say if you're using AI in employment, here are the things you've got to consider.
Here are the ways that you have to maintain data in the types of things you have to do as an employer and the types of information that you have to make publicly available, so that we can determine whether or not, there are discriminatory outputs. So the uses the term automated employment decision tools or AEDTs. It requires that those tools be used to audit tools forays, and gender bias. Race and gender bias. Only race and gender bias, right? [laughter] As Maria said early it excludes disability, there's not enough data on disability if you require there be requirement for data on disability there might be some data on disability. Even if the infant say we're seeing a lot of problems with the New York City law I read an article recently that just said it's been a bust so far.
Researchers at Cornell said 18 of 391 New York City employers has actually posted their results online. I think that they're there are going to be arguments made down the road while you're requiring us to maintain the data we'll assert some confidentiality and privilege over it we can't make it publicly available we can, we can comply with the law by still keeping the data and maintaining the data. It also, there's some other issue with the law itself. It provides a very narrow definition of AEDTs, so it gives some loopholes. Really compliance is like, it is the honor system. Just do it, you're going to do it, right? All right, great! So finding ways to enforce the law and pushing and requiring companies to do it. I think it is at least in the United States going to be the first kind of Guinea pig to see all right what is it going to require? How is it going to be enforced what are we going to do to monitor?
Let's talk a little bit about other areas -- New York City is really only one that has developed and passed a comprehensive law. I always like to give a plug in Maryland we've got a law that prohibits employers from using facial recognition tools in job interviews without notice. Privacy requirements. It doesn't touch on disability and entitle II reasonable accommodations it address AI in some respect. We have got a number of other states also looking into passing comprehensive AI bills, particularly with respect to employment, Washington state, Hawaii, washing, DC has tried and failed being they're trying again, which is great.
We're also seeing at the state level, a number of Executive Orders similar to President Biden's, so Maryland, great. Governor Moore recently issued an Executive Order in January of 2024, calling for the responsible use of AI in government and encouraging administrative and agencies in the State of Maryland to use AI but to use AI responsibly. Virginia, California, Pennsylvania, have all followed suit, I suspect more will come down the road. Let's talk about the litigation effort because I think that's why we're all here, this is the -- I mean this is really, we're the -- the next generation. We're on the forefront of really pioneering a new area of civil rights laws so we've got to be creative about this.
So as I mentioned earlier the EEOC has made a biggest and made a targeted enforcement effort on the use of AI.
So one of their more recent successes is in a case called EEO v. iTutorGroup, it is an online organization that provides hires individuals in the United States to do teach English as a second language in Asian countries. What the EEOC found was that applicants of a certain age of higher age groups applied for positions were automatically getting screened out by the AI tool. So in September 2023 the EEOC settled entered into a Consent Decree, we love them, because they have a lot more teeth than a typical settlement will have. It allowed, among other things, it allowed for 365,000 dollars for individuals screened out. It gives robust injunctive relief, it required training on biases on certain employers it requires monitoring, monitoring in the group to report to the Court what it is doing to maintain, to monitor the results to see if there are discriminatory impacts and what it is doing to fix those things.
So really one of the first of its kind. The next case I want to talk about is Mobley v. Workday, in this case, Mr. Mobley, did a class action suit against a Workday, is a third party vendor to screen resumes he claimed he provided 80100 jobs never got a response he alleged, that workday is an employment agency, again we talked about that earlier today, right. The theory is an employment agency, and therefore its discriminatory use of AI violates the law.
This case, in particular, well, so the case was dismissed in January of 2024 because a Federal Court in the northern district of California said that the allegations in the complaint weren't sufficient to enough to establish a employment agency. It did dismiss the case without prejudice, in fact, Mr. Mobley refiled the suit, the real key take away and important piece in this particular case this allegations climbs under Title VII under ADA and ADEA. You've got all those 3 big laws at play here. This case is really I think one to watch. It is going to be one where we're going to see how the interplay of these laws work how they work when it comes to AI and employment discrimination.
And really, what happens down the road is going to being really interesting I think that the dismissal of this case is a really good framework for me us to think about how we plead the case given the facts of the plausibility to dismiss. Another case is real women in trucking v. Meta platforms this is a case pending in EEOC right now under investigation in the EEOC involves a group of female truck drivers who allegation that Meta or Facebook was screening out and failing to provide opportunities for employment opportunities to women truck drivers.
So it was the algorithms being used is only -- was filtering these opportunities to males of a certain age and was excluding them from the opportunity to even have employment.
And so, that's another case that I think is really interesting we've got to figure out -- really how the EEOC process itself can be used as a great tool to gather information and figure out what is exactly happening here. So in this case, you've got the -- the real women in trucking these are available online I know you can find at least the complaint and the underlying documentation that real women in trucking submitted.
But for those attorneys and for those Plaintiffs in that case, getting a statement, having the employer provide a statement, or -- not employer but in this case the third party Meta, provide a statement why they're not doing it. This is a statement. This is a case to keep watch and keep in mind, there are a number of cases going on around the country, to think how we'll approach these cases what are the best practices what are the -- what information can we gather that we can all share with each other, so that we're bringing these cases to develop the law in the most productive way without creating barriers in the law itself.
You know I don't think it is lost on anyone here that often times when -- at least for my perspective as a Plaintiff's attorney when we bring lawsuits we say oh, no the ADA law applies to ADEA in employment. The Defendants say it cannot be done, it doesn't mean it doesn't apply, someone has to be first. Convincing not only the other side but a judge that yeah, I get it. These are high contents I think, but at its basic core, what we see is a discriminatory output. There's something going on here, so there's got to be some opportunity to develop some discovery and learn facts. I want to talk about best practices things to consider, if you're an employ applying or an individual employee or you're advising employee and also if you're a employer things to consider and think about. If you're an employee, one thing I tell everyone, know your rights. Know the law.
Familiarize yourself with the EEOC guidance, know what the ADA says you don't have to be an expert or go to law school. That's one of the best things with the guidance we see issued from the Federal agencies is that it really is a very you know -- I won't say basic it is a very easy to understand and easily comprehensible guidance that you can kind of understand what your rights are. Familiarize yourself with that also know what is going on locally and in your own you know whether it's in the city like New York City or in a state.
What efforts are being used to curb discrimination, what does the law require, when does it take effect, know that, so you know while applying for jobs or at a job that uses AI, know what kind of laws would apply so you have a better knowledge, probably than your employer does. Know when to exercise your rights, right. Know when to request accommodations. I think this is very tricky, right. Not everyone -- it's not apparent is that AI is being used, you have a questionnaire that I have to check box and answers question that is AI, it may send different answers that you're giving or may give you a different answer if you're responding with a question, in AI, if you're a person with dexterity issues and pause to respond. That can be held against you. Things like that. You have a right to request reasonable accommodation as Bryan pointed out earlier today, what that reasonable accommodation is and whether it is actually going help you or not, is really kind of concerning.
So, if we're using AI to interview an individual, and look at their face and we decide well, we need a -- we need a reasonable accommodation because of you know my -- because I'm autistic what is the reasonable accommodation, do we not use the tool at all? Does it not get used at all in your decision-making? Do we have a human do that instead? Does that fix the problem? And I think I was at this conference yesterday and gentleman who is very involved in the tech piece of it said, I mean he opined without me even speaking about it, that you know once you ask for a reasonable accommodation does that get held against you? Are you now this person they already know has a disability? You're asking for things that we're going to have to give, does that get held against you? Is that going to result in a discriminatory output?
One thing I also encourage if you're in any position that uses AI to track your productivity things like that, asking your employer for your own data, if they're maintaining data on you they're evaluating you based upon the use of AI, ask to see it. We're going to suggest early on this is an early form of discovery you feel like there's going to be discrimination, you feel there's going to be a claim down the road, it is nice to preserve that now. Get the information where you can, so that when you -- eventually have to knock on my door I can say, this is great. We've already got some pieces in place to know there's at least, there's a good plausible allegation we can make better based upon the facts we know.
And again, file a charge with the EEOC. And engage in a conversation, have the EEOC do some work for you. Can you also file a complaint with the DoJ in certain instances we know the EEOC and DoJ they have a lot more expansive ability to get more information that private citizens don't have access to. That's one way to look with local PAs I love shout out to D RN disabilities right of owes owe and Connecticut you're all great. One way you can get more information.
On the other side for employers, again, familiarize yourself with the law. Understand it. It doesn't really change whether you're an employer or employee, know what the law is and know what the law requires know how to make sure you're using AI you're doing in a way that it doesn't run afoul of the laws. The biggest piece of advice for employers is question, question, be aggressive if you're using vendors hold their feet to the fire.
One thing that Bryan mentioned earlier was you know these vendors will listen to their clients. So if you're a client you're saying I need to know XYZ. I need you to explain this to me. If they're saying it's proprietary information, we'll not share with you. Maybe you need to go to a third party vendor. If they're using proprietary information, guess what? You're going to still be responsible. EEOC guidance and Executive Orders use other companies that consider potential discriminatory impacts of AI, IBM is a good one that does a good job of internal monitoring to ensure their products are not resulting in discrimination and keeping good data on it.
Routine internal audits are also critical to the responsible use of AI and employment. Conducting those audits looking at the outputs looking what is going in and going out. Constantly trying to tweak, figure out those results are coming out and fix those so you're not seeing discriminatory impacts. And again, question vendors if you're an employer you don't know what questions to ask -- hire an attorney. Hire me I'll go in and ask I don't have any problem asking questions. So that's one thing that I would highly recommend.
And again, when we're talking about reasonable accommodations, privacy notices, you have to give employees adequate notice and a process for requesting accommodations. I think in one instance there was an employer who used AI that required an individual to use the entire tool which I believe was inaccessible. Finally had someone allow -- assist them with completing it, once it was completed -- they fixed it to make it accessible so they could complete it, it was also discriminatory. At the end it was by the way do you need a reasonable accommodation? We've already completed this, you got results without an accommodation. Does that get -- do you just put that away? What happens again kind of see the same themes overlapping.
So one point I will make it is not just AI impacting people in employment. It is not just AI impacting all facets in our life, but us, as lawyers, we are absolutely impacted by AI and so this last clip I'll show, is by an individual name Eric. He is the Director of the Stanford's digital economy lab he has some thoughts on AI in particular how it is used in the legal field.
(captioned video)
[laughter]
Anthony May: Coding joke I didn't get it at first. So there's a GIF of Seth Myers I wanted to give a shout out to the other late night host, I have a couple of minutes I am happy to answer questions it is a lot of information in an hour, I'm happy to answer any questions.
(applause)
I can repeat them on the mic so folks can hear. Yes question
Attendee: In the benefits context, folks have used a due process challenge to kind of get through this black box issues and I'm just wondering if you have seen that, I mean it would only be for public employers, in the private context, are you using that -- you're terminating taking negative action against me we don't know why, we can't tell the steps because the algorithms is not available to us, have you seen that in the employment context?
Anthony May: No thank you for that theory I'll apply that. We have just seen a lot against private employers. So I haven't come across that yet, it's something I have not thought about or come across, we have seen litigation against public employers that brings in Title II, really from my vantage point, everything in the kitchen sink, let's see what is out there, let's think about it individually and collectively.
Figure out if it's a good approach obviously you don't want to throw everything out there and create bad law. But I think that is particularly for government employers, which is why I think we're seeing a lot of these Executive Orders to directed towards government agencies we're trying to prevent that from happening I think that's a great point and a great way we can think about bringing these kinds of claims in the future.
Attendee: I have a question about monitoring. Because you've brought up the point about I too take a medication that is a shot, and like I have never asked for reasonable accommodation I work at home I don't need it. Like so now that people more people work at home I've heard it's ok we're going to monitor them or monitor them, this sort of brings up the issue of, ok, well what does that mean and will more people come out of the closet with their disability, so to speak? It was a interesting phenomena to think about that. I'm in the corporate world, different than most of the advocates here but -- I hear a lot, we're not using it for promotion, we just want to see how many key strokes it takes. Like, really?
Anthony May: Yeah, right.
Attendee: Really that's all you want?
Anthony May: Yeah.
Attendee: But so, it is just interesting to me and -- just, sort of that long-term impact of what does that mean especially as more people are being asked to go back into the office, and if you're staying home we're going to monitor you. I don't know what that means?
Anthony May: Especially when you're getting a reasonable accommodation that allows you to work from home --
Attendee: Right.
Anthony May: We'll let you work from home as long as you're being productive, right. I think why having conversations in the space we're all in the same world, generally having these conversations with folks that would not normally think I can do my job fine, I don't need accommodations if that is a piece you're being monitored on your productivity, in generally don't work at a place that does that, I would be curious and want to know on a monthly or quarterly basis, well can you give me these outputs I can see what I'm doing, where I'm not being productive and why?
And if it is related to my disability then we need to have a conversation about why this -- I want to know how you're using the data, really using it to just monitor and figure out key strokes and what are you using that for?
And do I need an accommodation based upon that? It creates a situation where you need to have that conversation. I think just knowledge is power. I'll go over here and then here and then -- we're overtime but I'm happy to hang out.
Attendee: Hi, my question is like, if you raise the issue of like I use AI tool to screen read, that is protected against retaliation by an employer?
Anthony May: I don't think I understand the question.
Attendee: If you ask a employer, if you -- I use AI tools to screen performance evaluations, is that something in and of itself that you can be fired for.
Anthony May: For simply asking.
Attendee: Yes.
Anthony May: If they do, please call me, you know -- you're entitled to that information, you should be aware of the tools and how they're monitoring you.
Attendee: I'm wondering about like any potential overlap in bringing cases under ADA or AEDA under that the privacy laws like in Illinois, I feel like there could be a case to make if you're being monitored, I mean, very limited application obviously depending upon what jurisdiction you're in, you're collecting some sort of biometric information so, that cannot be allowed and then also this can be used for used against you for discriminatory reasons. I'm wondering if you've ever done that or ever seen that before.
Anthony May: Yes it's a great thought, what we just in general more in general think about is, all right, where ask the venue, where are the state laws maybe, better and more robust? California has really great protection employment protection statutes. So you might think well maybe I want to bring a first test case in a place that has both Federal, bring Federal and state law claims decide which forum I want to be in, which ones can be the best. You want to consider what remedies you're looking for, obviously. And whether the biometric privacy act only limits your remedies you want to have something more expansive that will give you anything you want.
Those are, again these are the kind of critical creative lawyering that we all need to be engaged in and engaged in collectively, because you know I've got, thoughts in my little office as I'm spinning my wheels last week I had a conversation with Megan and Bryan, you guys are great I'm learning more than I knew, I think continuing the conversation how we can strategically bring this kind of litigation using state and Federal laws kind of in tandem is the creative lawyering we need.
One other thing I will mention there's a great book it is called the algorithms. It is by Hilka Shelman just started she is an investigative journalist interviewed over 600 people, tech industry leaders, individual employees, lawyers, all about this very topic. It is a great, so far it's been a great book.
I'm book I'm listening to, which is also weird to me it is wonderful I encourage you to check it out, it gives you a lot of insight and real life implications some of the best stories of the employees found themselves out of gainful employment because of AI with that I'm going to be around. I'm happy to chat but have a great day. Enjoy lunch!
(applause)