AI’s secret super power: Fighting bias and promoting equal opportunity

foreign [Music] [Applause] so we're going to dig right in we've only got about 30 minutes and we've got this is a big topic to talk about and hopefully we'll have some time for you all to ask your questions as well so the main topic here I think that everyone wants to hear about is ai's ability to fight bias promote equal opportunity so you are on the front lines of this yes so what public or private examples private sector examples do you find out there in talking with your clients are the most promising and exciting certainly you know the um we've really entered a new age you know the AI age and you know in terms of equal employment opportunity and making sure that everyone gets to be a part of that that's something that HR professionals eeo professionals need to look at both in the private and public sector I know government Regulators will be looking at it they already are um but you know we we called this AI secret superpower because I think in the long run and by long run in the AI world that's not that many years it could be in the next even months but certainly in the next three five ten years I think you will see that AI will change the way we look at eeo in many ways that the structure that we have now in this country goes back to the 1970s and you have systems in place where you're looking for adverse impact you're looking to see if there's um basically under the four-fifths rule or using some sort of statistical analysis whether there are disparities in hiring or promotions and you know in that that framework remains here but AI has the possibility of eliminating bias like completely eliminating bias from the way that we look at employees because it is neutral and I think there's a lot of interest in um government entities like the one I led ofccp or or the EEOC and you had commissioner Sandra Ling here yesterday he's also at the Forefront of these issues I think there's a lot of interest from federal Regulatory Agencies and state agencies and what AI can do having said that there's a lot of skepticism as well so you know right now your question in the and I was the former ofccp director the office of federal contract compliance programs that's a federal agency that regulates Federal contractors many of you are federal contractors they use the statistics to assess whether there's disparities based on race and gender in hiring and in pay and and was the first agency and when I was director to publish guidance on artificial intelligence in employment and so um you know but I will tell you both ofccp and the EOC are behind the times there's a lot more guidance that they need to publish in this area because about 90 I saw a stat that said 95 percent of Fortune 500 companies are using AI already quite extensively in employment and of course Eightfold we're here today to talk about artificial intelligence in April um and and who has brought us together at cultivate Eightfold is a great example of that the use of AI and employment to try to enhance equal employment opportunity so um you asked me what ways it's being used in every way AI is being used through chat Bots to help someone an employee determine what benefits would be best for them it can be used to identify underutilized labor in your Workforce who where you could maybe give them an additional assignment that will be useful it can tell you through looking at skills who might be a good person to promote into another position and and you know the big one of course is an employee selection it can be used to identify uh look at 10 000 resumes and determine who would be the best candidates for a particular position helping you to use neutral criteria to identify the best talent so it's it's being used a lot now it's going to be used all the time in in just a few years I would say so you mentioned that various government institutions are a little bit behind the times with respect to regulations how do do you think policy makers assess and evaluate different types of AI that are popping up and determine when they should intervene yes so I'll tell you and you probably know this that you know you have the New York AI law you have the EOC who's published some guidance in the area of disability inclusion and AI you have the ofccp which has indicated that it's looking at this area you have the White House which has taken sort of a whole of government approach indicating that they're going to be looking at AI through numerous different agencies and they want to publish guidance and regulations in this area so you know you do have a lot of activity at the federal and state level but the point is that in my view even though there is a skepticism healthy skepticism toward AI because of the structures we have in place already and the concern that there will be because of systemic racism and sexism and employment that we still have not been able to eradicate as a society there's a concern that AI could could replicate replicate or even enhance that that's a concern the idea that you might not be able to get a job anywhere because you're just not you're not on the right side of the algorithm and that that may impact People based on their race or gender in a way that's disproportionate or causes adverse impact those are legitimate concerns but I will tell you um that concern exists today in terms of human decision making unconscious bias Affinity bias the way people look at resumes how many times have I know we train about this in HR and in eeo but how many times does someone look at a resume and look at a hobby or look at the fact that they went to the same school and that's all they talk about in the interview and oh that person's a good fit I mean that happens more than we would like to admit and AI doesn't do that and that's the benefit of AI so as long as the federal Regulators are interested in companies going in there having AI ethics councils like Eightfold has for example I'm on the AI ethics Council along with Vicki libnick who's the former uh acting chair of the equal employment opportunity commission you you want a AI companies having an AI ethics Council you want your own companies may want to have an AI ethics Council that you're considering AI you're looking at the impacts of AI you're making sure that it's not having a disparate impact based on race or gender and assuming you do that AI is going to be great for eeo because it can identify People based on their skills how much more job related can you be than picking people based on their skills and so even if there is an adverse impact in a given case but it's based on skills you'll be able to validate that so there's um in the long run in my view AI is going to be a boon for equal employment opportunity yeah I mean in a way do you think it might be more likely that organizations in the near future who don't use AI will be more likely to be penalized because in using humans they're going to be considered more likely to be biased I mean I think it's a various dew point I think anytime you bring in you you have human decision makers which every organization has of course um but anytime you don't use Ai and you're only using humans and you have a disparity what a regulatory agency like EOC or occp is going to consider is oh is this disparate treatment are you treating people differently because of their race because of animus because of historic discrimination because of stereotypes and stigmas that exist because of job segregation that you think men are appropriate for these jobs and women are appropriate for those jobs Antiquated thinking that is based on stereotypes Regulators are very concerned about that so if you're only using human decision makers there's always that concern but in terms of whether that will be soon or later I think I don't know if in the next six months to a year Regulators are going to be saying oh you're not using AI so that must mean that you're discriminating but I do think I'd say in five to ten years that if you're not using artificial intelligence as a significant part of your hiring process that Regulators are going to be concerned yeah because they know that the AI is neutral presumably by that time uh you will have The Regulators will have put regulations in place you won't be able to have ai that creates adverse impact or that is is discriminatory in nature and there are there's good Ai and there's bad AI let's be clear about that I I'm on the AI ethics Council for April April to me is good Ai and it could be helpful in the long run but there's bad AI too that's not that where they're not thinking about these things and the AI itself may be relying on stereotypes and stigmas about who should be in a certain job based on historically who was in that job that's not the AI you want or AI where you can't explain why it's reaching a certain decision like if you ask it well why are women in this particular group why are women being recommended for higher sixty percent of the time and Men eighty percent another time why is that Gap there you need the AI to be able to explain that to you you need to be able to say oh well it's because this particular skill was looked for and and the availability of that skill in this particular Workforce happens to be there's more men who have that particular skill than women or vice versa that's you need to be able to have an AI that can tell you that if not you're going to get in trouble with the Regulators right and to that point I think it's important to understand if you're looking at AI tools some of the best practices for implementing an AI tool within an organization so what what would you consider to be some of those best practices well one I would say that and you know I'm speaking to the choir here I think to some extent you're all here you're interested in AI there's many companies that are afraid of AI right now because they're seeing this regulation they're concerned about it and it may be because of chat GPT in particular and a sort of generative AI or telling people in their organizations hey we don't want you using AI at all we don't want you interacting on workplace computers with artificial intelligence you know that's not going to work for that long right because as a business you're going to have to be comfortable with AI to be competitive you're just going to have to be and if you want your if you want to compete for the best employees and compete with your your fellow companies in your industry and get the best employees you're going to need to employ AI to look through thousands of resumes and identify those that that top talent which incidentally would also most likely I'm speaking here from an affirmative action standpoint which ofccp that's the regulator of affirmative action in the United States AI can help you identify underutilized Talent you asked me before what are ways it can be used there's diversity job boards where you can match uh underutilized Talent underrepresented groups with skills that they have that you're looking for and that can help you in your affirmative action your Outreach and Recruitment and things like that but ultimately it will be able to help you compete for the best talent because you'll identify you're not going to spend lots of time I uh interviewing individuals who don't have the skills that are needed you're going to be focused laser focused on those that have those skills and be able to pick the best person yeah I mean I I think that it has so many points of value that we just need to take a step back and think about what are I mean somebody said it in the last session like what are the problems that we're trying to solve and just kind of hone in on the tools that are going to be appropriate for those to solve those specific problems you mentioned a couple of minutes ago disparate impact and the fact that AI tools can be used for good and they can be used the other way and one thing that just kind of always sticks in my mind is the fact that AI tools are by and large programmed by a very limited group of people who are you know in a certain demographic so that they might have this built-in bias associated with them so an organization that's deploying AI tools has sort of a responsibility to ensure that um the AI based technologies that they're using are in fact neutral so if if you're an organization who is deploying an AI tool how can you properly monitor your AI based technology to ensure that the tool itself is not having a disparate impact well you need to be testing it you need to be using common sense so for example and you probably need to audit it or at least have the company that you're using that provides the AI be doing an audit and and telling you the results of that so for example you know one concern in the AI world that I've seen among Regulators is let's say and this is this is something I used to talk about ofc at ofccp quite a lot there are certain Fields where if you look at the higher level positions in those fields and by by Fields I'm talking about law Finance Consulting accounting tenured faculty at universities Tech Executives where those executives are predominantly male and often White and so 80 male 20 female you see that in law firms even today uh you look at the statistics that are that are published by the the uh they're called the nalp nalp uh ethics survey and what it tells you is that 80 of equity Partners at law firms are men 20 are women those numbers have been sticky over time uh notwithstanding all the affirmative action efforts that have been done notwithstanding the fact that over 50 percent of those uh graduating from law school are women and about 50 percent of mid-level Associates are women and yet it's still 80 20. when you get to these higher level positions so one concern about AI is that if you go in there and you say okay here's our Workforce we have 100 Executives 95 of them are white men and we have five diverse Executives uh these are the executives that have done well the concern is that the AI will go in there and say oh well you have to be a white male to be a good executive because 95 percent of the time exactly it's a white male right and so you need to make sure your AI is understanding right in some way that there's been historic discrimination that the availability in in this area because of this historic discrimination it may have been 95 men 30 years ago but not anymore and that you need to look at availability in the whole Workforce and you need to make sure that you're you're not overweighting the demographic or waiting at all frankly the demographic consider characteristics of those individuals in that position when you're looking for comparators or examples that you want to find in the workforce generally you need to make sure that your AI is not biased towards men or towards whites or toward any group frankly uh that's really where Regulators where the New York AI law is concerned so what you want to ask a company and you could and this is what we do on the Eightfold AI ethics council is okay so what are what are we doing to see if there's any sort of algorithmic bias so if 50 men and 50 women apply for the same position and they have the same same skills same qualifications so you compare two sort of equal groups you switch their gender you see if that makes a difference you switch the race you see if that makes a difference there's a lot of different ways to test for bias but but that's you know that's one for example you want to make sure that you're looking to see if race or gender is changing a score or is changing an outcome and to the extent it is you need to eliminate that yeah I mean I think I think it's so important because AI does have the ability to amplify bias I mean if it's learning from the information it's being given right so if it's learning from your culture or from your individual organization that a certain situation is the case then it's just going to keep perpetuating that right sure like you have to teach it no actually this situation is in play because of a unique circumstance it's not something that you should continue to to facilitate yeah like if you ask AI for example if you gave gave AI 10 individuals five men five women who are leaders in our country and asked who would be the best president of these ten you don't want the AI looking at and saying oh well every president has been a man right every many presidents are This Tall which can sometimes be harmful to women exactly if you say like oh you got to be over six feet we often hear that about President things like that that can correlate with gender or or correlate for example without of being someone without a disability you want to make sure that you're not looking at these correlative factors that really are not job related how tall you are your gender doesn't make you a good president and so you know I just use that as an example because it's an example where we've never had a woman present hopefully we will soon but uh there are positions like that in Corporate America where there have been very few women or very few people of color in those positions and we need to make sure that the AI is not uh amplifying that as you mentioned and that's why you want an AI ethics Council that's why you want testing and auditing but in the long run let me tell you something if you ask most Americans and you still you still see this who would be a good next president if there's still a percentage of Americans who say that a woman shouldn't be president believe it or not and that's very sad you know that that still exists in this country but there is still bias people have bias that's why if you can remove the bias from the machine uh from from the AI which which you can and which will certainly be I think common in the next few years because of all these laws and Regulators if there is no bias it's better it will it will tell you no these three women would be the best next president or best next CEO or best next programmer right whatever it may be so speaking of the New York City AI law do you think regulations like this are going to become more widespread and if so how should employers be get ahead of this how should they prepare for something like this coming to their city or state well one I wouldn't I wouldn't run away from AI I would embrace it now I would start embracing it become familiar with it make sure your HR leadership is familiar with it is trained on it is is learning about AI use chat GPT I asked chat gbt today about the impact on of I asked it actually very bluntly I said does uh will is AI could it be positive for equal employment opportunity and the elimination of discrimination it gave me a very good answer it said uh yes it could be and it for the reasons I've given you but it also said But there are concerns that it could amplify bias and it was a fairly comprehensive answer it was a good answer I use AI um in enter my daughter is profoundly autistic she just difficulty communicating I've been using AI with her to help her to communicate with telling AI about my daughter and this was suggested to me by someone in the disability field where I tell AI about my daughter and I'll ask you questions and I'll say well my daughter said this one or two word response what could she mean by that can you elaborate on her behalf and then I'll show it to my daughter do you agree with that do you not agree and over time I found it to be very useful in it and it's suggesting to me things that my daughter would like to do and it helps her to be able to communicate so anyway there's so many uses of AI but it's important for your Workforce to become more comfortable with it if you tell them don't use AI period ever one they're still going to use it and frankly it's going to become impossible not to use it because it's going to be embedded in search engines and everything generative AI but you want to embrace that now and then the second thing is the AI companies that you utilize you need to be asking them do you have an AI ethics Council if we get audited and by occp or the EOS C or or any sort of government entity and they ask us why is there an adverse impact here based on race um can you explain it to us in a job related way is the ad going to be able to do that or are you going to be left telling the the regulator well it's a black box we don't know exactly why we're not certain if it's neutral uh but they tell us it is that's not where you want to be you want to be with a AI company that's going to help you if you get audited and be able to defend the AI as being job related yeah and by the way that's a skill set when people always say oh we're going to automate everything the AI will take care of that that's a human skill set being able to explain to decision makers why the AI is making the calls that it makes that's a job that needs to be created when you deploy AI within your organization so new human skill set that needs to be brought to the table all right so you mentioned Craig that many people are still afraid of AI granted that's probably not who's in our audience today but the people who are in our audience today do need to sell AI internally as being a Force for good so what leadership and communication strategies need to be in place in order to come away from a conference like this and say we absolutely have to be supporting hateful be supporting uh platforms and talent intelligence in our organization and being able to drive it Forward yeah so from the top down and and I think it's important to convey the message that we're embracing AI that we're at the Forefront of new technology in our organization uh it's the same with the dawn of the internet can you imagine companies now looking back that refuse to use the internet in Commerce or to have a web page or things like that you it just doesn't work or refuse to use Google or search engines or or Bing to be able to search things it doesn't work you know so it's the same with AI we're all going to be utilizing it in our daily lives just like our phones it'll probably be embedded in our phones it's going to be helpful to us and being able to generate documents generate schedules all sorts of things and to show from the top down that you Embrace that those companies that do that are going to be at the Forefront and then understanding always including though never forgetting equal employment opportunity or fairness saying we also recognize some of the concerns that exist about AI which are legitimate and we are on top of that we have an AI ethics Council we have a group of task force that are focused on IA to make sure that people are not left behind so yes we're going to be using AI to try to identify individuals in the workplace who could be able to advance you know maybe for you know in particular positions or opportunities but but look if you've if you it's already having an open door policy if you feel that there's a concern that you need an accommodation or that the AI is not doing everything is not identifying you for example feel free to reach out to this group and well you know it's that sort of Engagement that we're going to use AI we believe it's going to be helpful to you but we recognize the concerns and we have a task force that's dedicated to addressing it and then same with um like in terms of AI and employment like Eightfold it's the exact same thing saying that we're using AI to identify Talent it's job related we're identifying the best workers they stay longer at the organization all these sort of positive things that are helpful from using AI but that also we recognize that I AI can be concerned we don't want people to be left out for if you're individual with a disability and you're applying you can ask for an accommodation related to the AI now that doesn't really apply so much to April because it's looking at skills but there are some AI out there which will like as you all know which will uh interview you for you you'll be interviewed for example and I'll look at your facial expressions your eye contact it may look at how you speak or how you respond to questions sometimes this this is why the EOC public just a whole set of guidelines related to AI in equal in e in equal employment opportunity it basically said well let's say you're someone with autism and it's hard for you to give eye contact or you have you have anxiety disorder and you get anxious when you are in in a an interview where you're being interviewed by an AI or you're looking into a you're being taped or something like that and you need to be able to request uh accommodation so that maybe that that part of the employment process will be done separately or in a different way or or at least they'll recognize that if you don't have eye contact or if you take more time to answer questions or if you need to ask for breaks that that won't be held against you in the way that your interview is scored I mean those sort of you need to be able to request those sort of accommodations for some types of AI if your disability could be impacted by the AI so I mean these are things that you have to be thinking about as an organization yeah I think all good points of advice we have time for a couple of questions from the audience would anyone like to chime in with some thoughts for Craig hi um hi hi I'm Liz um and I was thinking about the comments from ashitosh around the advancement you know over the next five years the pace of technological advancement and then I also think about some of the Regulators that I've seen during Congressional hearings when we think about Tick Tock a few weeks back I'm not convinced that Regulators are able to keep up um with their technological advancements um and so what are we doing or what are they doing to help speed up that pace because with the growth that we're expecting I'm not sure that Regulators can even it's speak to some of what's going on now so I'd begin to get your thoughts on that sure I you know I don't think Regulators have have kept up I don't think they will it reminds me of other areas I I used to be a local government attorney and I remember when Uber and Lyft came into Miami-Dade County and they're like we're not letting anyone use Uber and Lyft and we're going to find you if you do and and and that didn't work I mean people used Uber and Lyft and they ultimately didn't find them you know so the and and you know that that that's improved Our Lives to be able to have ride sharing for example it's the same thing with artificial intelligence people are going to use artificial intelligence you can't prevent them from using it as a regulator we should be using it in employment we've seen that if you don't use Ai and you use human decision makers at least 80 men 20 women in a lot of these positions which is very harmful to the country because we're under utilizing our talent and it's the same for any protected group for African Americans for Hispanics for Asians for Native Americans for lgbtq plus employees for for women for women of color and for for people with disabilities all of those groups are underutilized in the workforce because of historic discrimination and AI is a way to address that and it's a neutral tool but it's a positive tool in the sense that when you ensure its neutrality and frankly if it's not neutral why are you using it for business either the whole purpose of using AI is to identify the top talent in a neutral way not in a biased way but as long as you're using it for that purpose and you're testing to see if there's adverse impact and you're addressing it over time that's what Regulators should want and that's what it should be that's what Regulators should be enforcing so I'll tell you on the AI ethics Council for Eightfold afold has always told us that they want more regulation that they care about regulation that they view it as a way for Eightfold to be at the lead among artificial intelligence companies because April is going to comply and other companies that don't comply are going to go away because it's too much risk for companies to use an AI um a platform that doesn't meet the regulatory guidelines that are set but at the same time for Regulators they need to be open to AI in the long run it's going to be beneficial for Equal Employment Opportunity it has to be one thing I like to often say again from my local government days is this we had the same discussion about autonomous vehicles oh well our autonomous vehicle is going to be harmful are they going to and yes there are individual cases where an autonomous vehicle can be harmful but in 20 or 30 years everyone's going to be an autonomous vehicle and to get a driver's license you're going to have to you know meet probably a very special test because it's going to increase risk and liability for everybody it's the same thing with AI if you're still letting humans do the entire process of of hiring in in five years in ten years knowing that there is unconscious bias knowing that and knowing that there's Affinity bias anchoring bias all these biases and and and and still not using AI which can be which can be tested for neutrality you're going to be asking for liability that's what's going to happen and you're going to be asked why are you not using AI when you know that there's unconscious bias in the workforce we would Craig for taking the time to chat with us today foreign [Applause] [Music]

Channel: Eightfold

Views: 79

Keywords: AI, DEI, bias, equal opportunity employment, HR, HR technology

Id: LSHL9cng080

Length: 29min 2sec (1742 seconds)

Published: Mon Jun 12 2023