Podcast: Making Artificial Intelligence Inclusive for Hiring and HR
Future of Work Podcast, Episode 18
Dan Nichols discusses how Candidit is using artificial intelligence to built a hiring platform based around inclusive talent competencies to reduce bias and better match employers to candidates with the specific skills they need.
This podcast is developed in partnership with Workology.com as part of PEAT's Future of Work series, which works to start conversations around how emerging workplace technology trends are impacting people with disabilities.
Intro: [00:00:01] Welcome to the workology podcast a podcast for the disruptive workplace leader. Join host Jessica Miller Merrill founder of workology dot.com as she sits down and gets to the bottom of trends tools and case studies for the business leader H.R. and recruiting professional who is tired of the status quo. Now here's Jessica with this episode of workology.
Jessica: [00:00:26] It seems like in the news we are bombarded with the fear of how technologies like artificial intelligence are eliminating jobs and putting us out of work. It's because of this I've been on a mission to learn more about technologies like A.I. and the different ways they they can be used to humanize not only hiring but also the workplace too. In 2018 only four out of 10 people with disabilities are reported to be employed as part of my partnership with Pete and Our Future of Work series. We are shining a spotlight on making the workplace jobs and employment more accessible. Today we're talking about how technologies like machine learning A.I. and block chain are making it easier for people with disabilities to connect with employers and for employers to hire people with disabilities. In my continued quest to find resources on the topic of Artificial Intelligence and how this tech is being used to create a more inclusive space I came upon a great company and a brilliant mind that I'll introduce you in just a minute. This podcast is sponsored by clear company. In this episode of the work ology podcast again it's part of our future FOX series powered by Pete. The partnership on employment and accessible technology. Today I'm joined by Daniel Nichols. He is the president and chief technology officer at candidate Daniel has an eclectic career as a military veteran and a lieutenant commander in the US Navy Reserve. He's also a former chief product officer at victory and the founder of STEM jobs. Daniel welcome to the work ology podcast.
Daniel: [00:02:00] Thank you Jessica. Glad to be here.
Jessica: [00:02:02] Let's talk a little bit more about your your background because it's a it's a varied one. Talk to us about that.
Daniel: [00:02:09] Well I've never fit a mold so all those people out there are probably feeling similar to me it's when you don't fit a mold you have to kind of reinvent yourself sometimes many different times and I actually had a time I was able to interview about 100 different executives and asked them sort of similar questions around this their background and one of the things I asked was you know if you could do it again could you repeat what you did and pretty much all of them said no they couldn't write we end up sort of where we're at almost accidentally. And so I guess what I've done in my career has followed interests and then picked up the skills that were necessary to sort of tackle the challenges that were along the way and I had some very large challenges presented to me like going to Iraq and assisting in the earthquake in Haiti. I think though the ones that interested me the most the ones that were about empowering individuals and that's kind of the sweet spot of where I've sort of lived and worked and grown and taken new doors and those have opened and made some other doors. And so here we are.
Jessica: [00:03:16] So what led you to starting this this new company. Like how did you how did you head down the path of of artificial intelligence.
Daniel: [00:03:26] So but interest in workforce development and what I've found throughout my career as a Navy chaplain U.S. Department labor health system working there with people is that the most powerful thing you can do for people is to connect them to to a career to a vocation that their love an environment in an environment where they can succeed. The Duchess dresses so many things sort of that economic peace in that sense of meaning especially in Western culture for people and that's how I viewed this this employment piece as being the key challenge that I've been interested in. And I've tried a lot of things along the way to sort of address the fundamental challenges in employment and the technology has not really really been there to do that on a mass scale until recently with cloud computing. And then the introduction of these sort of machine learning libraries which aren't necessarily like new technologies if you've been in data science for years you've been using a lot of the technologies or at least a lot of the algorithms that are necessary for A.I. but now we have the computing power that's able to handle the vast amount of data necessary to to really make powerful things with it. And that's that's sort of the difference that's available today. So done a few of these things pretty manually in the past and that's that's pretty slow.
Jessica: [00:04:46] It is slow and let's I just want to step back here and just remind folks that we're talking today about how to make artificial intelligence inclusive for hiring and human resources. There's been a lot of concern when it comes to A.I. and the potential for causing discrimination and inclusion. We've talked to Dr. Yu retrovirus director of the inclusive design research center about this topic in a previous podcast that will also link to in the transcript. She did raise some concerns that were really troubling about how A.I. pattern identification algorithms can eliminate candidates or employees who don't fit a common pattern. Can you talk us through about how your work is focused on changing that.
Daniel: [00:05:32] Sure. The like it to give a little bit of background or sort of environment with this. But fundamentally we lack like all of us together in workforce hiring etc. a common language for work without that common language there's nothing really linking the workforce together. That means the education system and the employment system people looking for work the whole thing isn't really connected. So when you don't have this connectivity you lack a fundamental thing which is trust. And that's kind of the one of the chief things we're trying to addressed and the result is predictable. You have to fall back to processes that are fundamentally biased human selection the interview your gut instinct. Do I like you or not. I mean that's really where things fall back when the complexity is such that humans. Is this the right human to do these kinds of different things and they're evolving over time. There's a pretty complex kind of thing. And in the industrialized world right. We've we've created the conditions that stamp out uniqueness. So this desire for work to be replicable and scalable and repeatable is translated to humans. When that happens we lose our identity and we lose our uniqueness because accounting for that incredible complexity is a very hard thing. So if A.I. is is leveraged in the same way that it has been then yes it's it's very powerful in screening people out but you frankly don't need it to screen people out. That's almost not using the power that we have at our hands and disposal with a eye for the right kind of purpose. I believe that A.I. can help us leverage the uniqueness the incredible complexity of human beings and there in our individuality. Unfortunately just most often wielded by the same hands that desire just the average out of people. And so that's you know that's not a direct answer of how we're using it but it's it's sort of framing kind of why and our approach to A.I. which I think is a different thing. If you're just trying to solve the same sorts of problems you're doing kind of manually and through your process today you're using A.I. to do it to screen people out then yeah you're probably gonna end up with a lot of bias that's already built into the system. If you can flip you're thinking about that too how do we invite more people lament and handle and deal with the complexity to find new ways to solve problems than A.I. is one of the chief tools that we have to really be able to do that.
Jessica: [00:08:01] Talk to me about how you've been able to develop a taxonomy and a language of employment even simple things like job descriptions and the skills required are so very different. And I'd also like for you to maybe talk about the consistency that your algorithm and the A.I. that you're creating and crafting and developing is designed to create consistency for employers and thus candidates.
Daniel: [00:08:24] So. Everybody's likely heard of it. All right. The Department of Labor's sort of taxonomy general taxonomy for occupations you've got like eight nine hundred different occupations within there you have knowledge and skills and activities and all these sorts of things are there if you take an occupation template though what you find is it's not very useful in the localised sort of state it will list every possible tool or technology that might potentially be used or sometimes it doesn't keep up it's not able to keep up with the sort of speed of new technologies that are out there. And so you're it's just way too general really is what it comes down to for a specific organisation to deal or handle with. So we we did start with that kind of framework and because you you've got to have something you can link things to that people can kind of agree on. That's got a lot of research behind it and extrapolated from there. What I call competency so are our main approach in creating this taxonomy is to design a set group of competencies so those things that people are good at that you can know that you can measure et cetera and looking at kind of competency based hiring and so we took these competencies and identified initially about thirty thousand plus. So that's a lot of competencies and each of those is dropped into four domains. So it's activity it's knowledge it's tools or it's technology that kind of fits the general Mark what we've done very differently though is to place each competency into this three dimensional vector. That's kind of where the math kind of gets a little bit crazy so if you if you walk it back and look at A.I. or machine learning right. But it requires a few things. One is it requires existing data so you don't have A.I. if you don't have a set of existing data that data must fit that be fit in a way that it can be compared to itself. So you have to have sort of standardization of that data which means you really need a taxonomy. The other is that A.I. is all math right. Specifically it is linear algebra which is essentially the math of relationships. So that's that that's what's being applied here is this math of relationships you know how compatible are you with X. It's a great question for A.I. and how I can kind of deal with that and I allows or allows to account for the uniqueness and individually at massive scale because it can deal with these massive numbers of things. So what we look at is you break a job down into a group or set of competencies an individual is a group or set of competencies things that they know do you know et cetera. Well and of course education is a group of competencies they're going to teach you how to know do know or do well but it's the sort of secret sauce of what we've done is is the math right because most efforts to date look at job title and the job competencies that you do or things that you do as just strings of text. And so how do you compare the meaning of a word to the meaning of another word. There are ways of doing that and that's most natural language processing is one way of doing that but it doesn't tell you anything about what that thing does. Right. So digging a ditch what is the you know is probably similar to using a pole to pole digger to dig it to dig a hole right. So those are probably similar kinds of activities but what does that even mean. And that's that's where we create these sort of domain vectors. So we look at every competency as a how general and versus how specialized Is that how process oriented is that versus how creative is it how physically active is it versus how cognitive it is. And so we're able to build something that these machine learning models can understand and could do can perform math functions on. And so that's that's really consistency so everything we're doing that is trying to help employers translate the way that they're doing things now into this consistent model that is much better suited for the mathematical computations in A.I. than existing sort of processes are because those processes have to take you know normal language novel language and convert that into math whereas ours is already done.
Jessica: [00:12:42] What about benefits for the job seeker. One of the things I like about what you're doing with your technology is that it provides value to the individual and not just the employer. Talk to us about how you're helping develop job seekers specifically people with disabilities.
Daniel: [00:12:58] One of the most terrible horrific things you have in life especially your career side is really finding a job. That process of knowing if you qualify not knowing if you qualify has its landed students and trillion dollars worth of debt frankly. So it's a terrible process because you don't when you apply for a job you may never hear back in if you do hear back you don't know if you actually qualify. You think you qualify and you don't have any. Guidance and so that's a key thing that we supply as we match it to every possible job is in the position that's in the system and we'll give you feedback on how well you match to it what your what competencies you're missing and how to find know how to fill those gaps. You may have an competencies and so it gives you a clear path from not only where you are to the current job that you might be looking at but to any potential position that we have in the system so that you'll know if you qualify or if you don't qualify. The other piece that we add because we have this sort of three dimensional matrix for every competency and it's not just an English text of what that is and finding equivalence and other text phrases we're able to see that you might have done something and how close you are to other types of things or how some of those similar those things are. So for instance with military personnel that vary if you try to match them to a sales position you'll get a zero percent match pretty much every time because they don't functionally do anything that is sales specifically sales they don't sell anything in the military but they do things like in the recruitment commands that are fairly similar to what you would do skill wise if you were in sales and real. This system is able to find those things into to bring those things out. And I think that's a very powerful thing for the user side is to not present them with closed doors but present them with all the varying options that they have and you need to have sort of robust computing power and capabilities to be able to do those kinds of things and that's important. The reason to do that you have kind of this forcing to the middle is risk. So from the employer standpoint everyone that you don't hire which is a lot more than you end up hiring is a potential risk to the organization. And so to minimize your goal really is to minimize the number of applicants deemed as qualified. And when you do that you almost always push out the outliers and that pushes out great people but it pushes out people with disabilities almost first because they are most definitely outliers in the system and that's why this mapping these pieces and then hoping that for the candidates they're looking for opportunities is just as critical as showing sort of matches for the employers on their site.
Break: [00:15:50] Let's take a reset. This is Jessica Miller Merrill. And you were listening to the work also podcast today we're talking with Daniel Nichols about artificial intelligence and inclusive hiring. This podcast is sponsored by clear company and is also part of our future work series in partnership with Pete. The partnership unemployment and accessible technology.
Break: [00:16:10] This episode has been sponsored by clear company complete talent management software provider clear company software solutions include award winning Applicant Tracking onboarding and performance management solutions. Power retain and engage more top talent with clear company.
Jessica: [00:16:26] Let's talk a little bit about your technology. It's called candidate and walk us through maybe some of the basics around everything not everything but a lot of what it does and how it's being used with some of the clients and organizations and people that you're working with.
Daniel: [00:16:42] Let me start with what we're trying to solve because we people on the same page there is usually easier to have those discussions so the primary challenge we're trying to solve is that of trust employers. Right. Looking at candidates that are out there for positions fundamentally don't trust them and probably shouldn't. Because thing was like 86 percent of employers found falsification and frankly lies on a resume made after they hired people candidates looking for job don't trust employers right. I guess if I go back to that that other one employers then run extensive background checks and all this sort of testing to figure out if someone is telling the truth about what they're doing. Candidates all trust the employers America's good employers are asking for the unicorn in their job description like people that don't exist or that have I love the ones said you have to have X amount of experience and X education while people get the education and then have the world they get experience if you have to have the education to get into the jobs or require that would give you the experience. So it's these sort of crazy things of this complete lack of lack of trust and that's that's I think fundamentally we're trying to do so there's two primary technologies that we that we utilize. The first is to account for the uniqueness of of individuals the second is to rely on verified information from trusted sources. So you know the first of those let's say let's look at that individualization of thing. We're utilizing machine learning and this competency framework to account for the bring out the individual nature of people if you recall the Super Bowl ad as well.
Daniel: [00:18:21] So the company did the big ad but you saw the kind of screen on there they're talking about military hiring go back to that going to do a lot of done a lot of work of that in the past. And so you see this eleven b right eleven Bravo and then it connects to the jobs at this eleven b would be fit for and that's a really cool kind of thing. A lot of military ammo so that eleven Bravo it means infantry that say what we call a military occupation. And so these translators are out there. But the problem I immediately saw in that is that not every eleven Bravo is the same as every other eleven Bravo in fact that is a terrible way to approach that whole thing. Every single infantry member of the infantry right. They go through different things they have different experiences there are different commands they learn different systems they add their own education during that time they're in the military the same can be said for individuals and majors and colleges right you have two biology majors even from the same school probably took different electives they did didn't you know perform differently on different portions of what they learned and we're unique we're unique individuals. So how do you get to the uniqueness of it when you know currently we go into higher we look at a degree is almost being binary we look at your military entirety of your military experience is almost being binary the power of of technology can let us get beyond that kind of thing and that's that's a key piece of what we bring to the table right is helping me helping to simplify the view and the understanding of the uniqueness that an individual brings to the table the other is that is the block chain side of things so we start on the block chain path and where you're trying to do it there is the verified information from trusted sources right. So who is verifying this individual's actually done and perform these particular types of skills and most employers accept the word of a school or an entity or another employer or institution much more so than they trust the individual. And so the individual you've got a lot of systems out there that are all about self attestation write your resume a is what you're saying that you can do by yourself and then somebody is going to have to prove that that's possible. The what our system looks at doing is trying to bring multiple entities together that might normally be competing with each other to help to verify and to provide that sort of verification of information. That's where block. Change comes into play it's that we can talk a bit more about what kind of block chain changes but it creates almost a superior safety net and brings a complete visibility to the supply of talent and the demand for talent so that there's there's a lot more clarity in the workforce system. And so those two primary technologies we're bringing together into candidate so that if an individual employer is bringing on individuals and screening them but not hiring that candidate then drops back into the total overall we call talent exchange or candidate exchange. So another employer that's hiring now has access to that individual. And so the effort for recruiting becomes much less and much less expensive as everyone shares with their recruiting efforts to build this pool into something that's that's a much more validated verified pool that you can pull from.
Jessica: [00:21:47] I like that you guys are doing this because when I think of what really the number one hiring source is and it's a standby for the entire time I've been in H.R. is referrals. And so you guys are using block chain and machine learning technology to offer employers another way.
Daniel: [00:22:07] Yes. Yeah. You have referrals or that. That's an interesting piece because it's a big barrier for individuals disabilities because they may not be. If you are connected to that network or referrals they get something like 7 percent of applicants come from referrals but 40 43 to 50 some percent of referrals or end up being hired. That's a gigantic imbalance for people that are on the fringe or that are outliers and not connected to those already employed. So yes having another way would be great because the referrals all about trust. Right I trust you. You've done a good job. If you bring somebody to I can be more likely to trust them. I would. Well that's got nothing to do with objective. You know can you do the job or not. That's got everything to do with this issue of trust.
Jessica: [00:22:56] Yeah. And typically with referrals and one of the I guess challenges for referrals is that it's also a good thing it's that trust piece. But it also brings the same type of people to the organization. So there's not different types of people. There's not different thoughts different experiences different backgrounds. It's not an inclusive recruitment recruiting strategy.
Daniel: [00:23:20] Yes absolutely. Absolutely.
Jessica: [00:23:22] Let's talk a little bit more about the machine learning block chain and the A.I. that you guys are using to really challenge the status quo of what I consider and I think many do a dysfunctional recruiting and employment marketplace. Can you maybe talk a little bit more about how these two worked together.
Daniel: [00:23:41] Sure.So let's let's start with with BLOCK Jane because that's a little bit tougher to understand because there's just so much information is kind of out there really what we're trying to do is to get employers to work collaboratively together in understanding and shaping and also collaborating typically with the government and with colleges and universities right so that we can bring visibility to to this whole talent supply chain right. We can understand what demand is we understand then if schools know what demand is they can prepare a curriculum that would then fulfill demand and we can even be predictive about that sort of thing if employers know what the town's supply is right that they can predict what their hiring potential might be or know that they need to start helping to invest in the whole workforce system thing.
Daniel: [00:24:33] So the block chain side of it is really about there that technology helps to automate economies so that kind of the scale of things is to say it's the right solution where you're trying to coordinate activities between multiple often competing entities and that's the big challenge you've got with this world of recruiting is that employers are the greatest access most expensive asset most important asset are people. And so there's high competition between organizations around the people and what employers hate to do is to train people that are only going to leave and go work for somebody else. Right. So there's there's a lot of sort of turf ism in there you can find employers that are in completely different industries that might work together but you really have to get it. Especially today with rising technology and the massive rapid shift of need for people to gain new skills advance and transition in skills there's a need for greater collaboration and block chain is one of the few things that does that because it creates it gives us the ability to manage the privacy of individual information to manage the safety and the security of this massive amount of data that's out there that isn't used in the wrong kind of way because you can set kind of contracts in place and you can I guess establish this sort of economic agreements between organizations that this is how it's going to work and that it becomes very visible very clear that this is how it's working in today's world it's just literal chaos. The systems that you have in place are really built around the idea that it should be really inefficient because it can make more money off of inefficiency of you know people taking having to apply to many many many different organizations and or organizations having to recruit over and over and over and over again for the same kind of people and finding the same people that inefficiency is how the system works and how it makes money today and I think there's entirely different models that would work way better for for everybody and could still be monetized right by those organizations looking to do that.
Daniel: [00:26:45] But it does require a technology capable of establishing rules and working off of rules that I think gets employers that might compete with each other more comfortable about working together the ways that we've been able to find at least initially is around the issues of inclusion around certain populations that are groups of people that employers can kind of agree to work with and so that would work around or work together for on behalf of veterans or one of those individuals disabilities would be another of those populations. And so I think the first forays of real advancements in utilizing this technology nature our world around the diversity inclusion initiatives that are out there that are most open to taking the steps necessary to kind of recreate this world how it works.
[00:27:35] So that was just one area to talk about the machine learning side of things but that that collaboration I think is is really key to this working long term I think it's sort of a future where things are and will go. And the question is who you want to be in charge of that framing that. I think multiple public open source kind of approaches is the best way to do that.
Jessica: [00:27:59] You've said to me kind of when we were having a conversation that I and these technologies bring humanity to hiring. Can you talk to me a little bit about what you mean. Because I think so many people think the opposite when they think of A.I. and automation and machine learning.
Daniel: [00:28:19] So what are the human factors right. What is what is humanity kind of mean to me that's that uniqueness independence liberty community. All right these are these are human factors making those factors work in harmony in the workplace is incredibly hard and incredibly complex. How do we take advantage of the uniqueness of individuals and and offer them a certain level of independence but also make that work in community right. These are those are tough things that make the mistake I think we've had in the past is to try to force everyone into a specific mold again industrialization is framed around this idea that needs to be replicable scalable repeatable. Right. And so everything moves to the center and we create something that we can repeat and and replicate over it over and over again. And that's that's what we need for this business to be most successful that that kind of thinking and industrialization led to the atrocities that ultimately led to World War 2. And if you apply that kind of thinking with a. You're going to end up I believe with the same kinds of your potentially that this that kind of atrocities that people talk about when they think about A.I.And unfortunately many people see A.I. being used to kind of enforce conformity. Right. Which which frankly wastes the computational power of A.I.. Because you don't need A.I. to force conformity. We already do that pretty pretty well. What I can do for us is help take advantage for the first time of the limitless complexity of each individual. It can help us to capitalize on the power of the outlier to solve problems in ways that the conformed masses never will or cannot. And so I it's exciting to me right that what we how we should be using and thinking about A.I. is is how do we use A.I. to tap into the massive complexity of the individual and deleverage that complexity to be able to solve problems in that unique kind of way by letting people be unique when they approach problems due to a different paradigm for sure. And it's one that is not really possible without more advanced computational power but it is something that becomes far more possible in the future so I see not a world that is more governed and more constrained. I see a world that is more open and more more free.
Jessica: [00:30:57] Well thank you for that because I think that I hear like I was at a class master's class on culture in the workplace. It's a group of leaders who are going through a master's program here in Austin. And one of the questions was artificial intelligence is that the end of all these jobs and I was like Well no and so we talked about you and a lot of the work that you and candidate are doing so I appreciate I appreciate that insight.
Daniel: [00:31:24] I'm sure many people will think I am just horribly naive but when you've built some of it you see some of the power of it. We're not just talking about the concept of A.I. right but we're actually talking about machine learning and the tools and how it can be used the data sets and how you actually build the stuff that that opens up your world to the potential for it. And I hope that more people can kind of grab onto that it's not you know rose colored. I don't I know that there are dangers absolutely for sure and how A.I. can be used poorly. And and we've talked with the Department of Labor. I think that's where the regulatory side of things becomes very important in this to to build in safeguards or to build ways to identify when it's being used appropriately. But you don't just throw out this kind of technology that can enable people to to really be their unique self. So again when you look at diversity inclusion I see that group of individuals as being ones that ought to know dip their toe a little more deeply into some of these new technologies because it will help us to be more inclusive and to be more diverse in how we approach things and I think that ends up in better results better outcomes for people overall.
Jessica: [00:32:38] Some Well Dan thank you so much for taking the time to talk with us. Where can people go to learn more about you and what you're working on.
Daniel: [00:32:47] So I think you'll share the LinkedIn profile I do have some things I end up writing I've actually been building a lot recently so I haven't written as much as I might like to. But some of the blogs and bots are on their red cell talent. Dot com is sort of the the parent corporation the candidate that's the product. So some of those blogs and things show up there. I'm working on some other other things you know later in the future maybe even a little book that we're kind of passing out now to get some of these ideas a little bit out there but yeah thanks for sharing. Let me come on. And I'm certainly happy to share ideas and our progress as we go.
Jessica: [00:33:23] Well thank you and we'll link to the blog and your LinkedIn and maybe a couple resources that you that you have for far listeners so check out work ology dot.com to listen to the podcast. But also the transcripts and resources.
Daniel: [00:33:40] Thanks Jessica. Keep up the great work. And thank you for having these good dialogues.
Jessica: [00:33:43] I love the work that Daniel is doing and how he and candidate is helping to create opportunities for job seekers to get feedback on how they can improve their skills abilities and experiences to find work. I also think it's great how they are thinking about how to leverage the different technology to create new opportunities to put talent in front of employers particularly people with disabilities.
Jessica: [00:34:07] I think it's important for us as H.R. leaders to take time to understand and be exposed to how technologies like block chain machine learning and A.I. are being used and developed in our space. We will include some great links and resources in the transcript of this podcast. I want to thank Pete again for the partnership on this feature book series as well as our sponsor clear company. Thank you two for taking the time to listen to the work ology podcast.
Closing: [00:34:34] The workology podcast Future of Work series is supported by Pete the partnership on employment and accessible technology Pete's initiative is to foster collaboration and action around accessible technology in the workplace. Peter's funded by the U.S. Department of Labor's Office of Disability Employment Policy. O DEP. Learn more about Pete and PEAT works dot org. That's PEAT t w o our K S dot org. Production services for the workology podcast with Jessica Miller Merrill provided by total picture dot.com.