Adina Sterling: How will artificial intelligence change hiring?
Ladies. And gentlemen, welcome. To the future of everything. Live. This. Show is being video and audio recorded. Please. Take a minute to silence all cell phones, and. Now. Please. Welcome bioengineering, professor, and the host of tonight's show, Russ. Altman. Thank. You thank, you testing, one two three sounds. Like you can hear me welcome. To the future of everything I am Russ Altman a few, little, notes before, we, get started we're very excited about two great guests we're really excited to have you all here today you know we don't we usually do this in a dark studio. Cave, and so, every now and then we come out of our cave and we do this with you and it's super fun so thank you for coming I know that it's especially on a rainy day this is an amazing, turnout. So, I just want to you heard the voice of God say, that. This, is being taped both on. Audio, and I believe video and it. Will be aired as a normal, show through, the magic. And, of. Editing. So. A couple, of things this is a radio show on Sirius XM a, Channel, 121, so, if you have a new car and you still have your Sirius XM or if you're a sirius, XM subscriber, I'll. See you on Saturday mornings, on the, 121, channel we also make it available on iTunes, as a podcast. With. With several other shows that are part of this the growing, Stanford, selection, of of. Radio. Content. And then Stanford. Radio dot, stanford.edu. Stanford. Radio all one word dot, stanford.edu, also. Has some, sound files that you can listen to it so thank you for your support, previously. Or thank, you for your support going, forward I want, to take a moment just to thank my producer, Brian, Pelletier. Who makes. Me sound good by editing out all my bad things, and I also want to thank Michael Friedman and Catherine McMillan, who helped, get the production under, order there are many others as well so. I think. To. Tell you we're, gonna have two guests, in one, guest we're gonna have a first, few minutes of conversation, and, then a short break not, enough to get out sorry. And. Then we're gonna finish that conversation have. Another short break we'll, bring on the second guest have one continuous, 23. Minute conversation and, then we'll stop we'll have a very brief break while you, come to these mics where, we'll have 10 to 15 or so minutes of Q&A, so do think about if you have questions, we, like short questions, so. That we can get a lot of questions in so. I think that's, everything that I had to say and I want to again thank you and I think we're gonna start the show now ah. One. Other thing I have to say which, is that I do use, notes so I can't. Memorize everything, so I want to apologize in advance you will see me referring, to notes as I give my introductions, and stuff and it's just it's part of what I need to do so I apologize, if your if that's disappointing, to you, having. Said that I will ask my producer if we're ready to go. Today. On the future of everything the. Future of hiring, finding. A job so. Looking, for a new job is exciting, but, petrifying. And. Some people do it a lot and some people do it every now and then it's, exciting because you can have new opportunities you. Meet new people personal. Growth a change, of environment. Often. A good thing you, can learn new skills it's. Also petrifying, its petrifying because you might not find a good job or a good match you, might have to move to, get to the kind of job that you want physically, move change where you live you're. Worried about the match of the culture, of the company to, your value system what you want to do you, want to you're, worried about your match to the other employees how. Is that going to go am I going to be compatible with them do are they going to be friends or or, frenemies, or foes, and. Is the mission of the company consistent. With the things that I value are, they doing more good than evil and, it's, really the same for the employer right the employer is trying to find just the right people to, bring together for, this common, mission rowing.
In The same direction. Singing. For the same choir booked all of those sayings. They. Need to find talent, and aptitude, they need to find communication, skills commitment. To the mission of the company. They. Want diversity they want people bringing in new perspectives. They. Want to make, sure they can pay them and afford their salary, that they're competitive, in their benefits and things like that and they want them to stay a while once, they're trained and functional. So. The way that people find, jobs and, the way that people post jobs has, changed, in many of our lives especially. Those of us who are a little older a lot, so, in, the, olden days we. Had. Newspaper. Ads. Magazine. Ads there. Might be radio, or television call, us up come to visit, us we're hiring and then, we had things like Craigslist. We, had monster.com. We, have monster.com. We. Have LinkedIn, as a way my. Children were told that their LinkedIn resume, was at least as important, as their printed resume when. They were in college preparing. For the job market Facebook. Instagram, whatsapp. I don't know if you find jobs on Instagram but we will find out if whatsapp. Or Instagram gets you jobs then, there's headhunters, and, in personal references, I work, at this company you might enjoy working at this company too, a benefit. Of the internet and all of these organized, sites is that, we can study kind, of they generate, data that we can study how, good of the match has been who's, working, who's not working how. Can we make this very, sometimes. Awkward matchmaking. Process. As as, good as possible, and. What. Are the policies, that we should implement at our company to, have the best most effective hiring, practices, and make, this whole process more. Exciting, and less petrifying, with. That I'd like to invite Adina sterling up to the to, the stage. Thank. You Adena. Thank. You. Adena. Sterling, is a professor, of organizational behavior, at, the Graduate School of Business here at Stanford, University, she, studies labor markets, social. Networks and hiring, practices. Idina. Hi. Hi. Given. All the new ways that we have to identify large. Pools of candidates, our company's. Better at choosing and recruiting. Employees than, they were 20 years ago or has. This not gotten, better. That's. A, complicated. Question. Perfect. So. In. Some ways. Companies. Have gotten better because they've become more, efficient, they. Have gotten better at, targeting. Particular candidates. That, would, be overall, a better fit in the organization. But, on the other hand they, also, have to deal with, the. Large. Influx, of. Candidates. Applications. That they're getting, and. So it's. I would, say as possible. Today as it was 20 years ago that they would overlook, a diamond. In the rough someone, that doesn't. Appear to have the skill sets on the surface, that they want, but.
Would Be fantastic, for the, organization. Yes. So going, back to this and the changes that have occurred you know I gave this quick little summary based on not really being an expert has. There been a, history. Of hiring and do. HR people think about it very differently and have the policies, that the company's evolved, and, what do they do about this firehose, of applicants. In terms of process, yeah yes the history that you gave. I think is a really important. History, so 20. Years ago 30, years ago in 1994. Monster.com. Started. And that revolutionized. The way that hiring, got. Done, companies. Up to that point might have 20 applicants, for a job, once. Monster came on the scene they could get 200 applicants, in the number of minutes and so. Overall. What they've become is, more technically. Savvy and, how, they, go. Through resumes, right, now. Around. Three quarters of all resumes, and all applications. Will, be touched by some form of AI they. Will be screened out early, in the process and. So that's that's great that's one way that some of the efficiencies, have come about but. When you've got 10x. The number, of applications, overall you're. Still getting a lot, of of, applications. So if AI is one of the first filters, that means if I have a touching, human, story about. Why I might be a very compelling candidate, that's totally, out of the blue I served. In the military and. You know was a very you, know an, inspiring, platoon, leader and, had showed leadership, skills, that are ridiculous yeah, that might not make it through an AI filter, that's right that's right you might not have the buzzwords as a part of that story that. Would allow your resume, to get the attention of a. Hiring manager so. Does that lead to systemic, problems, and systemic, misses so to speak I would. Guess it does on the hiring front, and do, they do the people who kind, of make these sins of omission, do they ever feel. It or know about it or is it kind of a blind a mistake that they're never really aware of because they never see the application, yeah so that's a another.
Good Question it has to do it kind of type 1 or type 2 errors, and and frankly. They. Often don't know when, there's a missed opportunity. Somebody, that would be just absolutely stellar in the company, but perhaps because they. Come from a, school that isn't, one of the typical colleges. Or universities that. They recruit from and particularly, maybe, they don't have a social, connection in the organization. That allows them to to, get the attention in the hiring process that, they need they, could certainly be be, overlooked, so I know that you I've actually have studied, this phenomenon, of the network, the. Network effects, of applying for jobs and whether the kind of network that an employee has and whether that helps or hurts them enter a new organization, and of, course on the positive side it would be like well if you have a good employee and they know another person who they think's good. Sounds like a good thing but, you found that there were also some downsides, to this so tell me about the balance, I mean the, network, that the employee comes in with or lives in and how that might impact the hiring processes, right so. Social. Networks in hiring are, everywhere. And it's. Hard to get a good estimate of just, how, pervasive, they, are. Low. Estimates. Are around, you. Know around 50, percent of jobs will be found through social networks but. On the, higher end of things. Some. People, have estimated that 75, to 80 percent of jobs are found through. Social networks so if I could just ask when you say through a social network is that the informal, network of friends or does that include things like LinkedIn yeah and these kind of reification. Z' of your network yeah yeah so it's both it's both so it's friends, family former. Co-workers alums. Of a school, that's. Part of your social network and it's. The reification of that, social network on LinkedIn if, you will yes and what, a lot of companies do now is have the API to be, able to say if you, give us access to your LinkedIn, profile, we'd, love that because, we want to know who you know in our company, and guess. What if you don't know someone in the company you're, kind of out of luck you don't raise to the top of the pile because. HR. Managers love when. You have a referral, into the organization, now, is this without question, always a benefit to the company or is, there a downside to this network effect of hiring yeah. So. It's, a it's a great question and it's what a lot of my research, is on and. What. One of the things that we know is, that there are benefits, people. Who are hired through social networks through thin friends, family coworkers previous. Co-workers, they are more likely to stay in organizations. Okay so turnover, tends to be less there. Is mixed, evidence that. Their, performance in, the organization. Is higher, in fact, and some of my work I'm finding that it's, not really. That's not really the case. Then on the flipside, not. Only could it not be. Confederates. To suggest that in fact performance. Could be lower if you're hired through a social network for some of the reasons, that we might think so, if I know. You and, I've applied for a job and you're sitting, in a company like Facebook you, might say oh it Dean is really, great just hire on the spot and lo, and behold that, hiring, process, then gets truncated right, iczer convinced, the quality. Checks, that's right that's right huh. And so, I think you've also looked, at so. Sometimes, people, apply. For jobs to kind of test the waters or even get competing, offers, that they can use as leverage in their other negotiations that. Knows I think you've connected that to network, social effects as well that's, right that's right so one. Of the things that we can think about on. The positive, side of things when it comes to social networks and hiring is that. When. Someone an applicant, knows an employee. And the organization. They. Can. Act non rationally. And what do I mean by that I mean that. In, one study MBA. And. Law students, I found that even, though MBA. Students, and law students, are highly. Strategic very. Smart, and, you. Know that they will benefit, from having multiple offers.
From. Companies that. They're willing to give those offers up early, if they, have a friend in the organization, that's offered them a job that. They don't intend, to keep in other words I'm willing to give, up my negotiating, power, with another, company if I, have a friend and then organize that just based on the kind of the very human I don't want, to kind of cause, trouble, for my friend by acting a little a friend to take a reputational. Hit because. Likely. They've endorsed, me and so, I don't want to, prohibit. That company, that my friend works for from moving on to other candidates, now. Of course when we think about, the. Use of AI and, as this has been in the news a lot there's, the possibility of these, AI algorithms. You. Implicitly. Having, biases, against. Underrepresented groups, women, minorities. Any group, because. They typically are trained. On data, most of these AI algorithms, you give them a set of successful. Hires or something like that you say please find people like that and if, there are. If. There are correlations. In, that data that you may not want to perpetuate they. Could create. Unfair. Situations, as, this come up in the work and are we seeing this happening, in these AI programs, yeah yeah so the. Way that I would describe it is that. That. I've been pleasantly. Surprised. And, encouraged, that, a lot. Of companies I think are moving, with a lot of caution in this area they looked they, looked at what happened with Amazon. Where. Over time they discovered lo and behold with, the AI and machine learning. Processes. That they were using that they were screening, out underrepresented. Minorities. Women and that they were hiring, all white, guys and, they. Said okay so we're, not going to continue. With that with, that algorithm. Anymore. And so. Whether. What I think is that, a lot of companies at least in. Interviews. With me are. Telling. Me that they are not willing right now because. They know that bad, data end means, bad data out and bad processes, out they're not willing to kind of let the machines, go. On their own that at a minimum they, understand, that a supervised, machine, learning process. Is necessary and, that. They've got to move with a lot of caution when it comes to to, AI and machine learning for this very reason so that's good to hear although when, I think about the very first thing you said which is that the first screen is always some sort of AI you. Could imagine that it's actually down in the trenches, and the weeds it might be a little complicated and, it. Might be difficult for them to fully implement. Totally. Unbiased, a screening. If they're depending on it to get from 200 to 20 yes yes and I'm, sure at some point those discussions, might become awkward with you when, you push the I don't. Know. Either. They will. Frequently lie. Or I thought I don't want to say that. Here's. Here's what I would say I think that that. There's been a lot of attention to these issues and, we've. Had great. Great. Books like Kathy O'Neill's weapons. Of mass destruction, books and things like that that there's a. Good. Deal. Of kind, of knowledge and caution, in, the industry, about these things and so, I'm hopeful, that the, implementation as, we, continue to learn more will be done with a lot of care so let me just ask is there a way that some of these AI technologies, technologies, might be good is there a way that we can turn them around yeah and have them be a source of diversity, and open-minded. Hiring perhaps, even more than, they would naturally do if it was an entirely human yeah so. One of one of the, real. Areas, of growth in AI is, with, sourcing, and a. Number of recruitment. Tech companies, have come on the scene in the last couple of years that have explicitly.
Focused, On, targeting. Underrepresented. Minorities. Women and, candidates. That companies. Wouldn't otherwise have. Exposure to and they, do that through sophisticated. Targeting. Methods that have to do with understanding. Frankly. The massive, amounts of data that, people are providing. On the internet. About their lives and so they. You. Know the. Algorithms, know. What they blow right with a fair amount of accuracy. Race, gender sexuality, and some some, other aspects. This, is the future of everything I'm Russ Allman I'm speaking, and we, will have more with Adina sterling, about hiring bias, AI. Next. On Sirius XM insight. 121. Thank. You end of section. So. We're gonna go there, next. Segment okay thank. You for your patience we're just gonna go right into the second segment I do. Want. To check my little list of things to ask you absolutely. Yeah. Good. I'm. Ready Brian. Welcome. Back to the future of everything I'm Russ Allman I'm speaking with Adina sterling, about hiring AI, bias. And fairness, so, in the last segment we talked a little bit about, these AI algorithms, and how they could, be bad they, could be good another thing that you've studied is gender. Pay gaps yeah and that's on a lot of people's mind and probably at least half the people in this audience so. Let me ask you what, have we learned about gender pay gap and is things getting better or worse. So. One. Of the things that we know is that it's real, and, even. In very sophisticated. Models. Where we have lots. Of information about. People's. Test scores the schools they went to. Other. Aspects. Of their human capital we still see based. On all of those observables. That women, on average are paid less than men now. The the. Statistic. That tends to get thrown out there about an 80, percent women making 80 percent of what men make, does. Not take into account the, fact that men, and women do different jobs, by. The time you look within a job, that. Wage gap significantly. Narrows but it is still there, and it, varies but anywhere from five to ten percent. To. Be still, enough to be concerned that's right so. Are. There strategies, because, a lot of this has to do with the initial salary right when you get when you start and then you get on a curve and that curve is just very hard to move the time so are there strategies that. You've, studied or that others have studied to, continue. To narrow that gap and, what do we need to do yeah yeah so, your. Question I think is spot on in the sense that when. Women start out making less than men, then the cumulative disadvantage. Process, gets put into place and so, one of the things that can. Happen from the start is that, employers, get better information about, people, in other words let, me not rely on my. Stereotypes. Or beliefs about what a leader. In a fortune 500 company ought, to look like let, me bring people into the organization, spend. Time looking. At how they perform, on a team, how, they lead within the organization, and then, mate let me make salary assessments. And that's, one of the things I did in the study where, we had a, on. Average, eight to ten percent gender. Wage gap but. After an employer, brought, people into the organization got. To see what they were really good at that. Gender wage gap completely. Reversed. So this is the I saw. Some of your papers this is the literature on tryouts, or it may be an internship maybe, that's not the right word so how do you logistically, do that though so how do you as an employer let's say you're an employer you have the best of intentions you, read, all of your scholarly, work and you say yes I need to bring these people in and give them a test is there, ways to set that up to make it practicable.
In Terms of you. Would need a very high rate of good workers because if you're constantly saying oh this is not a good one this is not a good one that could end very quickly so how do you actually how. Do you do that scale well, I I think that we're not quite there from, a technology standpoint, but we're getting close so, you can imagine that instead, of bringing people in to an organization, for three months you, instead have them do some sort of simulation, or. VR where. They. Can spend two or three days, operating. As they would within the organization, as. A shortcut, if you will so that you can see them working in, real, time versus. Trying to make an assessment about how they how good they would be based on observable. Characteristics, and so is there emerging. Evidence that even those virtual, reality, or very short, experiences, can still give. Enough information to the employer, to make that better decision, because that would be great that means you, can really compress it to something that's very doable yes, so so, the answer is yes and. So. Yes there's evidence, that it helps, as long, as employers, don't, shoot themselves in the foot and what I mean by that is frequently, they, will couple those. Tryout. Experiences. With, more traditional hiring practices, and then still fall into some of the same sense if you will where, they're expecting. Say. People. From kind, of a certain class to behave like another class, and. When you are poor underprivileged. Etc, you're not walking into a room commanding. Attention then quite the same ways and so when employers are still looking for that that means socioeconomic background, can still matter in ways that perhaps it ought not this. Is the future of everything I'm Russ Altman I'm speaking with the Dean of Sterling about hiring fairness. Bias, and. On, this issue so, I wanted to ask about. What. Did I want to ask about. Whether. There's a tipping point so well let me step back you, you mentioned that at the end of your last comment that there. Might be expectations. About the coming the person coming into the job and I think if I remember looking. At your resume you, have written about the, star. Person, who comes in where they here this guy is good, or this woman is good and that, that sets a whole stage, for, success or failure, that's very different from somebody who doesn't come in with that expectation so. What. In the world can people do about that kind, of very, human, oh I hear this one is a real great one or I don't know about this one how, do we manage that is that manageable, yeah. Well. Some, of it I think starts, with awareness so. The very fact that that. You're even asking that question I'll tell you that it's not something that crosses most HR managers, mind and. Managers. Minds and so, we, do know that. That. There's this expectation when, you're good that you're gonna that, you're going to get, better and you're going to get the lion's share of the resources, one. Of the things you can do is, place. The, people without say, as. Much. Attention early, on in the organization, with a really good supervisor, somebody, that knows, how to mentor and train them and bring, out the best in them, maybe that. Manager, is high. Status, in the organization. And so that, is a that then, helps. Shine a light on that on that employee so you just kind of have to be aware of it and proactively. Try. To be fair about expectations. And so, to. Kind of to end up I wanted to talk about this new gig economy, right the 1099, economy, is not going to be a source of more fairness because it's where contractors.
Like An uber driver you, know I don't believe the uber al the uber algorithm, takes, any. Consideration. Of whether the driver is a man or a woman does is this a great leveler, or not hmm. It. Depends, I, think. It depends there there's, been a number of studies. Done. On platforms. Like Airbnb, of. Course that show, that we're far from leveling, the playing field, in, on. Those sorts of platforms when people, are. Interacting and, kind of these more, private, or intimate, spaces that. Gender. Stereotypes. Racial stereotypes, stereotypes about. Class they'll come to the forefront and, that, there's still a lot of discrimination on, those on those platforms and, so I think again it goes back to a careful, design of the technology. But. Then coupling, that with a set of policies, and programs that, sometimes. Say we need to educate our, consumers. And our, content, providers, or whatever. The case might be on. Ways. To be fair so, unfortunately. The same biases, that we can see in normal, hiring, practices, can just play themselves out in the gig economy where, you're just picking a contractor, for a one-off relationship. But, even in that one decision you, can bring in a whole world of, perhaps. Non optimal decision making technologies. That's right well. Thank, you for listening to the future of everything I'm Russ Altman coming. Up we'll talk with former Facebook executive Alex, Thomas on Sirius. XM insight. 121. Thank. You. You.