How AI Evolved the Hiring Process

PODCAST TRANSCRIPT

Speaker 1 (0s): You are listening to TribePod a podcast series of interviews of interests to the HR community. It is hosted by Jim Stroud, sponsored by Proactive Talent and enjoyed by you. Today's episode begins right after this.

Speaker 2 (42s): Let's face it we're in a whole new world. Now We know that the reactive old way of hiring in the post and pray model is expensive and it's getting more expensive every years. What employer brand does is it is a long term strategy that will help you get better at hiring faster. And at a higher quality.

Speaker 3 (1m 1s): 75% of candidates will research a company before even applying. And 86% of candidates will not work for a company that has a bad or non existent employer brand. Some of the many benefits of having an effective and strong employer brand include doubling the amount of applicants you get per job posts, decreasing your cost per hire by 40% improving employee retention by 60% and overall just yield better Glassdoor reviews.

Speaker 2 (1m 29s): We know that companies with stronger employer brands spend about 10% less overall for talent,

Speaker 4 (1m 35s): Proactive Talent helps out clients with their Employer Brand like going in and working with them in several phases to learn more about the culture, the people, what are the important values to each and every employee. And then to share that story and refresh the Employer Brand, or build it from the ground. The benefits of having an effective Employer Brand is that you're going to be able to attract the talent that you really want to join your company and not just people who would be simply applying for whatever requisitions you have out there. They generally love your message, love your culture, and are there to be busy for the long haul.

Speaker 1 (2m 10s): For more information on Proactive Talent, visit us online at Proactive Talent dot com or click the link in the podcast description. Hello, dear listener. And depending on when you're listening to this podcast, good morning or good afternoon or good evening, welcome to another exciting episode of TribePod with me today is a very special guest special guests. Tell us who are you and what do you do?

Speaker 5 (2m 32s): Hi everyone. My name is Mark Gray. I'm currently the head of hiring at a invisible technologies. Thanks for having me on

Speaker 1 (2m 41s): My pleasure. My pleasure, you know, we were having a rather interesting conversation just prior to recording about AI and how it has affected recruiting. I was reminded of a blog post our wrote sometime ago, and by the time this podcast Pell style, I'll put it in the podcast description. But essentially what I do was I looked at the job advertisement for recruiter into year 2000, 2002, somewhere around there. And then I looked at a job advertisement from today, or I think from 2018 and compared to two.

And although there was something that stayed the same, there were a lot of changes in, and the, the root of a lot of those changes was technology, you know, and I, it made, just made me when I wrote it. It made me just wonder how much things have changed in recruiting because of AI. Can you speak to that? How has in your opinion, AI been applied to the recruitment process overall in recent years, as opposed to the year 2000 2002?

Speaker 5 (3m 46s): Yeah. Even based off that, you know, what's super interesting, you know, how technology has changed hiring, you know, you look at that historically, it's like, well, you know, one of the greatest inventions of, of hiring was the fax machine, you know, and granted send your CV to post. Then obviously the internet arrived and job adverts. And then obviously LinkedIn, and, you know, I truly believe AI is that, you know, the next box, the machine of hiring. So I think what people don't realize is it's already been utilized heavily in the hiring side of things.

And it could be as simple as, you know, CD screening with an, an applicant tracking system and even the some machine learning elements to how to craft the most language neutral job specs out there. And, you know, in some case, you know, there's some AI based hiring assessments. So multiple categories, one is video-based analysis of the personality traits.

Others are AI based kind of logic, risk tolerance games. So it really is kind of exploding right now in terms of what people are benchmarking and what people are measuring and how they're utilizing it.

Speaker 1 (5m 7s): You mentioned bias in, in your, in your commentary there, do you think AI can really strain out bias? Cause I hear this argument that yes, it can screen out. So it's, it's awesome. It's perfect. And then I hear this kinda argument that well computers or AI or whatever are programmed by machine, by men, by human beings. So human beings have their own inherent biases. So it's going to come out and the tools they produced.

W w what do you fall into in that argument and that debate?

Speaker 5 (5m 41s): Yeah, it's a question I get asked a lot. I think it's more the structure of the question that interests me than the answer, but I will obviously answer and that's, we have a tendency to go, you know, AI is dangerous, it can be bad, and it has inherent flaws. And that's a true statement, but it's rarely said, is AI better or worse at bias that humans? And, you know, those are the two options when it comes to hiring either a human's going to check your CV and put you through the hiring process, or some form of AI is going to do it.

You know, the research on this is actually quite clear what school of economics recently published a paper, essentially stating that across multiple studies across multiple functions on across multiple different varied AI tools, the AI was less biased than humans. So the answer is, yes, it is definitely less biased. Is it perfect? And that's the question I usually, definitely not.

You know, we are still very much in the kind of embryonic phase of how AI is being utilized in higher. You know, it was what eight years ago, when like the first idea, even cropped audit and the first kinda beat models of AI and hiring a broad audience. So it's still very early days, you know, how, how long did it take to build an airplane? You know, the first place they built didn't fly too well. So we're in the back kind of similar kind of structure where we've made huge leaps and bounds with what we can do, but there's definitely still a long way to go before we could accurately predict to a certain depth of understanding of how accurate those results are.

Speaker 1 (7m 35s): I remember being freaked out the first time I saw a chat bot being used in recruiting and I, and I saw how someone could book their own interview time. And so a few screening questions, and I was like, oh my gosh, this is, this is amazing. And then I was thinking, okay, this is good. On one hand, other hand, all recruiters are done. You know, what the machines are taking over. Oh my gosh. But it didn't happen. Of course. What do you, what do you think about that initial fright when you first heard of chat bots or first heard about AI entering into recruiting space, do you think the end is the end is near?

Speaker 5 (8m 14s): Yeah, I think my initial is, you know, I think everyone, no matter what industry they were in, when they were like, whoa, what is this AI, you know, will I have a job in 10 years? I think there, there's definitely some people actually some functions that are on their way to extinction. Unfortunately, the next kind of 50 years, one of the big ones being radiographers I know it's a pretty big one for that AI image recognition software it's really come a long way, but in the context of higher, I think if you break down what hiring is to its core principles, you are, you have to judge, is this person competent and will they have to do it?

I put in the function, you're hiring them in and then you go, okay, well, what does that require? It requires the person doing this to know what exceptional looks like within this function. The ability to break down that function into all of its core components, and then replicate that and trying to figure out what kind of human do this. So you're talking about hundreds of variables. I'm being able to judge those variables now kind of AI help with some of that definitely kind of AI help with some of the more intangible aspects of hiring, not in the next 10, 15, 20 years.

And that's where the whole, my view is. AI would definitely be supplementary to the hiring efforts. I, I made the whole, you know, I think the talent market, if we look at it as a whole, you know, the job market people apply for jobs, quitting jobs, for whatever reason, you know, hyper inefficient. I think, I can't remember the number off the top of my head, but there was, there was some research done on essentially the efficacy of how an optimal hiring market would look like.

And at the moment it costs the us economy, something staggering, like $300 billion a year. So inefficient. So, you know, people going into a role where actually, if there was a competent hiring process would have filtered them out. And then this person that joined them and quit six months later, and then you lose them. You have that productivity gap of, well, we spent a month hiring this person three months, onboarding this person. Now they've last. And we have to start that again. So it is a hugely inefficient function globally.

So anything that can aid in moving that needle, you know, I I'm all for it.

Speaker 1 (10m 52s): You know, I, I hear some, some good examples of AI in, in, in practice, but I've also heard some, some harsh stories. You know, I hear horror stories a lot from Amazon for better or worse. They're, they're always trying to push the needle forward. You know, I think that I read where they could fire people by text, if they don't read certain thresholds and it's all automated when a manager doesn't even have to intervene. And then there was this other example, famous example, people were talking about it so much on Twitter and in conference chats about how Amazon had this algorithm to find the best software developers, but because of the parameters they put in place, no women were, are, were considered among that.

So it was like the software screened out women entirely kind of a czar. The, do you know of any other horror stories around AI or,

Speaker 5 (11m 47s): Yeah, there's one that I won't mention who, but it's a US-based AI company, essentially anyone that has a, a brief understanding, all, you know, very basic social science or very basic organizational psychology kind of looked at this product. And the question marks immediately because we were looking at it and going, how can they pull that up from a video when, you know, essentially academics kind of do it from a study.

And essentially some of the claims were, you know, from a short form video, we can deduce their level of intelligence. And that's very dangerous because a lot of the companies that use AI in video, it's, it's essentially based in, you know, hard science things like the big five personality traits, you know, th that's pretty much the benchmark for most modern psychology. Being able to measure that to a high degree of accuracy, you know, it's pretty much that's the, the, the, the, the entry point into this market.

But I think when companies start to claim, they can do things that essentially isn't empirically backed. That's when, you know, you need to kind of really worry about what kind of software you use.

Speaker 1 (13m 8s): Sure, sure. I I've even have heard about a particular company. I won't say their name, but when you're looking at them on camera, supposedly they could in, when the camera is scanning your face, it could pick up on the blood rushes around your eyes and under your skin and detect whether or not you've had a sudden rush of blood around a certain question and effect making it sort of like a lie detector in a way.

Speaker 5 (13m 35s): Yeah.

Speaker 1 (13m 36s): And I thought about that,

Speaker 5 (13m 41s): It's look, I might, even if you can like that, that, that is a Marvel of technological achievement, but I think if the premise of your hiring process is to see if someone's alive, you've kind of missed the point of while hiring process should be. Yeah. So yeah, there is, but I think with Andy you space, you know, there there's going to be good and bad doctors, you know, that's just the history of capitalism, unfortunately, is, you know, if there's, you know, if there is a technological gold rush in a new vertical or a new industry, people are gonna go and try and claim a percentage of the market.

And I think it would be irresponsible for me to say to all of them have bad intentions. I think sometimes, you know, it's, it's naive it to you as well. So look, as I've said, there's definitely room for improvement in this industry as well. And I also think there's definitely a lot of companies and organizations doing phenomenal work.

Speaker 1 (14m 44s): Sure. So specifically, if you could talk about some of the things that AI does in recruitment process, I've mentioned about how a chat bot can help you book scheduling, but what other things have you seen or heard that AI can do and in the process?

Speaker 5 (14m 60s): Yeah, I touched upon a bit of a start, but, you know, I know there's some companies, you know, large enterprise companies can't remember which organization now, but, you know, massive, massive multi-national. I think they mentioned they received something like three, 400,000 CDs a year, which is that there's no, there's no size of a recruitment team. That's going to get through all of those. And they utilize some kind of AI to essentially benchmark candidates based off what's written in their CV.

It's fairly rudimental, it's, you know, keyword searches, sentencing structures looking for, you know, written language capabilities. So that's what some of the big organizations are using. As I mentioned, there's some AI tools on assisting companies on how they write job specs, or they're not using any bias language or using any gender language or any language that historically kind of attracts a certain type of person, which is, you know, it's an interesting space as well.

The, the big one is obviously the video based assessments. That's kind of one of the biggest markets on the utilization of AI is, is it possible for us to extract some type of information? Will I recruit or having to spend 30 minutes squeezing the cochlea? Those are the kind of main three I know there is, like I said, some like game gamification utilizing AI. So, you know, I remember seeing one, which is like, oh, there's a certain pattern as to you hit the space bar, the balloon fills up.

And if you hit it too many times, the balloon explodes and then you figure out the Potter and then they'll see how close you are to blowing up the balloon while the tide takes God to see what your risk tolerance is. So there's certain gamification aspects of utilizing AI with as well, but I'm sure there's a few I missed, but the main kind of areas focused heavily on, you know, AI trading and they, I recruitment through video.

Speaker 1 (17m 13s): When you mentioned that bit about the balloon and it reminded me of a case study I've read, and I'll just always quote it because I just saw it, never it by the results they received. I think it was Unilever. They hire like 500 students out of college with this technique. They posted a job online and it actually wasn't even a job. It was an invitation to play a game with some sort. If I remember correctly, you come to his website, you play a series of games. It's about 15 minutes and they're assessing your skill sets.

And I guess doing things like the balloon thing you're talking about, and then you, you get a certain score. And then the people with the highest scores are invited to interview via higher view. And so you, you answered some pre-screening questions that way. And then those who score well there get to meet with the hiring manager and then they're, they're hired on the spot after they have a personal evaluation. And I thought, man, that is awesome. And I was like, cool, in that case study, but what I haven't seen, and I will love to see this.

Maybe you've heard of it somewhere, a company that has implemented a lot of AI in their recruitment process, but then checking on them maybe 90 days, sick, 30, 60, 90, a year later to see how much of those people have they been able to retain on the workforce. Cause I see studies that hear a lot about AI making recruitment process better, but how has it was the impact on retention? Have you heard of anything like that or seen anything along those lines?

Speaker 5 (18m 43s): Yeah, so I think it depends how you work the problem box. So I think one of the strengths of how I view device in the past is, you know, to form the basis of what you're trying to hire for, you need to understand what your best people are, your best people deeper than the surface level. So if you can build that image up and then try and replicate it in certain paths, especially you'd like toss with a tangible I put. So sales is a great example. There's a very clear correlation to how people are performing in terms of retention.

So that's where it gets a bit more complicated because hiring my general view is if someone leaves within the first three months of that joining, that's a feeling of higher anything after that, it, you know, it's a variety of reasons. It could be the manager, it could be the culture, it could be, you know, changes in how the company was doing things that there's a million reasons. So I think retention is actually a much more complex issue because if you look at what was it, it was texts that did the personality treat based model of performance.

And he goes, okay. So if we look at how people perform, it's like, you know, you have their core traits, how that interacts with the tasks. They do the variety of those tasks. The T dynamics and team dynamics is like very complicated because it's like, well, how does one person's interaction affect their work based off that person's interaction based off their tricks. And then, you know, it also feeds into their like intrinsic and extrinsic reward centers and then not judges will I put, looks like, so we, you put all these together.

It's like, how does a person stay the business for less than a year? Well, obviously there must be some critical breakdown in some of these factors. So, you know, maybe this manager has a certain personality type that favored people that were highly agreeable, but this time they came in and they were very disagreeable and there was clashes based off of that. And thus, you know, you have something called like the chameleon effect where you go, well, I think this person probably isn't good and they'll say, we'll treat them like, they're not very good.

And the person I do think is good, I'll treat them like, they're amazing. And there's actually some, you know, performance enhancement and reductions because of that, because then the individual thinks, oh, I must not be good. Cause I thought about was a little, so, you know, it's really hard to nail like retention is a really complex problem. I don't see AI fixing that in the long run, unless you could, you know, tangibly link retention with certain key factors. So maybe, you know yeah. As I said, like someone's reward center.

Okay. Do people that say in business X for over three years, you know, is part of their intrinsic reward center somehow tied to the culture or the type of work. And if you can prove that that's really exciting.

Speaker 1 (21m 42s): Yeah. And then as you were saying that I was thinking there are other factors that I don't think could be quantified in any kind of algorithm, like who could anticipate a pandemic or a war on Ukraine, you know, on a personal level when someone's gonna fall in love and decide to move to another city or country.

Speaker 5 (22m 2s): That's a great, that's a great point. You know, technology will only take us so far, you know, I'd say what the heart wants is that far more complex that,

Speaker 1 (22m 15s): You know, what, usually when topics of AI around recruitment comes up, I noticed that there is a lot of attention on the ATS system from job seekers and they're saying, okay, what keywords can I put in my resume to make sure that, that, that, that the robot picks me up? You know, it makes me sort of wonder what should accompanies say, well, what kind of disclaimer, maybe they should develop for the sake of job seekers, you know, saying, Hey, we use fancy robots and algorithms to decide for the best candidates for us, you know?

And then some job seekers may see, that'd be like, I feel, I feel swindled in a way, you know, if your machine doesn't find me, how do I know I'm going to get a fair shot? So it's, I can see how it all makes it easier for the company, but for the job seeker, I imagine they are a bit nervous about it. That what would you say to them?

Speaker 5 (23m 11s): Yeah, I at firstly challenged the assumption that even if there wasn't AI they'd get a fair shot, you know, I wouldn't even know how to define what a fair shot looks like anymore. You know, I take hiring these days. It's just so difficult, both from a candidate's point of view, a recruiter's point of view, you know, as I mentioned, w when you break down the core components of what hiring is like, no one really knows how to hire, you know, it is so complex. We have some guidance, you know, some north stars as to, you know, best practices, but no one truly, really knows how to hire properly with all the complex pizza balls.

We can do it to a percentage. We can get it to a point, but from a combinate perspective, you know, I understand that fear, you know, I personally, we don't use any AI, you know, CV scanners or anything like that. On the flip side, I think it would actually, there's a huge opportunity in the market for a company to, you know, paying, paying a certain companies, that job sites to see if they use AI like a best practices section for a confidence to, you know, oh yeah.

This type of AI favors X, Y, and Z use these keywords. But yeah, I think there's going to be an evolution because essentially the benchmark of AI CD scouters assumes that a CDs are accurate, be CDs, tell a full story. I see CVS can predict future performance, which is, you know, once again, that's been clearly debunked, you know, past performance is no reflection of their future capability. So it's also why I think it's AI in general is a tool.

And if you use the tool correctly, you get, you know, great output. If you use the tool poorly, you'll get poor outputs. And I think that's actually the bigger disclaimer that AI has this like really weird paradoxical kind of marketing appointment, where on the one hand, like all it's, it's the great savior. It's going to figure out all the things as humans can to make it make everyone's lives easier. But at the same time, it's also this, you know, force for evil and we should be afraid, but no one really talks about it.

Like it's still a tool. And you know, if you don't set it up correctly, you're not going to get anything useful out of it.

Speaker 1 (25m 35s): Yeah. I tend to think of AI in recruitment. I made the comparison that it's not the Terminator. It makes us all Tony stark. Well, we're all like iron man. You know, it's, it's a tool like a tool of armor that we could use, but ultimately we're making the decisions. Some of the things that I like about AI as well is that, and I heard his argument from the beginning and I think it still holds true is that it allows recruiting more time to do things that AI presently can't do. And ideally, we'll never be able to do such as managed relationships.

You know, recruiting is all about relationships. You gotta be able to talk to the person, convince them that, that your opportunity is better than the one that they're already in. You know, you gotta have managed relationship with the hiring manager, convinced them that, you know, this person you found is awesome. That'd be perfect footage role. I don't see any kind of algorithm or technology being able to facilitate that human interaction to the point where people have a gut feeling and say, yes, I feel it in my gut. This is the person I need to hire. I don't see machines doing that to you.

Speaker 5 (26m 39s): No, I actually, I, I kinda go gut feeling is a dirty word in the recruiter world. Second, a hiring, I think gut feeling is something you earn. If you, if you'd been a hiring manager for 20 years, it holds a lot more water than a, you know, a new graduate. So, but I get the point you're making and it completely, and that's, you know, that's, you know, going back to like AI is still like, so early on, you know, how would you even, I couldn't even begin to fathom how you build that kind of neural network to teach human interaction, you know, sent you to AI recruiter.

You know, I think, I don't think we're anywhere near that. I don't even know if we can ever get there maybe like 200 years or something, but yeah, the impact of the personality and who you are and how you come across, it still holds a lot of weight, you know, at the end of the day, you know, an AI can give this person to score, but, you know, even if this person scores highly, but they're, you know, a nightmare interview and you're like, do I want to bondage this person? You know, there's still, you know, some human tricks that wall was kind of balancing, but I think also, you know, one thing that doesn't really get talked about is, you know, most developing worlds, you know, the birth rates are done.

You know, I think the us is, you know, for the first time ever had more deaths than births last year.

Speaker 1 (28m 18s): Yeah. We're in that, I think it was called super aged nation status or getting close to it where like a 20, over 20% of the population is over 65 or something like that.

Speaker 5 (28m 29s): Yeah. So, you know, the, the, the amount of people in the job market is going to decrease significantly over the next 20, 30 years. So I even think the bigger question is, you know, yes, AI recruitment is hyper useful, but you know, how does automation play into the future of jobs in terms of tasks? And I guess, you know, segue, that's essentially what invisible is, is focusing on heavily, but also, you know, yes, AI will help people, help companies find jobs, but I take the reverse of it is going to be true to the future, as well as how will AI help candidates find jobs.

Because if we're seeing a huge decrease in the workforce, the one thing that we're going to have to tackle is that inefficient marketplace, you know, as I said, $300 billion a year, that's significant. So what can we do to move the needle at both ends both from an employer perspective, but also a Catholic perspective. And I think that's a space that really hasn't been explored extensively yet is could we get to a point where, you know, candidates companies are on this unified platform, we can read what accommodating wants, what the rewards are, what the traits are, what they're competent that, and match them up with businesses where they'll actually be like hyper fulfilled, you know, really high, highly engaged and thus have a very strong output.

So I think that's also an interesting perspective that doesn't really get discussed on.

Speaker 1 (30m 3s): Yeah, I think that would be a, the next killer app. If someone could develop that. The only thing close to what I think you're suggesting that I have seen is this tool called Esther bot Esther BI is a chat bot that a, I think a software developer created because they were getting inundated by recruiter requests to talk about software developer jobs. I think it was, I think it was all the developer jobs. And so what they did was they said, Hey, instead of talking to me, talk to my chat bot first, cause the chat bot can answer all of your, all the frequently asked questions that I get.

And then if you get past it, the robot, then we'll talk. I thought that was, I thought that was pretty brilliant. I, at the time, and maybe even now I see that as a tool for people who are constantly being harassed, so to speak, I won't say harass, but constantly approach tutors. So I tend to think software developers, but for the average, Joe, I don't know. I think it would be really cool if someone could develop something that would find a job for the candidate or at least find opportunities.

Speaker 5 (31m 10s): Yeah. I leave it, not the whole process, you know, hiring is very inefficient. You know, you, you pick a hundred people, maybe three or four get back to, so, and that's part of the bigger problem, but, you know, imagine a scenario where you get a match from a, you know, a job perspective, a function perspective, a growth potential perspective, but also a compensation perspective. Cause that's, that's where some of the inefficiency lies as well as like, oh, well, can I make more here would be more happy there will that, you know, figuring out that, that mess, that's the kind of next that the next great implementation of AI, but I still feel we're quite far away.

One thing we are close to, which is very interesting is with all this trade analysis software, I spoke to one of the, the doc, one of the CEOs of these companies, he's like, we're essentially one or two years away from being able to identify dark traits. So, you know, you know, things like narcissism, psychopathy, Machiavellianism, he's like, we're not far away from that. I thought that was very interesting because you know, the, you know, the initial reactions were like, oh, this is terrible.

You know, there should be something shouldn't be discussed, but I was just thinking, well, in some roles, that's actually a positive, it's like

Speaker 1 (32m 37s): A dating app.

Speaker 5 (32m 42s): I think it's, I think, you know, there's some, there's some crazy things in the pipeline for most of us that said job seekers, but maybe even for dating apps. So yeah, it's, it's an interesting place to be

Speaker 1 (32m 57s): Most definitely most definitely. I think that until AI and machine learning can, can learn intuition. I think human beings are going to be okay because I can see setting up a process to create a phone. You know, you making iPhone 10, 12, whatever the number is nowadays, you know, you make the next iPhone, but having a machine constantly crank out I-phones is not going to give someone an inspiration to make the next Samsung phone or the next whatever else phone could that takes intuition, takes an idea, takes a creativity.

And I, I, I can see machines imitating creativity where, you know, we put in different variables that it may appear random, whatever, but in that truly random because someone programmed them at some level, you know, so until until machines are sentient, like, I don't know, data on star Trek or some others, I think we're, I think recruiters are going to be okay. I have really enjoyed this conversation. If someone wanted to reach out to you personally and continue the conversation on how can I find you online?

Speaker 5 (34m 9s): I'll set up a chat bot that no, they can reach out to me on LinkedIn, Mark Gray, G R a Y, feel free to reach out, happy to have a lively discussion about the future AI in hiring, or just any general questions of how I utilize it and implement it. And yeah. Thank you so much for, for having meal

Speaker 1 (34m 40s): Again, sir.

Speaker 5 (34m 41s): Thanks.

Speaker 1 (34m 52s): Thank you. Thank you. Thank you a thousand times. Thank you for listening and subscribing to our podcast. If you have any questions, comments, or suggestions, please send them to us. You can reach us at TribePod that's T R I P O D at Proactive Talent dot com. We look forward to hearing from you.