Reducing Bias w/Lindsey Zuloaga

[00:00:04] Welcome to The Recruitment Flex with Serge and Shelley. I'm Serge.

[00:00:09] And I'm Shelley. And we talk all things recruitment starting right now.

[00:00:17] Bonjour and welcome to The Recruitment Flex. Shelley,

[00:00:21] more interviews, more guests. We're learning every day.

[00:00:24] Yes. So Serge, I'm really excited because I think this would be the very first time

[00:00:30] we have a chief data scientist on the show.

[00:00:33] Yes, it is.

[00:00:34] I am absolutely thrilled. What an opportunity for us. Who we have joining us today is Lindsay

[00:00:42] Zulaga, who is the chief data scientist with Higher View. Welcome to the show, Lindsay.

[00:00:48] Thank you both. Great to be here.

[00:00:50] As I said, this is a first. This is just an awesome opportunity. And I have to admit, Lindsay,

[00:00:57] I was checking you out on LinkedIn and I nearly fell out of my chair.

[00:01:03] Your background is so fascinating, which only a recruiter would say because that's what we do

[00:01:09] for a living, right? Like we study people's career tracks and work histories. So I would

[00:01:15] love it if you could share with the audience a little bit about you, your career path and your

[00:01:20] journey. How did you end up in HR tech? Yeah, definitely not planned that way.

[00:01:27] I'll start pretty far back. As a kid, was a pretty scientific thinker, but I definitely

[00:01:34] didn't really see myself as a scientist. I was really good at math, but I didn't really

[00:01:40] understand what math was actually for. I remember being in eighth grade and having someone ask

[00:01:46] the teacher, when are we ever going to use the quadratic formula? And the teacher was like,

[00:01:49] you won't. And then when I got to high school, I had a really amazing physics teacher. I

[00:01:59] actually wrote a little essay that I have posted somewhere back years ago on LinkedIn about him,

[00:02:05] but an amazing physics teacher that really opened up my eyes to like,

[00:02:09] oh, this is what math is for. Actually math describes the world and there's quadratic

[00:02:15] equations used in the equations of motion of how things actually move. And that was huge for me

[00:02:22] to connect why math is important. So I was the first person in my family to go to college.

[00:02:28] I had very little confidence with kind of what I was going to do or even going at all. I

[00:02:34] didn't know the first thing about college, but I kept chipping away at it and I worked

[00:02:39] several jobs as I went. And I was very intimidated by studying physics or majoring in physics,

[00:02:45] but it was the only thing that really I was passionate about. So I went for it and actually

[00:02:50] started getting into research in my undergrad and then went into a PhD program as well.

[00:02:56] Never saw that coming, never excited to do that, but ended up getting a master's and

[00:03:00] a PhD in physics and doing a postdoc and surprising myself and my whole family and

[00:03:07] doing that. And when I decided to transition from academia to industry was when I learned how

[00:03:13] broken hiring is, and I was shocked. I was working in this space that's pretty competitive

[00:03:19] and I was doing pretty well in academia. And so I thought I should be fine getting a job,

[00:03:24] right? But I went out and applied for many jobs and it's just a black hole.

[00:03:30] And it's been a weird situation where I was overqualified for a lot of jobs,

[00:03:37] but also had never had a job before in the real world as I call it. So it was hard and that

[00:03:44] ties into my passion for working in this field was this, you go into an applicant tracking

[00:03:50] system, you upload your resume, then you have to manually re-enter everything that's in your

[00:03:55] resume because it didn't parse it very well. You make a cover letter and a specific resume

[00:04:00] with the keywords from that job posting, and you finally press submit and you never ever hear

[00:04:06] anything ever again. So that was really rough. I did end up getting a foot in the door in data

[00:04:11] science, which with the amount of math that I had had was pretty easy to pick up, right?

[00:04:17] This was before there were data science programs and data science majors. I taught myself

[00:04:22] this online very quickly, I liked understanding algorithms and machine

[00:04:26] and got into that in the healthcare space initially. And then found HireVue pretty

[00:04:31] soon after and was really interested in what they were doing looking at video interviews

[00:04:37] and analyzing video interviews to predict how well someone will fit in a certain job.

[00:04:43] And I'll admit I had my skepticism going into it, kind of understanding how they were

[00:04:47] doing that, what data they were using. And it was early days. HireVue had been around for a

[00:04:52] while, but they were just starting to use machine learning or AI in the process. So

[00:04:57] it was really exciting to get in early when they started doing that. And I feel a lot

[00:05:01] of ownership. It's been almost eight years that I've been at HireVue, seen that product

[00:05:06] go from its infancy to what it is today and we're adding more and more products and

[00:05:11] capabilities all the time. So it's been an exciting journey.

[00:05:15] I was trying to figure out the timelines exactly. So you would probably have started around 2016,

[00:05:21] correct?

[00:05:22] That's right.

[00:05:23] And HireVue came in fruition when around 2014?

[00:05:27] We actually were founded in 2004, but it was very small time. So yeah, our founder

[00:05:34] was actually doing his MBA at a small university here in Utah and he could not get an

[00:05:40] interview at Goldman Sachs because they didn't recruit in his school. And Goldman Sachs was

[00:05:46] literally like, you could see it from his window, but he couldn't get an interview there. So his

[00:05:51] idea was we could open up the funnel more if we ship people webcams and they could record

[00:05:56] themselves. And so that was the original idea. And for a long time we did, we shipped

[00:06:01] people webcams. We still offered to do it not too many years ago and no one had asked

[00:06:05] in so long that we stopped offering. But that was the original idea and it grew pretty slowly

[00:06:11] for a while and then took off right around like 2012, 2014 more and more.

[00:06:16] Shipping webcams. It feels like sounds that would be really hard to scale, correct? If

[00:06:23] you think about how many interviews are running through the HireVue platform right now,

[00:06:28] it'd be almost millions of cameras you'd have to ship. Thank God for the advent of

[00:06:33] laptops and computers getting webcams. On that note though, tell us more about HireVue.

[00:06:38] For the audience that's never heard of it, can you give it a little bit of a breakdown

[00:06:41] what problem it's trying to solve? Yeah, originally we were this video

[00:06:45] interviewing or asynchronous video interviewing where candidates can record themselves answering

[00:06:50] questions on their own time and then recruiters and hiring managers can watch those videos.

[00:06:55] A big challenge that we still saw is there's still hundreds or sometimes thousands of applicants

[00:07:00] for every role and it's hard to watch all those videos depending on the role. That's where

[00:07:05] we started thinking more about assessing the videos through automatic means or through

[00:07:11] machine learning. We built a product, it's not all of our interviews, a lot of people think

[00:07:15] that if they're taking a HireVue interview where their AI is being used on them.

[00:07:19] It is disclosed if it's being used. It's not even the majority of the interviews actually.

[00:07:24] Some of our customers do use that AI scored interviewing where candidates have the same

[00:07:30] interview. We're measuring things that are related to that job. We put a lot of work

[00:07:34] into that side of it. We have a big team of PhDs in industrial organizational psychology.

[00:07:40] We're assessing this job to measure the things that matter. We've trained algorithms to score

[00:07:46] on a standardized rubric. We're predicting things like teamwork ability or communication

[00:07:51] skills or things like that. Those scores help guide human reviewers as they evaluate candidates.

[00:07:59] We've also added with time more products, things like games. We have acquired other companies

[00:08:05] that do things like coding challenges, virtual job tryouts, scheduling. Scheduling

[00:08:11] interviews with people is a big time sink. A lot of it's trying to automate the tasks that are

[00:08:17] really awful for humans to do so they can focus on the more rewarding things and the more human

[00:08:23] things in the process. We see a lot of our customers have automated a lot of those pieces

[00:08:29] and they see a lot of great results from that. I'm just wondering if you could go back just a

[00:08:34] smidge because you talked about the origins of the async video interview and the challenge being

[00:08:42] you've got just the inherent bias of somebody watching a video of someone, right? How could

[00:08:48] you possibly get through a thousand videos? Where is HireVue at today for clients that aren't

[00:08:54] using the AI scoring tool? How is that working? Is it still asynchronous? Is there some sort of

[00:09:02] filter up front and only a small percentage do an async video? It really depends on the company.

[00:09:09] Yes, that's true. Sometimes there's filtering that happens before someone's invited to an

[00:09:14] interview. We do have customers that watch every single video, right? They have teams of people

[00:09:19] that are evaluating. There's reasons for that and sometimes it's concern over using AI,

[00:09:24] which we have a lot of great data that's very convincing to show that our system is

[00:09:30] better than humans in every way we've measured because of this consistency, less bias,

[00:09:36] and better at predicting good hires. We got to prove it out further down the road with our

[00:09:41] customers that you're seeing the results a year in or whatever it may be that these are actually

[00:09:48] better hires and they stayed longer on the job. Those people that we put in the top tier.

[00:09:55] Actually, Lindsay, I'm a little bit surprised. I thought the product was stitch because I

[00:09:59] know there's been some discussion was brought up in a book of the algorithm as far as

[00:10:04] leveraging AI to visually assess. It's a live product right now, right? It's available to

[00:10:10] customers. It wasn't shelved for a moment in time or did I read that completely wrong?

[00:10:16] You mean using the actual visual component of the video.

[00:10:19] Exactly. Yes.

[00:10:21] Yeah. We did that for a time and we discontinued it

[00:10:24] four or five years ago. It's been a while.

[00:10:27] Yeah, exactly. That's what I thought.

[00:10:28] Yeah, it's absolutely gotten a lot of attention. The history there is early days we were

[00:10:35] interested in muscle movements in the face as you can imagine why. That's a big part of how

[00:10:40] people express themselves, particularly for certain roles where you're looking at someone's

[00:10:45] a flight attendant or a customer service role and we want to see if they smile or those kinds

[00:10:50] of things could be related to their performance in that role. What we found through our research

[00:10:55] is that we've always seen language had the most predictive power and what someone's face does

[00:11:02] aligns very closely with their language. It didn't actually add more value beyond the language.

[00:11:09] Also, these large language models have just gotten so much more powerful. The value we

[00:11:14] get from language is just better and better. At the same time, we had concerns over

[00:11:21] the visual aspect of what are you looking at in the face? If I move my eyebrow in a funny way,

[00:11:27] am I not going to get the job? People are concerned that there's something they can't

[00:11:30] control there. Of course, throughout the whole process we're testing the algorithms

[00:11:35] for bias. We're looking into a lot of things. There are issues around lighting or skin color

[00:11:43] that do make a challenging address. At the end of the day, we just found it was more

[00:11:48] concerned than it was really adding value. We did away with any non-verbal communication

[00:11:55] back in 2020 and then phased that out. Just to clarify right now, when you're doing

[00:12:02] any type of AI assessment on a candidate is based on language? Is it words used? Do

[00:12:09] you look at accents? No, it is the transcript. The transcripts are fairly robust now. These

[00:12:15] systems are really accurate, transparent people. The only way accent would come into play is if

[00:12:20] someone had a very thick accent, their transcription accuracy could be less, which is true for humans

[00:12:26] as well. We have to have ways of looking at thresholding or flagging for human review if we

[00:12:31] feel like someone's very hard to understand. At the end of the day, it is looking at

[00:12:35] the content. These new large language models are really good at getting at context and nuance.

[00:12:41] It's less about the actual words you chose or where you're from in the country or the world and

[00:12:47] the meaning there. I want to touch on standardization in hiring. There was a recent

[00:12:54] article, I think in Safety Mag that talked about AI and automation bringing much needed

[00:13:00] standardization to the hiring process. Quite honestly, I couldn't agree more.

[00:13:04] But would you mind just elaborating a little bit on how this standardization

[00:13:09] is going to help reduce bias and improve DENI in hiring?

[00:13:15] Yeah, we've seen it with a lot of our customers. It's amazing how having a process that you're

[00:13:21] looking at things that are relevant to the job and nothing more just naturally gives you a

[00:13:26] boost in diversity. AI or not, having something that is highly standardized, we see this.

[00:13:33] That means that we're looking at something that's very job-related. You're applying

[00:13:37] to be a teller. We have you do an exercise where your accounting changed or something like that.

[00:13:42] This is highly related to the job. We don't have any gut feelings or any judgments based on

[00:13:49] your name or where you're from or what school you went to or your GPA.

[00:13:53] You immediately get a boost from that and that you're giving everyone that same chance.

[00:13:57] Further, when you do use AI, we've all seen the headlines of AI can be biased and of

[00:14:03] course it can. You have to be careful that it doesn't repeat the mistakes of the past.

[00:14:09] Like any powerful tool, you can use it for good or you can use it for harm.

[00:14:13] You've got to be careful, but you can actually mathematically tune the algorithm to ignore

[00:14:20] certain factors. An example would be in a video interview, you might have even training data

[00:14:26] that has bias against women. Well, there are things that are hinting towards being a woman.

[00:14:32] You talk about child care or something like that. With an algorithm, you can automatically punish

[00:14:38] that feature so you can say ignore that word then. If there's a word like that or a subject

[00:14:44] matter that differentiates men and women, the algorithm should be blind to that word.

[00:14:49] That's something you cannot do with humans is just kind of tune out certain aspects

[00:14:54] that can lead to bias. There's a lot of promise there.

[00:14:58] So is that built right into the product regardless of the client or is it up to the

[00:15:03] client to say, okay, when I am configuring my system or configuring my version of higher view

[00:15:12] or is this right across the product line? It's across the board and we published a

[00:15:17] peer-reviewed paper on this last year in the Journal of Applied Psychology.

[00:15:21] We build it into the optimization. So we're not just trying to make the algorithm predict

[00:15:28] an outcome accurately like your team orientation ability or something like that, but we also are

[00:15:35] penalizing it for having group differences in the outcome. So it's built in that it's

[00:15:41] optimizing those things at the same time. There is follow-up for the individual customer.

[00:15:48] So when we release that algorithm into the wild and people are using it,

[00:15:52] is there anything unexpected? Like now we're using it on a different population.

[00:15:57] Is there any bias that creeps in because of the way you're sourcing? Like you're sourcing

[00:16:01] from this particular college that has more people of color but this other college doesn't

[00:16:07] and that's a different population than we trained on. So we would want to check that

[00:16:11] regularly and perhaps mitigate further on that particular population.

[00:16:17] So is that higher view doing the check or the client has to put up their hand and say,

[00:16:22] I need you to double check?

[00:16:25] No. Yeah. It is part of our statement of work that we do that at least annually.

[00:16:30] That's best practice in assessment in IO psychology. In the US, we have the New York

[00:16:35] City law now, which is the first AI and hiring law that requires we do third party audit

[00:16:43] annually, which is pretty much that. It's an adverse impact check with a third party every year.

[00:16:49] We've been doing that internally for a long time and we've done several third

[00:16:52] party audits as well, but that's an annual thing that we're doing going forward to comply

[00:16:57] with that. Obviously if I'm a TA practitioner, an HR practitioner, there's a lot of concerns when

[00:17:02] it comes to AI tools and what happens if something goes wrong? What happened if the

[00:17:08] tool is biased or flawed in any way? It looks like the liability is going to be the employer

[00:17:15] or the end user that's leveraging it. So I'm just curious, what advice would you give to

[00:17:20] employers that are looking at these types of tools? Like what should they be assessing? Should

[00:17:25] they have peer review studies like you guys are doing third party audits? What should they

[00:17:29] look at? So AI regulations are in their infancy and people are trying to wrap their

[00:17:36] head around this and it's hard. I think they're going to need to be specific to the industry.

[00:17:40] Like when you say regulate AI, that's so broad and no one really knows what it means.

[00:17:46] Everyone's saying here are the guidelines, the framework, and they're all saying similar

[00:17:50] things but they're not specific enough. So as a data scientist, you think about fairness or

[00:17:56] ethics and what that means. That's really hard to define for some data scientists. If you work

[00:18:02] in a social media company and you're trying to figure out what content to recommend to someone,

[00:18:07] no one has even defined what fairness means or what you should even think about. Really,

[00:18:13] that's like a completely new thing. Hiring is not a new thing. So I think we have to

[00:18:19] communicate that to people sometimes. Like this is not totally new. We're using new tools

[00:18:23] to do something that people have been doing for a long time and there's a lot of established

[00:18:29] stuff here. So we're building on decades of science and best practices, but we're using

[00:18:35] new tools. We still have to follow all the rules and that's something the federal government has

[00:18:40] come out and said, hey, if you're using AI in hiring in the US, you still have to follow the

[00:18:45] Americans with Disabilities Act. It's like, well, no shit. Of course you do. All these

[00:18:50] things are kind of obvious but some people have reacted like these AI systems, they're just

[00:18:55] doing whatever. They're doing whatever they want and of course that shouldn't be true.

[00:18:59] You should look for a vendor that is an expert in hiring and knows the space, knows HR,

[00:19:04] knows TA, following all those classic, we do our adverse impact checks at this frequency.

[00:19:11] This is what we look for. These are psychologists that are trained in the industry to know

[00:19:16] this is what it looks like when we have a problem. How do we build up the evidence

[00:19:21] that this is working? It's all this science that's been around for a really long time,

[00:19:26] but you've just brought in new tools of assessing people.

[00:19:30] TITLE There's a ton of noises when it comes to this space, AI and HR tech,

[00:19:36] and as a practitioner, you're getting hit from everywhere and you don't know what's real and

[00:19:40] what's not. Has that been a challenge for HireVue because you're competing with a lot

[00:19:46] of startups that might not have a team of data scientists. They might just be leveraging

[00:19:51] ChatGPT plugin to do assessment on candidates, which is doable. I don't recommend you whatever,

[00:19:57] but how does HireVue counter against that noise in the marketplace?

[00:20:01] TITLE Yeah, and it's going to get more interesting to your point. ChatGPT can

[00:20:05] do a lot of things. It's really cool. But how do you put guardrails on it? How do you

[00:20:11] make sure you have science behind it? ChatGPT wasn't trained to be accurate or truthful,

[00:20:17] necessarily. Yeah, I think we will see in every industry more startups come online that

[00:20:23] are doing cool things really quickly. We've been set up really nicely through

[00:20:27] being a pioneer of using AI in this space. We've had to get through a lot of scrutiny

[00:20:33] and it's made us stronger. We've done third-party audits. We have an explainability

[00:20:38] statement. We've been pushed to be more transparent and that has been really good

[00:20:42] for business. It's been good for us. I was at Unleash a couple of weeks ago or HR Tech in the

[00:20:48] fall. Everyone's talking about skills and they say we infer skills. And I'm like,

[00:20:52] how well do you infer skills? I mean, do we have an accuracy on that inference or is it

[00:20:57] just check the box? And ChatGPT is really good at inferring skills by the way. So

[00:21:02] can we all do that now or are we even going to discuss how accurate we are or how good we

[00:21:08] are at inferring skills? It comes back to these questions of scientific rigor. And so I think

[00:21:13] as you're looking at vendors asking for that, if they can't explain what they're doing,

[00:21:19] any technical ability, then I would be concerned. And even bringing in your own

[00:21:23] people from your company that might be more science-y or math-y to ask questions,

[00:21:28] even if they're not in HR, dig at it a little bit. I think can be helpful as well.

[00:21:32] One of the things that we noticed going to HR Tech in 2022, everyone had D, E and I everywhere.

[00:21:39] Right? And then after ChatGPT came out at Unleash the year after, it's like they put

[00:21:44] a sticker of AI over the D and I, and that's what their tool is doing now. And we come

[00:21:50] across exactly that situation when you're asking how they're leveraging AI on the floor.

[00:21:56] The majority don't know. And when I said the majority, it's probably a salesperson they hired

[00:22:00] and obviously it's not the founder. It's not fair, but you gotta explain exactly how the

[00:22:05] tool works. There was a lot of buzz earlier this year. There was a book that came out

[00:22:11] called The Algorithm. And boy, did they ever pick on HireVue. Nante, Heike Schulman.

[00:22:19] She was very critical of HireVue in that book. So I would love to give you the opportunity to

[00:22:27] counter or maybe respond on what HireVue does to ensure that the data used to train its algorithms

[00:22:35] is diverse and unbiased. Yeah, I talked to Heike, she quoted me in the book. I think

[00:22:42] there's definitely, as I mentioned on the visual aspect and the nonverbal,

[00:22:48] there's a lot of outdated information in the book. So there's a lot of focus on things that

[00:22:52] we haven't done for a long time, which I think is unfair given where we're at now.

[00:22:57] She's a very smart person and I respect her a lot as a journalist. I think that

[00:23:00] there's a lot of good questions that she brings up that we've grappled with for years.

[00:23:05] And having those conversations pushed us in the direction of being more transparent,

[00:23:09] having better practices. I think our practices are very good. We've set the bar in the industry

[00:23:16] for sure, but because we are a pioneer, we open ourselves up to be scrutinized. When we build our

[00:23:23] training sets, when we train our algorithms, we control a lot of that. And when we started

[00:23:29] doing this, we didn't necessarily, right? We would get our data from our customers and we

[00:23:33] realized pretty early on that we want to have a lot of control for this exact reason. The

[00:23:38] data that you use is so important. If we're trusting our customers, how are they measuring

[00:23:43] performance? Do we know that they are not biased versus, hey, we're going to have trained evaluators

[00:23:49] that evaluate interviews that we know have some background in this and we're giving them the

[00:23:54] exact rubric. We're having multiple reviewers evaluate every single answer. We are comparing,

[00:24:01] we're looking for discrepancies, we're discussing, we're looking at the diversity

[00:24:05] of that group and the diversity of the training data. And that is all published in our explainability

[00:24:11] statement. Our explainability statement, the short version is 30 pages long, but it's open to anyone

[00:24:16] who's interested to go through. We're just updating it right now, but it's a living document.

[00:24:20] We'll change it as we get more data, as we improve our practices or anything that we change

[00:24:26] in how we train algorithms, et cetera. But it goes through all of that. So I'm really proud

[00:24:30] of the work that we've done. Like I said, we've learned through time and dropping the video

[00:24:37] aspect of the evaluation was one of those ways where it's like, hey, we saw that this was not

[00:24:43] worth the concern that it was causing and we're willing to admit that we could do it a better

[00:24:47] way. And we did. We see it as a journey that we're on. And there was a lot of focus in the

[00:24:51] book of where we were four or five years ago, in my opinion. So. Thank you. Thank you, Lindsay.

[00:24:57] Cause as I was reading it, I was thinking, wait a minute. I was pretty sure that you had sunset

[00:25:04] that part of the product. Thank you for clearing that up. Search over to you.

[00:25:08] Yeah, absolutely. And I'm glad it was brought up because we read the book,

[00:25:12] we had Hilke on the podcast as well. And I think she called out some really good things

[00:25:17] in the sense that, Hey, we got to start thinking about this, but I agree it was unfair

[00:25:21] to a lot of vendors that were named in the book as far as what their actual practices are.

[00:25:25] So thanks for clarifying that. Now let's look in the future. We're in 2024 and we're moving

[00:25:32] really quickly. Seems the last couple of years, technology has advanced to the pace

[00:25:35] that we've never seen it. And that's always been the case, but AI has just sped that up.

[00:25:41] So what is the world of work and recruitment going to look like in 2030? Are we going

[00:25:47] to have robots interviewing robots? Yeah. I always think about that. Are

[00:25:53] we just going to have robots sending our emails for us and other robots reading those emails?

[00:25:57] And then we're not even talking to each other anymore. No, I love this question. I think you're

[00:26:01] right. We're on this exponential curve. So we should expect the unexpected and we should expect

[00:26:06] to be blown away several times in our lifetime. These big things like the printing press or

[00:26:12] antibiotics or the internet. Like maybe we're in one of those moments right now. It's hard to

[00:26:17] say when you're in it, but it feels big. And there's a couple of things for the future of

[00:26:21] work. Generally, I'm pretty optimistic about us as humankind being able to adapt like we always

[00:26:29] have. There will be a lot of automation. There will be a lot of jobs that go away,

[00:26:34] but there will be more jobs that are created. I know a lot of people are worried that this

[00:26:38] will be the final time and that it won't happen again, but history has proven that

[00:26:42] that typically we see things shift in a way we don't yet understand. And we can't really predict

[00:26:49] yet what that will look like. We're already seeing it. We're seeing layoffs, but we're seeing a lot

[00:26:53] of new job creation as well. Different types of jobs that we just have to be more agile. We

[00:26:59] have to understand skills and jobs and people better. And I think AI is going to really help

[00:27:05] us do that. So particularly in the world of hiring, I'm really excited about AI just helping

[00:27:12] people transfer their skills to different roles. And when you're hiring for a job that no

[00:27:18] one's ever had before, you can't really do it the old way of looking at the resume and saying,

[00:27:23] I'm just going to hire someone who's had this exact job before because no one's had that job

[00:27:27] before. So you have to say what are the jobs that are close to this job? And it might be

[00:27:32] something you didn't expect, but this job is similar to a mail carrier or something. Maybe

[00:27:38] you didn't see that, but AI can tell us this has a lot of the same skills

[00:27:42] and that's the people we're going to want to move over. And I'm excited for candidates

[00:27:46] because I think it means a lot more people are going to have options and they're going to be in

[00:27:51] jobs that they really like. And hopefully applying for a job doesn't mean that you're

[00:27:55] one by one entering different funnels where only one person comes out the bottom in this kind of

[00:28:01] requisition based model. And there's more of a multi-dimensional space where opportunities are

[00:28:07] available to you. And you own a lot of your own data, including your assessment data,

[00:28:12] more robust data than just a resume or a LinkedIn profile.

[00:28:15] I love it. One last question and I'm curious for HireVue itself,

[00:28:21] is there anything exciting coming up in the next year? Like what is in the roadmap for

[00:28:25] HireVue as far as new products, new innovations?

[00:28:30] Now I'm going to say something I'm not supposed to commit to yet. Let me think.

[00:28:35] That's fine. No, no, no. Don't worry about it.

[00:28:38] My product leaders like weed. I can't date on that. No, kind of what I was talking about,

[00:28:44] we are thinking a lot about talent acquisition and talent management and how those things overlap.

[00:28:51] And as a person who's not from this space, I think a lot about why are those things separate

[00:28:55] or why do you lose a candidate? Like they apply for a job and then if they didn't get that

[00:29:01] job, they're just gone. A lot of companies will say, hey, we want to keep you in mind if

[00:29:05] something else comes up and they have no process to do that. So in our system,

[00:29:10] we're starting to think more about, hey, we have this candidate. They just applied for a job at

[00:29:15] your company. You know they're looking for a job. They want to work there. And here you are

[00:29:19] posting a really similar role or even basically the same role again two weeks later. They've

[00:29:26] already taken an assessment. Here they are and they come back in. So I think that's

[00:29:31] just a no brainer. We're thinking a lot about that and it ties into my future of hiring idea

[00:29:37] around being resurfaced for opportunities rather than applying for hundreds of jobs

[00:29:42] and getting denied every time. Perfect. We're going to get your chief product officer on the

[00:29:48] show really soon and we'll question him on that. We have a new one and she starts tomorrow.

[00:29:52] So we just throw her right in. Throw her right in. Perfect. Well, I really appreciate you coming

[00:29:58] on the show. If anyone wants to find out more about Lindsay, what's the best way to get a hold

[00:30:03] of you? Yeah, you can find me on LinkedIn. So Lindsay Zulaga. And for HireVue, obviously

[00:30:11] hirevue.com is probably the best way to find out more about HireVue. Perfect. Thank you so

[00:30:17] much. We really appreciate this amazing information. Can't wait to have you on the show again.

[00:30:22] Nice to meet you, Lindsay. Thank you.

[00:30:34] Shelley, let's face it, texting candidates is the easiest way to hire quicker today,

[00:30:40] but your cell phone doesn't connect to your ATS. You're sharing your personal number with

[00:30:44] strangers. That's pretty scary, right, Shelley? And it's not even legally compliant.

[00:30:50] This is where our friends at Rectex come in. They've created simple yet powerful text recruiting

[00:30:55] software that works with your ATS. Plus, it's designed by recruiters for recruiters,

[00:31:01] so you know it works. To learn more and book a demo, visit www.rectxt.com,

[00:31:11] mention the recruitment flex, and get 10% off annual plans.

[00:31:14] Do you love news about LinkedIn, Indeed, Google, and just about every other recruitment tech

[00:31:20] company out there? Hell yeah. I'm Chad. I'm Cheese. We're the Chad and Cheese Podcast.

[00:31:26] All the latest recruiting news and insights are on our show. Dripping in snark and attitude.

[00:31:32] Subscribe today wherever you listen to your podcasts. We out.