Payscale's Sara Hillenmeyer on Data Science and AI, Here and Now

[00:00:00] Welcome to PeopleTech, the podcast of WorkforceAI.News. I'm Mark Pfeffer. Today, my guest is Sara Hillenmeyer, the Senior Director of Data Science at Payscale.

[00:00:22] It seems like we're all talking AI all the time, but Sara is one of the people who turns the technology into actual useful applications. In a lot of ways, that makes her unique. So we're going to talk about AI in the context of what's real. How could AI realistically impact jobs? How can developers keep up with the changes that are coming so fast and furious? And what's more stressful, keeping up with the market or keeping up with the technology?

[00:00:50] All of that on this edition of PeopleTech. Hey, Sara, welcome. A long-time listener, first-time guest. And we're glad to have you. Let's start by sort of stepping a little bit back. And I wonder if you could put AI in perspective for me. If you're involved in any kind of business technology now, especially HR, it's really the conversation.

[00:01:16] Pretty much everything is being wrapped into the term AI. And I wonder if you could kind of help us sift through, you know, what's real and what's not. Could you put AI in perspective for me? There's so much talk about it now everywhere in the technology world, the HR tech world. But it's not always clear what's really true and what's just sort of talk. Can you help us figure that out? Sure.

[00:01:45] So as you know, Mark, you've been around a while. Everything goes through a hype cycle, right? The new technology gets developed and folks get really, really excited about what it could possibly do and what it could possibly be. And then as those applications come to life, that hype gets grounded a little bit in reality. And we find the tools that really work and solve real problems and stick and then iterate on those tools and make them better and better over time.

[00:02:14] The AI hype is funny to me because I've been doing this for a long time, since before it was cool. And we used to call it statistical learning. And then we called it machine learning. And now we call it artificial intelligence. And really the math and the guts of all of this is fairly old. We didn't have computers that could run it fast enough 20 years ago. But the way that it works is fairly well studied in academia and fairly well developed.

[00:02:42] And we were just waiting, really, for the computers to be able to catch up and do the computations fast enough that we could build the kinds of applications that you're seeing today. The large language model explosion started in 2017. With a paper called Attention is All You Need. And that really set the groundwork for how artificial intelligence can be used to generate language and to process language and to translate language.

[00:03:09] And so the boom that we're seeing right now is coming out of that paper and that invention that's really been transformational in how we use AI on text and generation. But we've seen other booms like this over the last decades. There was an image processing boom before the language boom, which you might remember if you use Google Photos or even Facebook.

[00:03:38] They were recognizing who is in the picture well before ChatGPT came out. Those applications were already there. So as a practitioner in the field, I'm very, very excited about this rapid explosion in applications. But the science itself has been plodding along pretty steadily for a long time, making gain after gain after gain.

[00:04:04] And this one just has so many different areas where good applications could be built to help solve real world problems. But it's really been a powerful cultural shift and not just us math nerds sitting at conferences getting excited about incremental improvements. Which was kind of my was going to be my comment. I mean, it sounds like this is a technical step.

[00:04:31] But, you know, people are assigning it all sorts of things, characteristics like, you know, it's about to become conscious or something like that. Is that is is that just a part of the hype cycle? I think so. I don't think we're anywhere near what's called artificial general intelligence or anything close to sentience.

[00:04:53] I do think we are at a cultural society transition point of how do we how do we leverage these tools in our world and what makes sense? You know, I was I was talking to somebody recently and they're like, well, if if AI is writing all of the code, how do we get senior software engineers? Because we never need to hire junior software engineers to to write code. How do we get people that have that expertise?

[00:05:20] Or even in the comp world, there's a little bit of fear of like, well, if if AI is doing the job matching, what do our comp analysts do and how do they learn the field that we need them to learn and build that that knowledge? And I look at them with a little bit of a cross eyes, I think, because, you know, I'm a math person and I was born well after the calculator was invented. And somebody taught me long division and I learned it.

[00:05:48] And then I didn't do it again by myself for a very long time until my daughter was learning long division and I practiced again. Once I learned the tool, then I once I learned the math, then I adopted the tool usage to make that part of doing math easier. And now, you know, math hasn't died. Mathematicians are still alive and well. We're just doing more and more complex things, building on that tool usage. That's become the second.

[00:06:17] Second brain for us to not have to go do all of that long division. And I expect the same thing to happen in all of these other domains where AI tool usage feels a little scary. Like right now, we'll have to learn how to teach folks the skills they need to be a successful user of tools. And that may look a little different than doing the same thing that the tool does over and over and over.

[00:06:43] But I expect that the kind of work we'll be able to do with good AI power tools is quite a bit better and more sophisticated than the work that we have to do in place of those AI tools. Have you heard of this company Mechanize? I think where, you know, their goal, they say, is to eliminate work.

[00:07:06] And they don't really get into what impact that's going to have on society or the economy. They just say it'll be better. They're going to eliminate work. You know, it touches a question that's been going on and that's been discussed a lot. And that's the impact of AI on jobs. I mean, I think I personally think these guys are stretching.

[00:07:31] But as companies keep talking about efficiency and as, you know, more applications are developed that make jobs, make work more efficient, you know, is it really leading to just an evolution of jobs into more complex jobs?

[00:07:54] Or are we going to look at a lot of job elimination and people out of work because there's just not enough work for them to do? Two things will happen. First off, our labor pool is shrinking, at least in the U.S. We are at peak 65 right now. We have more 65-year-olds than we've ever had before and then we'll have again.

[00:08:18] And the pool of folks that are coming in, being born, is not as large as the pool that are retiring at peak 65. So there are definitely labor market demographics that indicate to me that being able to run a business with fewer employees is going to be a big advantage in the coming years. That said, I don't think jobs are going away. I think people like to work.

[00:08:46] And the other kinds of technology that we've seen that have eliminated workload for us, we still, we just find other work to do. People thought that the internet was going to kill libraries and that no librarians were ever going to be needed. And now you can go to the internet, you can use the internet at the library. That has shifted in our society. Or we used to spend a bunch of time washing our clothes by hand and now we put them in the washing machine. But then we don't sit around and read the rest of the time.

[00:09:17] I think we as people are just always striving to be useful and to be productive unless we're spending three weeks on a cruise. We really do have this drive to contribute to society and to build things and make things and serve each other in this way. So I think that there's basically zero chance, in my opinion, that all jobs go away and AI does all the work for us and we set my ties by the pool. Because I don't think that's what we want.

[00:09:45] And ultimately, we'll find something to do. And my guess is that something is a more higher level work, a more strategic layer of work that is fueled and powered by the kinds of tools that are coming. I think jobs will change. I think there will be bumps in the road over the next several years as that shifts. We've seen that before. We've seen that with manufacturing and with anything that can be done by a robot.

[00:10:15] You know, those jobs have changed and shifted and those people are doing other things now than they used to be doing. And I expect that same kind of shift. I think data analysts are probably not a job that will be here in 20 years. I think using Excel really, really well is not going to be a skill that's super valuable in 20 years. I think writing basic code is not going to be a skill that's very valuable in 20 years or even five.

[00:10:44] But I don't think that we will let the AIs do all the work while we... The barf. Breaking news, acquisitions, research, funding. It's the week that was. It's a wonderful way to catch up on the week that was the week prior. Kind of a John Oliver-esque type of show. Ryan and I pitch each other stories. We laugh. It's fun. You can watch it on YouTube, but you can also subscribe to the podcast. Give it a listen. Give it a look. Thanks.

[00:11:14] Well, we sit idly by. Ah, da ist meine Prime-Bestellung. Was? Jetzt schon? Nur eine Sache begeistert so sehr wie die schnelle Lieferung mit Prime. Und zwar das tolle Unterhaltungsangebot von Prime. Was schauen wir uns an? Heads of State. Die neue Action-Komödie mit John Cena, Idris Elba und Priyanka Chopra-Jonas. Sieht mega aus. Schnelle, kostenlose Lieferung, geniale Unterhaltung und mehr gibt's bei Prime. Für nur 8,99 Euro pro Monat werde jetzt Prime-Mitglied. Die Inhalte können Werbung enthalten.

[00:11:43] Fortlaufende Mitgliedschaft. Mehr auf Amazon.de slash Prime. Visionary Voices. Bold Breakthroughs. This is Shally's Alley. Where top recruiting minds spill their best kept secrets live every Friday at 1 p.m. Eastern. Raw unfiltered conversations on sourcing, hiring and the future of talent. Where we ask the tough questions no one else will. Tune in and level up. As somebody who just got off a three-week cruise, I just want to, you know, confirm what you said.

[00:12:14] People on three-week cruises don't spend any time at all thinking about how to better society. Or contribute or anything like that. Let me shift gears a little bit. Sure. Because I think you have the coolest job. I mean, to do product development and, you know, exploring AI for, you know, for a company like this. What is it like for you trying to keep up with the market for AI technology right now?

[00:12:42] There's two things, really, to answer that question. One is keeping up with the exponential growth in the kinds of foundational models and tools that are available to application builders to build things. And that is a deluge right now. It's always been a rapidly changing field.

[00:13:05] But there's so much happening right now that I spend a day a week reading, watching YouTube, listening to podcasts. Mostly at the, not quite at the academic paper level anymore. More at the, here's an early prototype that somebody built that they're showing off. Or here's this new protocol that folks are starting to rally behind and get used to.

[00:13:28] And I encourage both my, my whole team and our partners in product and engineering to bake in that same kind of, of protected learning time. Because it's, it is moving really, really fast. And the kinds of things we can do this year are dramatically cooler, more powerful than they were even a, even a year ago. So, so there's certainly learning to keep up with.

[00:13:53] The second piece is, is keeping up with the market demand around this hype cycle. And as, as you may have, have experienced, and I certainly have, there's a, there's a, almost a market pull for just, I just need AI. I need AI to do this. Whether that, and it's not always specific. It's not, oh, I need AI to help me solve this problem.

[00:14:18] But more of a, sometimes from the top of a business, sometimes from a fear of being left behind. Folks are like, oh, I just, I just need to, I'm a little scared. I'm behind. Like, I need AI and give me AI. And that's a really interesting spot to be in.

[00:14:36] I think the, the companies that are doing, doing really well and, and Payscale is one of them are, are solving real problems using the AI tools rather than just adding AI to things so that we can say that we, that we have it. But there are certainly a number of companies that are, their whole business model right now is we can give you AI and it doesn't necessarily solve problems, but it plays into that fear that, that folks have of being left behind in this AI market.

[00:15:07] So, so what do you think is the harder part for you? You have to keep up with the technology, but also keep up with the market expectations and, you know, what sort of the flavor of the moment of what people are looking for AI to do? Which is harder? That's a good question. I'm a learner by nature.

[00:15:31] You know, that's how I ended up in this, in this field to begin with was I just like learning, learning math, learning science, learning tech, learning computers. And part of what I drew me to it was that the problems just kept getting more and more challenging. It's not the same job every day. Once you solve a problem with a computer, you move on to the next problem because you can run that same solution over and over and you don't need to do it again.

[00:15:58] So the fact that the technology is changing is definitely a positive piece of the job for me rather than a negative one. And the fact that the market is demanding it is a positive piece too. When I started on this career journey, I very much expected to be a professor somewhere writing papers that nobody, nobody reads.

[00:16:23] And because the market is demanding so much AI, folks like me have come out of the woodwork to go, to go build it because there's, there's a market need. There are good business opportunities. There's good professional opportunities. So I'm just super excited to be in this world. I didn't expect, I didn't expect to be living, living in a hype cycle when I started this journey. And it's, it's just really fun. I don't think there's a part that's more challenging or another.

[00:16:53] I earn good talent. That's actually the, probably the hardest part. There's such a demand for talent that getting folks that have the right skills to, to come work is, is probably the hardest piece. You know, and you know, everybody talks about that, the talent shortage. What are the kind of people you're referring to here? Is there a specific, you know, data scientist or, or something like that?

[00:17:20] Or are you seeing, you know, increased demand for AI skills across the whole spectrum? We are. We are. And I can, certainly from my experience as a, as a people leader and hiring manager, the kinds of roles I'm hiring are data scientists and machine learning engineers and folks who have a little bit of both of those, those flavors.

[00:17:40] But across the data that we see at Pascale, we're seeing a huge demand for AI plus job, whether it's AI product managers or AI strategists, or even in manufacturing and other sectors where you wouldn't expect necessarily AI skills to be super important.

[00:18:01] We're seeing jobs pop up with those kinds of requirements as companies look to get, to, to make use of the technology, certainly, and to get on, on the bandwagon and figure out what that could mean for them in terms of the efficiency of their work or the efficacy of their work too. Sarah, thank you so much. It was really, really fun talking to you. It was really great. And I hope you'll come back so we can do it again. Yeah. Great to meet you, Mark. Thank you. Thank you.

[00:18:31] My guest today has been Sarah Hillenmayer, the Senior Director of Data Science at Pascale. And this has been People Tech, the podcast of WorkforceAI.news. We're a part of the Work Defined Podcast Network. Find them at www.wrkdefined.com.

[00:19:01] And to keep up with AI technology and HR, subscribe to WorkforceAI today. We're the most trusted source of news in the HR tech industry. Find us at www.wrkdefined.news. I'm Mark Pfeffer. I get it. I get it. The podcast just isn't enough. That's all right. Head over to your favorite social app.

[00:19:31] Search up Work Defined. W-R-K Defined. And connect with us.