Stacia Garr - HR Tech 2024 - Can AI Deliver on Its Promises for HR?

[00:00:00] The world of business is more complex than ever. The world of human resources and compensation is also getting more complex. Welcome to the HR Data Labs podcast, your direct source for the latest trends from experts inside and outside the world of human resources. Listen as we explore the impact that compensation strategy, data, and people analytics can have on your organization.

[00:00:25] This podcast is sponsored by Salary.com, your source for data, technology, and consulting for compensation and beyond. Now, here are your hosts, David Turetsky and Dwight Brown.

[00:00:38] Hello and welcome to the HR Data Labs podcast. I'm your host, David Turetsky, and we are live at the 2024 HR Technology Conference in beautiful Mandalay Bay in Las Vegas, Nevada. I am here with Stacia Garr. Stacia, how are you?

[00:00:51] I'm doing great. How are you, David?

[00:00:53] I am awesome because we get to talk about HR technology.

[00:00:58] Amazing.

[00:00:58] Yes.

[00:00:59] What we're all here for.

[00:01:00] Oh my gosh, this is so exciting. The vibe at HR Tech this year is kind of electric.

[00:01:05] Yeah.

[00:01:06] I mean, it's not because there are downed wires that we're trying to jump over. Actually, there's a lot of really good conversations happening.

[00:01:11] Yeah. And it feels like it was pre-pandemic. I was talking to somebody about that yesterday. The energy seems to be back, the number of vendors. I mean, I know we kind of had that bump, but it all still kind of felt different.

[00:01:25] Yes.

[00:01:25] But now it feels, I don't know, like its own real energy that we've been missing.

[00:01:31] It feels better.

[00:01:32] I think so.

[00:01:33] Yeah.

[00:01:34] Yeah.

[00:01:34] And if you look around, there are so many exhibitors here.

[00:01:37] So many and so many big booths.

[00:01:39] It's crazy.

[00:01:40] Which is really interesting.

[00:01:41] Yes.

[00:01:42] But before we get to our topic for today.

[00:01:45] Yes.

[00:01:45] What's one fun thing that no one knows about Stacia?

[00:01:48] One fun thing.

[00:01:50] Um, I don't know if this is fun, but my first job was when I was 11 shoveling horse poop.

[00:01:57] And that was my introduction to the world of work.

[00:02:00] You know, I actually think that a lot of people believe that their first job was shoveling poop, but not necessarily.

[00:02:07] Literally.

[00:02:08] Figuratively.

[00:02:09] Well, that's how I paid for my horses.

[00:02:11] Oh, that's awesome.

[00:02:12] So I had horses and, um, my parents were like, well, if you want this, you got to work for it.

[00:02:18] Wow.

[00:02:18] So every Sunday morning for six hours in the Arizona heat.

[00:02:22] Wow.

[00:02:23] We shoveled, I shoveled the poop.

[00:02:26] My dad drove the tractor.

[00:02:28] Uh, you got to talk to him about that balance of power.

[00:02:31] Uh, come on, dad, at least trade off once in a while.

[00:02:35] Let me tell you, by the time I was 13, I knew how to drive a tractor.

[00:02:38] Of course you did.

[00:02:39] Yeah.

[00:02:39] Yeah.

[00:02:40] Which I'm sure helped you out later in life with how to drive a car and other things.

[00:02:43] I, you know, I can drive, I learned on a stick shift.

[00:02:47] Did you learn a stick shift?

[00:02:48] I actually didn't.

[00:02:49] But then when I lived in London for Morgan Stanley, I actually learned on the hills of,

[00:02:53] uh, Belsize park in where Hampstead is.

[00:02:56] And yeah, I got a lot of horns blowing at me.

[00:02:59] Yeah.

[00:03:00] Well, but I can, I'm sure as you can now drive a stick shift in San Francisco.

[00:03:05] So, you know, that's, that's a level of accomplishment.

[00:03:07] Oh my God.

[00:03:08] Lombard street is a particular horror.

[00:03:11] Um, and everybody knows Lombard.

[00:03:12] That's the snakey street.

[00:03:13] Yep.

[00:03:14] But that's actually not the scariest ones.

[00:03:15] The scariest ones are where you actually have to be in gear, not burning out the clutch

[00:03:20] or, or having your handbrake on at like the top with a stop sign, like almost being 30 or

[00:03:27] 40 degree grade.

[00:03:28] Yeah.

[00:03:29] And then somebody is like four inches behind you.

[00:03:31] Cause they have no, cause they're in an automatic and they don't understand the concept of a

[00:03:35] stick shift.

[00:03:35] Yeah.

[00:03:37] I've done that too.

[00:03:38] Yeah.

[00:03:38] Like an idiot.

[00:03:39] I drove a stick in Sonoma and then me and my, my first wife went to, um, went through San

[00:03:46] Francisco to go home.

[00:03:48] And yeah.

[00:03:49] Yeah.

[00:03:49] Yeah.

[00:03:50] We'll never forget those things.

[00:03:52] I was going to say, you're not going to do that one again.

[00:03:53] No, no, probably not.

[00:03:54] No.

[00:03:55] Even though it was fun.

[00:03:56] No, no.

[00:03:56] Yeah.

[00:03:57] But, but also it was a rental car.

[00:03:58] So that clutch, you know, exactly.

[00:04:01] So no, that is fun.

[00:04:03] Yeah.

[00:04:03] And I will never forget that one.

[00:04:05] Yeah.

[00:04:05] You know, it's early formative experiences.

[00:04:09] Yes.

[00:04:09] Teach you how to work hard.

[00:04:10] That's yes.

[00:04:11] And work very efficiently.

[00:04:14] Yes.

[00:04:14] And get up very early.

[00:04:16] Yeah.

[00:04:16] In June and July.

[00:04:17] Yes.

[00:04:18] Yes.

[00:04:19] Where I'm sure the flies are very happy to see you.

[00:04:24] Yes.

[00:04:24] Yeah.

[00:04:25] Yeah.

[00:04:33] Now let's transition to our topic, which we're going to touch on HR technology and let's

[00:04:40] touch on how HR tech has evolved or what you think of like what we typically have done in

[00:04:46] the past is say what's different this year from last year.

[00:04:48] But there have been so many evolutions.

[00:04:50] Yeah.

[00:04:51] In HR tech.

[00:04:51] Yeah.

[00:04:52] I kind of sigh because it's like, I was having this conversation over dinner last night because,

[00:04:57] you know, I think one of the, one of the cycles, and maybe this is what hasn't changed is the

[00:05:01] cycle, but the cycle of, you know, the, the companies, the big companies kind of getting

[00:05:07] bigger and being monolith.

[00:05:08] And, you know, you remember talent management suites and you know, all that stuff.

[00:05:13] And then, and then you kind of go through the, um, innovation explosion and you get all

[00:05:17] these smaller vendors who are more agile doing more interesting, you know, point solutions,

[00:05:21] blah, blah, blah.

[00:05:21] And then they get acquired.

[00:05:22] Exactly.

[00:05:23] And we're, we're kind of in that, you know, acquiring part of the phase.

[00:05:27] Right.

[00:05:27] Which I think is, is interesting, but it is just this cycle that we see, you know, I've,

[00:05:32] I've been in the space now 20 years.

[00:05:34] You've been in it at least that long.

[00:05:38] Yeah.

[00:05:38] The great beard should give it away, but I wasn't going to say the exact time.

[00:05:43] Um, but, uh, but so I think that that's one thing that's consistent, but, but what is maybe

[00:05:48] different is, um, you know, a couple of things.

[00:05:52] One is that some, many of the acquisitions that are happening now, I think, and I think,

[00:05:57] and will happen across next year, I think are less about the technology and more about

[00:06:01] the data that those companies have in the access to that because right now is, you know,

[00:06:07] a bit for, for a while, really, it's been kind of this data gold rush, but now with Gen

[00:06:12] AI, it is even more so.

[00:06:14] And so I think some of the acquisition will be done because of that desire for the data.

[00:06:19] Is it finding new sources or creating new types of data, or is it being able to leverage

[00:06:26] the technologies to acquire data in a different way?

[00:06:30] Kind of like, think about it, like screen scraping is always, has been around for a long

[00:06:33] time, but is it, is it about being able to make that mousetrap just a little bit different,

[00:06:39] a little bit better so they can find different sources of data?

[00:06:42] What's the new data trend?

[00:06:44] I mean, maybe, but I think that more, more specifically, you know, to, to power these large

[00:06:50] language models, you just need an immense amount of data.

[00:06:53] And the, for our space, the, the public data sets are not adequate to do a lot of this

[00:06:59] final refinement.

[00:07:00] And so particularly for these bigger vendors that have deeper pockets to, to do this, they

[00:07:06] are looking to how to develop the best models or to train and refine their models.

[00:07:10] And I think that they're looking to just simply bring in more data to do that training and

[00:07:16] refinement.

[00:07:16] I think it will lead in, you know, in some instances to better mousetraps.

[00:07:21] Like you take the example of workday and hired score, right?

[00:07:25] Like that was, I think to, to build a better mousetrap and hired scores is all this amazing

[00:07:30] orchestration capability that can then get integrated into the rest of workdays.

[00:07:34] So that, that I would say is more of a mousetrap acquisition.

[00:07:36] Yeah.

[00:07:37] But I think that there's going to be more that are looking particularly in the next year

[00:07:41] for the data itself.

[00:07:44] One of the things that fascinates me about this is that there's a lot of private data

[00:07:48] that companies want access to.

[00:07:51] And there are some companies that have it like the HCM or the payroll companies, they have access

[00:07:55] to all that data.

[00:07:57] And we all know ADP.

[00:07:59] Right.

[00:07:59] Exactly.

[00:08:00] Well, we all know ADP, how they pay one in five, one in six people.

[00:08:03] Yep.

[00:08:04] And, you know, they do a lot of work on mining that information.

[00:08:08] Yep.

[00:08:08] So do you see it like these companies are being able to invest in their client organizations

[00:08:15] and be able to say, hey, listen, we're giving you a lot of functionality.

[00:08:18] And now we're going to leverage the data that you have in our databases for other purposes,

[00:08:23] like benchmarking and other things.

[00:08:25] I mean, so much of it depends on the contract.

[00:08:29] Right.

[00:08:29] And the data culture that exists.

[00:08:33] Right.

[00:08:34] And then also, honestly, on where they're operating.

[00:08:37] Right.

[00:08:37] So so much of the EU's legislation limits what you can do also limits, you know, things like forget ability

[00:08:47] and all the rest of that.

[00:08:48] And that's, you know, that's just GDPR.

[00:08:50] That's not the EU AI Act.

[00:08:51] Exactly.

[00:08:52] But so I think that there's there is going to be a resetting probably of norms around what the data can be used for.

[00:09:01] So in general, what I'm tending to see is is folks saying, hey, we're not going to necessarily train the overall models with a customer's data.

[00:09:11] But we will train, obviously, their customer specific data and refine it in that sort of way.

[00:09:16] And and I think that some of what gets learned will necessarily pass on.

[00:09:21] But the data itself in any sort of recognizable form will will not.

[00:09:25] So it's in a walled garden.

[00:09:27] Yes.

[00:09:27] For the client.

[00:09:28] It's not being leveraged beyond the walled garden.

[00:09:30] I mean, obviously, there are some companies that are leveraging it.

[00:09:34] There are some.

[00:09:34] But to your point, they're not actually leveraging it to build other data sets.

[00:09:39] They're potentially leveraging it to create other products that can be that can be utilized within the context of the client.

[00:09:46] Exactly. Exactly.

[00:09:47] So it's not it's not benchmarking.

[00:09:49] Right.

[00:09:49] It's not we're taking.

[00:09:50] But we may be using the data to train a model that will then have broader capabilities.

[00:09:58] But it's not like the data is being shared and then kept.

[00:10:00] Right.

[00:10:01] In a benchmarking type of way.

[00:10:02] Yeah.

[00:10:03] And this kind of this gives me some kind of pause because in the past, and I'm talking about from my old times beyond the 20 years that you were talking about before.

[00:10:13] We use benchmarking for everything.

[00:10:15] Yeah.

[00:10:15] You know, when when you went to your CFO or your CEO and you said, this is the spend I want to, you know, have for next year for my merit budget.

[00:10:23] You had to have, you know, here's what all the other companies are doing in our competitive space.

[00:10:27] And here's what SHRM says.

[00:10:29] Here's what World at Work says.

[00:10:30] Here's what all these others say.

[00:10:32] So do you see that benchmarking as a as a function then?

[00:10:36] And I'm not just talking about with comp.

[00:10:38] I'm talking about more generally with HR.

[00:10:40] Do we actually need to prove it anymore?

[00:10:42] Or are the systems that we're talking about with the walled garden?

[00:10:46] Is it enough that the data is being mined for us that the systems are telling us this is what we need to do?

[00:10:53] Do you know what I mean?

[00:10:54] Or have we stopped benchmarking?

[00:10:55] So we we haven't.

[00:10:57] But when our clients are coming to us, a lot of time what they're saying is, is if if they are going down the benchmarking path, we are doing it because somebody else is asking us to justify not because it used to be like kind of like you said in the old days, people would be like, well, I'm not sure how much we actually should be spending per employee on whatever.

[00:11:18] Right.

[00:11:19] And so I'm going to I want you to tell me what others are doing now.

[00:11:24] What it tends to be is somebody coming to us and saying, my CEO, my CFO wants to want some sort of benchmark.

[00:11:34] I don't think it's actually going to impact what we're doing, but they want it.

[00:11:37] And the reason I think that we have moved away from the benchmarking is because we have so much better analytic capability that we can do the work.

[00:11:48] We can do the insight for that individual client to say, actually, at salary dot com, these are the things that drive better performance or drive better engagement.

[00:11:59] And therefore, you should do X, Y and Z from a spend perspective, et cetera.

[00:12:04] But that's different at I'm going to say workday because I'm literally looking at a big workday side.

[00:12:08] But that's different at workday because their drivers are A, B and C and therefore they should do blah, blah, blah.

[00:12:14] So we have a level of insight in detail and specificity that we didn't have 20 years ago.

[00:12:21] And I think that is why we're seeing the move away from as much of a focus on benchmarking.

[00:12:27] The other thing I'll say as a researcher is benchmarking is hard.

[00:12:31] Yes. Right. It's hard to do.

[00:12:33] It's hard to be get accurate numbers.

[00:12:35] It's hard to be confident in those numbers.

[00:12:37] And then people don't want to pay that much.

[00:12:39] Well, they definitely don't want to pay that much.

[00:12:41] That's certainly true.

[00:12:42] And we've certainly seen that, you know, in the compensation space.

[00:12:47] Compensation has always been built on benchmarking.

[00:12:50] I mean, obviously, I mean, we don't really do anything.

[00:12:53] Compensation is a little different than some of the other spaces.

[00:12:54] Oh, sure. Yeah, sure.

[00:12:55] But like even when in the old days when we had Saratoga, remember?

[00:13:00] And we used to go to Saratoga for what are the ratios?

[00:13:03] What are the HR ratios or what are the ratios of this group versus that group?

[00:13:07] You know, that was something that we, especially for growing organizations,

[00:13:11] we use that to understand where should we be right now?

[00:13:14] How can we grow and are we growing in the right way?

[00:13:17] And our leadership really cared about being able to grow like best practice companies grow.

[00:13:23] And so now I see what you're saying.

[00:13:26] You're saying, well, we have better analytics about what drives the company.

[00:13:30] So instead of going toward, you know, a benchmark, which is a lot of art versus science.

[00:13:36] Yeah.

[00:13:37] Let's actually just use what the analytics or what consultants who are using analytics know about us.

[00:13:42] Yeah. Yeah. Let's just use the science on us.

[00:13:44] Right.

[00:13:45] And that's not to say, I think, that benchmarks aren't useful.

[00:13:49] Right.

[00:13:49] Like, I think that there are times when say, okay, let's just gut check this.

[00:13:55] Right. Like, okay, this is what's driving this.

[00:13:57] But are we wildly out of whack with whatever those are doing or the like?

[00:14:01] But then I think there are other times when benchmarks are actually not helpful.

[00:14:05] So my primary example for this is on DEIB.

[00:14:10] You know, the average organization is not actually very good at it.

[00:14:14] And so benchmarking yourself on that isn't going to evolve the whole space.

[00:14:19] It's just going to kind of put you at not very good.

[00:14:21] And right now, that's even thinking about the word benchmark in the context of DEIB gets you in trouble.

[00:14:28] Yes.

[00:14:28] With the political winds.

[00:14:30] Yes.

[00:14:31] It shouldn't be.

[00:14:32] But, you know, that's what's happening.

[00:14:34] Yeah.

[00:14:34] The culture wars have kind of chilled a lot of the DEIB conversation.

[00:14:40] We are, as Red Thread, are trying to continue to have it where appropriate.

[00:14:47] Because, one, our clients are telling us that they haven't stopped doing it.

[00:14:52] Right.

[00:14:52] They are just not talking about it.

[00:14:54] Right.

[00:14:55] And, two, for us, there isn't the same.

[00:15:00] Let me just put it this way.

[00:15:01] We are a really good almost broker of information where people who don't want to talk about something but know that it still needs to be talked about can talk to us.

[00:15:08] Right.

[00:15:08] And then we talk about it kind of on their behalf.

[00:15:11] Right.

[00:15:11] Which is wonderful.

[00:15:13] Yeah.

[00:15:13] Thank you.

[00:15:14] Doing our best.

[00:15:15] Thank you.

[00:15:15] Well, and I think that's what we need to do right now.

[00:15:19] We need to do the best we can, given the situations that a lot of companies find themselves in.

[00:15:25] And a lot of employees find themselves in, too.

[00:15:27] Especially if you've been part of some of these employee resource groups that are now a little bit worried about whether they're going to survive or not.

[00:15:36] Yeah.

[00:15:36] Yeah.

[00:15:37] It's a tough time.

[00:15:51] So let's transition a little bit and go to one of the more fun conversations that I've been having,

[00:15:56] which is more about that interesting two-letter acronym that is on every single sign throughout the entirety of HR technology.

[00:16:07] And it's not HR.

[00:16:09] It's artificial intelligence.

[00:16:11] A little earlier in the alphabet.

[00:16:12] Yeah.

[00:16:13] A little.

[00:16:14] Well, yeah.

[00:16:17] So beyond the hype cycle.

[00:16:20] Yeah.

[00:16:21] What is going on with this?

[00:16:23] And really, where is it landing in the world of HR today?

[00:16:27] Because we've been told a lot people are scared in the world of HR.

[00:16:33] But beyond the hype, where is it?

[00:16:36] And where is it landing right now?

[00:16:37] Yeah.

[00:16:37] It's interesting.

[00:16:38] I have felt like vendors and practitioners were kind of talking past each other on this.

[00:16:44] And then we just released a study last week that shows that they are talking past each other.

[00:16:50] Wow.

[00:16:50] So what we found was that we asked our People Analytics Tech vendors, we said,

[00:16:55] how many of you have AI, ML in your solution?

[00:17:00] Any guesses?

[00:17:01] A hundred percent.

[00:17:02] Very close.

[00:17:03] Ninety.

[00:17:04] I was kind of shocked there were 10% who said it didn't.

[00:17:08] Really?

[00:17:08] That's what they tell us.

[00:17:10] Wow.

[00:17:10] Anyway, and so then we asked these same vendors' customers, not other random customers, these vendors' customers.

[00:17:18] Right.

[00:17:18] What does your solution have AI, ML in it?

[00:17:23] What percent do you think said yes?

[00:17:26] 50%.

[00:17:27] Close.

[00:17:28] 40.

[00:17:28] Wow.

[00:17:29] 40%.

[00:17:30] Wow.

[00:17:30] So they're not doing the job of being able to articulate what these things are, what the functions are.

[00:17:37] So that's what I thought.

[00:17:38] Right?

[00:17:38] So here's where it got pretty interesting.

[00:17:40] So we then, I said, what, like you, I was like, what the heck is happening here?

[00:17:45] Right?

[00:17:45] And so we went and we looked at the data by contract tenure.

[00:17:50] So we looked at pre-deployment, zero to six months, six to 12 months, 12 to 24, 24 and above.

[00:17:57] And what we found was that in pre-deployment in zero to six months, remember that number was at 40%, it jumped up to 60%.

[00:18:05] Right.

[00:18:06] And then it declined.

[00:18:07] Of course.

[00:18:08] And so we thought, my thinking was, okay, well this means that in the sales process and the deployment process, they're doing a really good job of articulating what they have.

[00:18:17] But folks who are later in the contract don't have that.

[00:18:22] Right.

[00:18:22] But here's where it gets even more interesting.

[00:18:24] This is why you always talk to practitioners.

[00:18:27] Is I brought this, we have a people analytics advisory board and I brought this to the folks and I said, you know, what do you guys think?

[00:18:32] Here's my hypothesis.

[00:18:34] And they're basically like, Stacia, you're so cute.

[00:18:37] That is not what's happening.

[00:18:39] Really?

[00:18:39] Yeah.

[00:18:40] They said, what's happening is once we get, we're sold something and once we get past deployment, we find that it can't do what we were told it would do.

[00:18:49] And so it basically is like, it doesn't actually exist.

[00:18:52] So the numbers went down for those later down into the low thirties and then slowly climbed up back actually to the 39%.

[00:19:01] It was at the 24 months and above.

[00:19:04] That's disappointing by the way.

[00:19:07] Yes, it is.

[00:19:08] But again, you and I have been in this space a long time.

[00:19:12] This is brand new tech.

[00:19:13] No one really like anyone here in this entire space who tells you that they really know what they're doing with AI is not being true.

[00:19:22] It's like, are you a princess bride person?

[00:19:25] Oh my God.

[00:19:25] Yeah.

[00:19:26] Right.

[00:19:26] So when, when Wesley says life is pain, anyone who tries to tell you differently is selling you something.

[00:19:34] Right.

[00:19:35] Yeah, exactly.

[00:19:36] It's the same thing with AI.

[00:19:38] Right.

[00:19:38] But that's both the frustrating because everyone here is talking about it.

[00:19:43] Right.

[00:19:43] And it's frustrating to talk about something that people are not totally clear on, but it's also the opportunity.

[00:19:48] Right.

[00:19:49] I'm, I'm, despite these numbers, I'll also tell you in our data, um, the presence of AI did not have any statistical relationship with renewal or a sense of value.

[00:19:59] What about satisfaction?

[00:20:01] Yeah.

[00:20:01] Yeah.

[00:20:02] No, there's just no relationship.

[00:20:03] Yeah.

[00:20:03] So, um, so what it says to me is that the AI is not yet delivering the value that, that people are hoping that it will when they're, when they're selling it or when they're buying it.

[00:20:13] But I think it will.

[00:20:15] We're just not at that point yet.

[00:20:17] But, but even the context of the conversation we're having, you use the, the words or the acronyms ML and AI in the same sentence.

[00:20:24] And yeah, machine learning is kind of a flavor, but we all know that most people are thinking about JetGPT or Gemini or, or, or some kind of like co-pilot because those are the more commercial aspects, you know, which are Gen AI.

[00:20:39] And as you said, but what are they expecting?

[00:20:43] I mean, cause many companies will use whether it's ML or some kind of statistical algorithm in what they're doing, you know.

[00:20:51] And have been for years.

[00:20:53] For decades.

[00:20:53] I mean.

[00:20:54] Decades.

[00:20:54] Decades.

[00:20:55] Yeah.

[00:20:56] We built a predictive model at ADP when I was at ADP.

[00:21:00] We did build a predictive model in like 2015, 2016.

[00:21:03] Mm-hmm.

[00:21:04] Yeah.

[00:21:04] That's almost 10 years ago.

[00:21:06] Crazy.

[00:21:07] But, but, but, but, you know.

[00:21:09] Yeah.

[00:21:10] No.

[00:21:10] So, I mean, the, the thing that, as I looked at it, I remember people analytics tech people are amongst the most sophisticated of our HR buyers.

[00:21:20] Sure.

[00:21:21] Because they're, they're data people.

[00:21:22] Right.

[00:21:22] They're usually tech people, that kind of thing.

[00:21:25] What it says to me is that a fair number of them probably know that they've got AI in there because AI has been in there for a long time.

[00:21:34] Sure.

[00:21:34] Right.

[00:21:34] It's the, the gen AI, which is not yet delivering what, what folks hope.

[00:21:39] You asked what, what do they expect?

[00:21:41] Right.

[00:21:42] Right.

[00:21:42] Well, I mean, I think what they expect is what everybody here is selling them, which is that it's going to, you know, it's going to generate content where appropriate.

[00:21:50] It's going to be doing increasingly.

[00:21:53] And I know we've got a question about like what trends are we seeing, but, you know, with agents and seeing agents, you know, being able to access information in all the places.

[00:22:02] You know, right now we're seeing a lot of glorified.

[00:22:03] I mean, somebody is going to skewer me for saying this, but basically glorified chatbots.

[00:22:07] They're not, but they, but from a functionality perspective, that's what it feels like.

[00:22:11] I think that's fair though.

[00:22:12] Yeah.

[00:22:13] Okay.

[00:22:13] Why would you get skewered?

[00:22:14] I mean.

[00:22:14] Because people are going to tell me, Stacia, you do not understand the difference between this and this.

[00:22:19] And I do.

[00:22:19] It's just from a functionality perspective.

[00:22:21] I think it feels like a glorified chatbot, even though it's not.

[00:22:25] And an end user, that's what an end user would believe.

[00:22:28] Right.

[00:22:28] Yeah.

[00:22:29] They're not going to understand everything that goes underneath it.

[00:22:32] And honestly, they don't, they shouldn't have to.

[00:22:34] Right.

[00:22:34] Like why?

[00:22:35] But anyway, so I think that's a lot of what we're seeing at this very moment.

[00:22:40] And, you know, we're beginning to see some interesting use cases, particularly in the analytics and data side, because of the ability to use natural language to ask questions.

[00:22:51] You know, there's, you mentioned your former employer.

[00:22:56] Yeah.

[00:22:57] You know, they launched a new technology, which is basically, which is their new HCM.

[00:23:03] And the vision is that you can get everything you need from search, not through menus and dropdowns.

[00:23:09] And I said the vision.

[00:23:10] Those of you who are not on the podcast cannot see his face.

[00:23:15] Oh, and look at that.

[00:23:16] But, yeah, you're right.

[00:23:17] What's the future, right?

[00:23:19] And by the way, it does it.

[00:23:20] I use it.

[00:23:21] I've used that system.

[00:23:22] I work on it with clients.

[00:23:23] And you can search in that search bar and find pretty much everything, whether you're typing in a person's name, whether you're typing in analytics and it takes you to that analytics dashboard.

[00:23:32] No, no, it does do that.

[00:23:34] The reason why I'm saying that is it's a menuing system.

[00:23:37] Yes.

[00:23:37] But the vision, again, we're back to vision and what are people trying to sell and what are people wanting?

[00:23:42] The vision is that it's not.

[00:23:45] That I can say, you know, change David's salary to 15% higher than it is right now.

[00:23:53] Yes, please.

[00:23:53] And it will do that.

[00:23:55] Okay.

[00:23:56] But that's merging two different things together.

[00:24:00] Yes.

[00:24:00] And to the extent at which it knows who David is and it knows what the basis is that we're talking about, it knows the rules and all that stuff.

[00:24:09] That's awesome.

[00:24:10] Right.

[00:24:11] But if it isn't and it takes you to the page where the manager self-service is, where it starts the process, even better.

[00:24:19] That's great.

[00:24:20] It started it out.

[00:24:21] But it's not getting you all the way there, is it?

[00:24:24] No.

[00:24:25] And I think that's the point about the drop off in what people are selling and what the reality is for folks.

[00:24:31] And I am not at all picking on this technology.

[00:24:34] Sure.

[00:24:34] It's just an example.

[00:24:35] It's...

[00:24:36] Neither am I to my ADP peeps.

[00:24:37] I still love you.

[00:24:38] We're not picking on it.

[00:24:39] We're just using it as an example.

[00:24:40] No, we love you.

[00:24:41] You're loved.

[00:24:41] I'm making a heart emoji with my hand.

[00:24:45] But that's, again, it's what's the expectation versus what's the delivery.

[00:24:49] We've been dealing in AI for a very long time on our consumer side with Siri and Alexa and whatnot.

[00:24:55] And it's constantly disappointing us because it doesn't understand what we're asking or it doesn't understand the complexity of the query that we're giving it.

[00:25:02] We have to become better at writing prompts, ostensibly, right?

[00:25:08] I'm waiting for that moment when someone says to me in HR that we don't need to worry about prompts anymore.

[00:25:14] Yep.

[00:25:15] That it happens for us.

[00:25:17] Yeah.

[00:25:17] And we're not there yet.

[00:25:20] Hey, are you listening to this and thinking to yourself, man, I wish I could talk to David about this.

[00:25:25] Well, you're in luck.

[00:25:27] We have a special offer for listeners of the HR Data Labs podcast.

[00:25:31] A free half hour call with me about any of the topics we cover on the podcast or whatever is on your mind.

[00:25:37] Go to salary.com forward slash HRDL consulting to schedule your free 30 minute call today.

[00:25:46] Is 2025, if you don't mind this one, I like this one.

[00:25:50] Is 2025 the year that HR embraces and adopts what we've been talking about in the world of artificial intelligence?

[00:25:58] In some degree.

[00:26:00] In some degree.

[00:26:02] Yes.

[00:26:04] There has been 2024 was the year of experimentation.

[00:26:11] And some of those experiments, I think have gone really well.

[00:26:15] And some, the AI has actually added work.

[00:26:19] Yeah.

[00:26:21] You know, I was listening to a, I think it was a podcast that was done by folks over up at the Upwork Institute.

[00:26:31] Kelly Monahan and Gabriela Berlecua.

[00:26:34] I don't know if you follow either of them, but they're brilliant.

[00:26:36] And one of the things that they, they pointed out, it was like, it was a concept I've been trying to articulate.

[00:26:41] And then they set it so perfectly, which is this idea that old IO psychology idea of the concept of demands versus resources, right?

[00:26:52] In organizations.

[00:26:52] And so we demand things from, from employees, but we give them resources to do it.

[00:26:57] And that balance has to be right.

[00:26:59] With AI, what leaders have been thinking is that they're giving resources.

[00:27:05] And what employees have been seeing is demands.

[00:27:07] And so there has been this disconnect between the expectations and the adoption and all the rest of that.

[00:27:16] Because for employers, or for employees, excuse me, it takes time and energy to figure out how do I use this new tool and integrate it.

[00:27:24] And this tool might be a threat to my own job.

[00:27:28] Sure.

[00:27:29] Right?

[00:27:29] Yeah.

[00:27:30] On the employer side, they're like, well, I just want you to experiment and try and learn.

[00:27:35] And it's going to make your job faster and more efficient.

[00:27:39] And employees are like, yeah, but what then?

[00:27:41] So, I mean, there's this whole really interesting dynamic that's happening that is impacting adoption.

[00:27:50] Yeah.

[00:27:51] It's definitely true.

[00:27:52] And it's also impacting HR.

[00:27:54] Right.

[00:27:54] Right.

[00:27:54] And so, back to your question, which is, you know, is this the year that we're going to see this?

[00:28:00] I think we will, but it's going to continue to be limited and in low-risk areas.

[00:28:07] So, we're going to put kid gloves on.

[00:28:09] For sure.

[00:28:09] And we're still going to be dancing around it.

[00:28:11] We've been talking a little bit on the podcast this week about the concept of the AI bot that does those small, minute tasks.

[00:28:23] Yeah.

[00:28:23] And their specific purposes they fulfill, whether it's a project manager, whether it's a, well, I forget what specific terminology that was used.

[00:28:33] But they just do their small task and they actually either talk to people or other bots that are doing these small tasks.

[00:28:40] Oh, the agents.

[00:28:41] Yeah.

[00:28:42] The agents.

[00:28:42] Agentic AI.

[00:28:44] Right.

[00:28:44] Adopting those could be seen as more of a tool if the education comes with it to say, and here's how.

[00:28:52] Here's a hammer.

[00:28:54] Here's how you hold it.

[00:28:55] Here's how you use it.

[00:28:57] Yeah.

[00:28:57] Completely.

[00:28:58] I think I am very bullish on agentic AI.

[00:29:02] I think that it has incredible potential to do things that people don't want to do, to do things at scale that we haven't been able to do and to do them faster.

[00:29:17] And I think the ability to have agents that are specific to a given task and understand how to do that thing well.

[00:29:25] I just think there's a huge amount of potential here.

[00:29:29] The challenge is going to be, of course, in the orchestration of all of those agents and making that work.

[00:29:37] But I've seen, and again, these are demos.

[00:29:39] This is not live software.

[00:29:41] But I've seen some interesting demos and ideas of how this could work in ways that would really meaningfully change things.

[00:29:49] Because, honestly, I mean, who wants to be the one who's going in and, you know, handling the benefits change or handling, you know, all these minutia that has to get done.

[00:30:00] Absolutely.

[00:30:01] But, like, it's not thrilling for anybody.

[00:30:03] No.

[00:30:03] And it's repetitive.

[00:30:04] There's a ton of error built into it.

[00:30:06] Yep.

[00:30:06] And we don't really hire for those jobs anymore.

[00:30:10] Yeah.

[00:30:11] Yeah.

[00:30:11] It's interesting.

[00:30:12] I was at Oracle Cloud World a few weeks ago here in Vegas.

[00:30:18] And, you know, Larry Ellison, his keynotes are always interesting.

[00:30:24] But one of the things he was talking about, he was talking about in the context of security, of cybersecurity, but he was talking about how they are going to move to a world where the AI is doing basically all of the responses to cyber attacks.

[00:30:40] Interesting.

[00:30:40] And he said, and it's going to write more and more of the code.

[00:30:45] And he said, it's not just because we're going to write more of the code because of the speed, but that's one thing.

[00:30:50] But one of the bigger reasons is that the AI will not use code that can be left open to attacks.

[00:31:02] Like, it is more consistent in its code writing.

[00:31:05] It's neater in its code, et cetera.

[00:31:07] And, you know, it's kind of the same way to what you just said, to the same point you just mentioned, which is that if we have this technology, there will probably be fewer errors.

[00:31:16] It will be able to self-check its errors.

[00:31:18] It will have another agent that will check its work.

[00:31:20] I mean, who checks the work that goes in right now?

[00:31:23] You don't.

[00:31:23] Well, you're supposed to write unit tests.

[00:31:25] No, no, no.

[00:31:26] But, I mean, if you're actually entering the data, you're changing the benefits, right?

[00:31:29] That kind of thing.

[00:31:31] So, I think that there's a possibility that the quality of the work is higher and that it's done more effectively.

[00:31:38] So, when you're talking about this, I'm thinking about a dark mirror episode where a woman's walking into a building that was created for resettlement of people on another world.

[00:31:49] And there were these little nanobots that were created not only to create the buildings, but also to ward off attacks.

[00:31:56] And the nanobots started attacking the people because they weren't having a good day and they were trying to encourage people to be happy.

[00:32:05] And so, they would attack people when they got mad or sad.

[00:32:08] Well, that was self-fulfilling, right?

[00:32:10] And so, to me, that just smacks a little bit of the nanobots that are going to come after us.

[00:32:16] You're attacking me.

[00:32:17] No, no, I'm just trying to get some customer service while you're attacking me.

[00:32:21] So, I'm going to take you down.

[00:32:24] Sci-fi has an amazing ability to both predict and scare.

[00:32:29] And scare it does, as you mentioned before.

[00:32:32] Yeah, yeah.

[00:32:33] You know, I think there's...

[00:32:36] I'm sure you've been following this whole thing on Gen.AI about how the Gen.AI that's out there is starting to consume Gen.AI-created content.

[00:32:46] You know, it's called Eating the Tail.

[00:32:48] Yeah.

[00:32:49] And one of the things that is, I guess, kind of heartened me in this whole scare world is basically, you know, if the content that's out there is all Gen.AI or mostly Gen.AI, the thing breaks.

[00:33:06] Yes.

[00:33:07] Right?

[00:33:07] Like, we have this vision that it's going to, you know, take over...

[00:33:12] We, some people have this vision, it's going to take over the world.

[00:33:14] It can't take over the world if there's no new content to consume and to continue to train the models.

[00:33:20] Like, it's...

[00:33:22] So, the point being, that's a very real example of how the AI is not running away and attacking us or, you know, or the other things.

[00:33:31] But people still believe in the dystopian futures.

[00:33:33] They still watch these movies.

[00:33:35] They still, you know, cling to, but what if?

[00:33:39] What if it takes my job?

[00:33:40] What if it does this?

[00:33:41] What if it does that?

[00:33:42] And there's some really good science fiction out there, Stacia, that does lead us down this dark path.

[00:33:49] Yeah.

[00:33:49] And because no one really understands what AI really is, because it's a bunch of different things, right?

[00:33:56] That they all just say, well, it's the AI.

[00:33:58] Right.

[00:33:59] Yeah, no.

[00:33:59] No, no.

[00:34:00] It's...

[00:34:00] Yeah.

[00:34:01] I mean, but it is interesting, right?

[00:34:04] Like, and a lot of people like to say it's a generational technology.

[00:34:08] I don't know.

[00:34:09] We lived through another one, the cloud.

[00:34:11] Generational technology.

[00:34:12] And, you know, but I do think that it's meaningful.

[00:34:15] It will change the way we work.

[00:34:20] Just like computers did in the early 90s.

[00:34:22] Yeah.

[00:34:23] And we've lived through it before.

[00:34:25] I am a historian by education.

[00:34:27] Oh, excellent.

[00:34:28] And so I tend to take a slightly longer term view of things.

[00:34:33] And, you know, it's always scary in the moment.

[00:34:36] And there will be backlashes.

[00:34:37] There always are.

[00:34:38] And then, you know, we kind of get on about our world and do the things in a slightly new way.

[00:34:45] But fundamentally, the needs that we have don't shift.

[00:34:56] Well, next year, we'll come back, hopefully, and visit this topic and say,

[00:35:00] well, the bots did take over.

[00:35:02] Sorry, bots.

[00:35:03] I know you're listening.

[00:35:05] Yeah.

[00:35:05] Yeah.

[00:35:06] No, I want to...

[00:35:06] You had one question on here that we didn't get to that I want to just mention real quick.

[00:35:09] One of the questions you said was, like, basically, what's the next thing we're going to be talking about, right?

[00:35:14] And I think, like, the immediate next thing is agents, agentic AI.

[00:35:18] But I think the thing after that is going to be these large action models, which is kind of a step beyond agentic AI.

[00:35:27] Because with agents, the idea is that the AI agent will do a single prescribed thing based on data that has been pre-trained on, etc., etc.

[00:35:36] And the large action models are a little bit closer to your dystopian future in that they are, in theory, able to reason and make more kind of decisions on their own.

[00:35:50] And I think that as we project out, let's call it 12 to 18 months, I think we're going to be talking a lot more about those.

[00:35:58] Right.

[00:35:58] So for your listeners, that might be an area to start to get smarter on.

[00:36:03] And also, we should watch the regulations as well about where that might be curbed.

[00:36:10] Yeah, for sure.

[00:36:11] I mean, there's been so much in the headlines about Meta not rolling out some of its AI features in the EU because of some of the regulations.

[00:36:19] And I'm sure that there will be others who follow that path.

[00:36:22] And so there will be lots of questions about what can and should be done.

[00:36:25] And, you know, we didn't talk at all about responsible AI or ethical AI, which we believe very, very much so in.

[00:36:33] So there's a lot that's at play here.

[00:36:36] Right.

[00:36:36] But I think that just kind of having a sense of what's coming around the corner is always useful so you can develop a perspective.

[00:36:42] Well, and if you don't hear here on the HR Data Labs podcast, definitely follow Stacia Gar, one of the most brilliant people today in the world of HR technology.

[00:36:51] And I'm a huge fan of her and everything that she's been doing at Red Thread.

[00:36:56] And we're going to put some links to Red Thread in the show notes.

[00:37:00] But you really have to listen when Stacia talks as well.

[00:37:04] If you go to a conference that she's at, definitely be in the audience.

[00:37:07] You'll definitely enjoy it as well as learn so much from her.

[00:37:11] Stacia, thank you.

[00:37:12] Thank you, David.

[00:37:13] It was a pleasure.

[00:37:14] My pleasure.

[00:37:15] Take care.

[00:37:15] Stay safe.

[00:37:17] That was the HR Data Labs podcast.

[00:37:20] If you like the episode, please subscribe.

[00:37:22] And if you know anyone that might like to hear it, please send it their way.

[00:37:26] Thank you for joining us this week and stay tuned for our next episode.

[00:37:30] Stay safe.