The Skills vs Tasks Debate We Need - Angela Le Mathon & Sandra Loughlin - #177

[00:00:11] Hello friends of the podcast and welcome to Directionally Correct, A People Analytics Podcast with your host Cole Knapper and today's guest Angela Le Mathon, VP of Workforce Intelligence and Insights at Walmart and Sandra Loughlin, Chief Learning Specialist at EPAMP. In this episode we will cover if skills are a latent construct you can measure directly. Is a skill something you can even measure?

[00:00:41] Yes. Whether tasks or skills are the real unit of work analysis that organizations should be building around? It's not enough to just know tasks because then you don't know who can do them. How AI is changing the fundamentals of understanding work and the workforce? I do have a question, this is just an aside, I'm just curious on your opinions on this because I've had kind of gone in circles around it. Should you treat AI agents as workers or as technology?

[00:01:11] And what HR operating model actually needs to look like in two to three years? But like, HR is changing too. How do you see this AI workforce transformation changing not just how HR operates? Now let's get down to business. Hey Directionally Correct fans. This podcast is dedicated to you to help democratize people intelligence for the world of work.

[00:01:38] If you're looking to support the podcast, please make sure to listen weekly. Subscribe to the Directionally Correct Substack newsletter. Sign up for the Data Driven HR Academy at datadrivenhracademy.com. Purchase Cole's book, People Analytics. Or check out everything else at colenabber.com. Before we get into it, a quick word about HR Bench, the company powering this podcast.

[00:02:04] You know, when we all started in people analytics, we wanted to do strategic work. Building predictive models, workforce planning, advising the C-suite, and most of all, quantifying the impact for the business. Instead, we spend months building dashboards and reports that should already exist. HR Bench eliminates that entire phase. Your HR is connects, your metrics calculate, your benchmarks populate. This is not novel. This is day one, not quarter two.

[00:02:33] That means skipping straight to prescriptive analysis, storytelling, and taking action for the business. Want to learn more? Book a demo at hrbench.com. Slash Directionally Correct. Find out more about the company powering this podcast and building the future of people intelligence. As always, all opinions are our own and thanks for being a listener. I wanted to say congratulations to both of you in different respects before we get down to business.

[00:03:02] Angela, first of all, on your new big role at Walmart, congratulations. Sandra, you are also on your way quickly to becoming a published author as well. Congratulations on getting a book deal to probably talk about some of the stuff that we're going to talk about today. But I don't know if I need to do a whole lot of introduction to both of you. Both of you have been guests on the podcast before. Amazing episodes, some of the more, you know, top 10 percentile fun episodes.

[00:03:31] I love those kind and really glad to have you back today. But I wanted to frame today as a debate. It's not going to be a debate, but as much as I'll try to make it one, maybe it'll be one. But we keep hearing out here in the market about skills is the future of the work or skills. Some people say the future of work is tasks.

[00:03:57] Some people say that the future of work doesn't even have humans in it and, you know, everything in between and the combination of that. And so I guess maybe I just wanted to start on the skill side of the equation. And this is probably more geared towards you, Sandra, based on kind of your background. But is a skill something you can even measure? Yes. Yes. Sorry. Like, sorry, I will elaborate on on that. Yes.

[00:04:26] So skills are yes, they're measurable. I think you can pretty much measure anything relatively well. I'm like an empiricist. But certainly when it comes to skills, yes. The challenge with measuring skills, though, is that, A, you have to figure out what they are and, B, they're latent constructs, which means that they can't be measured directly. So unlike height, for example, a skill, first of all, depends on how you define it. And then it is contextually dependent.

[00:04:56] So it might show up in certain places, but not others. And it can decay over time. There's lots of different issues with skills, but bottom line, they are absolutely measurable. And I would argue, hence maybe the debate, that the future of business will depend on organizations actually measuring skills effectively and using that information to drive more informed practices. Absolutely.

[00:05:25] Well, let me let's take the other side of the equation here really quickly. And just to cover our bases, Angela, you've got a background, obviously, on the skill side of things, too. But what about tasks? What is a task? Can you measure it? What do you use it for? What are your thoughts on that? Yeah, I mean, I think so task has become quite popular because we know that AI, for example, is a technology solution, however you want to frame that, that allows you to complete particular sets of tasks.

[00:05:55] And so the interest around tasks has kind of kind of bubbled up as a result of that. Right. And so tasks are things that we are, you know, we're used to sort of executing on and delivering. And so the opportunity now becomes how do we think about that in a more structured way, in a systematic way, so that we can actually now try to harmonize this for a business impact. Right. And so, yes, tasks absolutely can be organized and structured.

[00:06:20] I think where the debate comes in less so than skills, maybe is the is the fact that like, you know, they can be measured quite simply, easily. The benefit of those tasks and the impact of those tasks, I think, are things we have to think through because not all tasks have the same impact. And some of them are probably more straightforward in terms of how you define them in terms of measurement and others could be more complicated because they're longer downstream impacts.

[00:06:42] But I think the focus right now on the interest we talk about running things at scale is how do we start to define those tasks where we think there's an opportunity for us to then influence them in a particular way? And either that's if I increase the skill of the person that then, you know, the ability to kind of deliver on that task looks different. Or if I improve my technology and the systems and the tools that are supporting it, can I now reframe a role because of the fact that I've got technology supporting it?

[00:07:08] So tasks have now become the focal point for many companies in terms of trying to drive impact. Why is it more simple to measure a task just out of curiosity? I think it's because in many cases you can see them. And so when you're thinking about things like what we're looking at solutions like summarizing an email or researching. Right. So a lot of times you'll see demos of agentic solutions where you're getting an agent to do sort of like these very basic tasks. And in some cases it's interesting and don't get me wrong, you can kind of eyeball that and go, yeah, that's helpful.

[00:07:37] Whether or not scaling that particular activity across the enterprise is going to get you a measurable impact that matters is a different story. But it's a thing that you can see and people tend to react to that. Right. Like if I can see it, it must be true. It must be real. So skills, because you can't always see a lot of them are where it now becomes a thing of like, did you really have to scale? Is it really true? Is that really the thing that's driving it? So I think that's what makes it slightly more challenging for different users. If you don't have a background like Sandra, for example, to understand what skills are.

[00:08:07] Sandra does have an amazing background. I will say this. We had an episode a while back with Neil Morelli from Salesforce. And in that one, we talked a lot about AI workforce transformation. I feel like the discussion we're having today is kind of the core building blocks of what is going to make AI workforce transformation possible.

[00:08:26] And so maybe we can just check off the end of this debate section in the sense to say, if you had to choose, is it task or skills in terms of the unit of measurement? Like if you had to put all your chips in and just say like, okay, I'm betting on the future of how to do AI workforce transformation. Is it task alone? Is it skills alone? Or is it both? And what other data sources would you need to do that effectively?

[00:08:56] I'd love to kick it to you first, Sandra. I'm assuming there's probably going to be agreement here. But let's see what we get to. I don't understand how this is even a debate. Like not in this context, but like in the field. Can I actually answer that? Yeah, please. It's because the vendor market is flooded with marketing about what is the future and it is only this or it is only that and only people who know how to do things are this.

[00:09:21] And I mean, Angela and I even had a debate at SIOP a few weeks ago about what this looks like by the time this comes out. And it was a fun debate, but it was it was kind of pedantic at times. I think you could say because largely it's it's all of the above. But I'll just say that to say kick it back to you Sandra.

[00:09:39] Okay, well, I think number one, we should not be building our world model mental models on what vendors like to talk about, because the fact is that you have work, which I think is beautifully like very, you know, concretely, granularly defined by tasks and you need someone to do the work. That can be humans, it can be AI, but that you and you don't just need to match them now, you need to predict how to match them in the future.

[00:10:07] That's where skills come in. Skills are like the things that people or AI in theory hold that they can then use to do work in the future. So it's not enough to just know tasks because then you don't know who can do them. It's not enough to just know skills because you don't know where to apply them. You have to have both. What do you think, Angela?

[00:10:30] Yeah, I mean, I think I echo what Sandra saying, but also I think it's important to just kind of be mindful of the fact that the ultimate goal in companies is probably more focused on not just operationalizing tasks and sort of officially executing on tasks, but, you know, developing skills within humans, I think is really, really important. And so not arguing that obviously you can't get agents to support and have a role in that. But I think at the core, the human component is really, really central to how we think about things and how we as a society operate.

[00:11:00] And so I think like Sandra kind of beautifully described, I mean, though you can't sort of measure skills and maybe the same ways that you can measure skills at a task level when it comes to how you think about this as an enterprise, they're really important components of a puzzle. And so you need to have visibility of them together. And I think they sort of complement each other beautifully. But I would be nervous about sort of the focus on the only debate, mainly because I think, you know, when people are talking in that,

[00:11:27] it's mainly because if I'm only selling one thing, I'm going to tend to make the argument that that's the thing that's the most important. But I think companies that are trying to solve problems and that are operating within larger ecosystems, you know, there's a lot more complexity they have to navigate. And so it's very important to be balanced when it comes to these kind of conversations. Yeah, I definitely agree with that. If you're a one trick pony, you've only got one trick, you know.

[00:11:51] The thing I think about right now and this is really kind of undergirding the whole discussion is like, how is AI changing kind of the fundamentals of the concept of work? I do have a question. This is just an aside. I'm just curious in your opinions on this, because I've had kind of gone in circles around it. Should you treat AI agents as workers or as technology? What do you think of this? Right?

[00:12:15] Because I feel like some people are like, well, we've got to anthropomorphize them. We've got to, you know, treat them as in them. But like a lot of times, AI isn't a human. It's a process like it's in is tackling different segments of a process or maybe the entire process in the end. I don't know. How do you guys think about this?

[00:12:33] Well, I think it's an empirical question. Actually, it was just looking at research yesterday or two days ago showing the human, the psychological impact on humans of treating AI like a colleague. Okay. Which to be fair, like wasn't good. Like it made like the humans like feel disenfranchised. It made them feel like.

[00:12:58] Like, first of all, there's I think there are structural benefits in certain instances for thinking about AI like a workforce element that has to be trained and managed and like whatever. But anthropomorphizing, I think is going to have negative returns on the humans. And ultimately, unless we believe in dark factories, which for one of course I don't. Well, they do exist. So you have to believe in them.

[00:13:25] No, I sorry. I believe that they exist. I believe that their utility is far less ubiquitous or it has it has much more limited utility in many industries than I think Silicon Valley would like us to believe.

[00:13:41] But anyway, but regardless of that, I think the future of the workplace is going to actually be more psychologically grounded because now for the first time, I think in at least in several hundred years, we actually have to have humans think and learn and work in new ways and lean in and contribute.

[00:14:06] And all of that requires a degree of understanding of the psychology of people, which for what it's worth, I don't think companies have actually done very well to date. And therefore, thinking about not just what makes sense from an efficiency perspective in terms of like how we treat AI as a colleague or a technology, but the impact on humans who have to work in that workplace too, I think is non-trivial.

[00:14:35] Yeah. Yeah, I would add, I mean, I think, you know, I think it's a really sensitive thing because I think we don't want to lose human to human collaboration and human contribution. And I think when you start to go down that path, you muddy the waters in a way that I don't know is super helpful, especially when you think of long term implications and how these systems are actually set up and what they're actually doing.

[00:14:55] So I do think it's an area where there's probably context or situations where, you know, I've seen people naming, you know, and having a sort of attachment to some of these tools. And I think in some cases that are quite, you know, isolated, you know, fine if it helps you get through work better. But I would kind of look at companies and kind of say, look, long term or just as a strategy overall, it's probably not the direction of travel. I don't think you want to create that confusion with your human workforce.

[00:15:24] So I think it's a space to kind of watch and just be mindful of because you've got to be thinking of not just short term implications, but to longer term impacts in terms of what capability as an organization you're building. What muscle are you trying to sort of establish or stand up? And I'd be very nervous if we're not paying attention to that and being super intentional about the implications just because we've got a shiny new object at the moment that looks cool and it's responding to us or at least largely saying the things that we want because we've programmed it a certain way.

[00:15:52] I think we just have to be really mindful about what this is actually doing to our workforce. Yeah. And in prepping for this, I can't remember which one you said there's cool thing energy. There's a lot of cool thing energy to it. People like to be cool, right? Let's be honest. Oh, gosh. Well, I'm thinking like this is changing. I mean, obviously it's going to change and transform how businesses operate kind of soup to nuts, as they say.

[00:16:23] But like HR is changing too. How do you see this AI workforce transformation changing? Not just how HR operates, but what are the problems it tackles? Where is it relevant? You know, what are the skills that in tasks that HR itself needs to be engaging more in or less in? I mean, go go whatever direction you want to go. What thoughts do you have on this?

[00:16:50] I mean, my first reaction is I think assuming you get the building blocks, all the pieces aligned, because I think sometimes the ambition and the reality are really different. And so HR functions can hear all of this stuff and go, oh, my goodness. And then the reality is like they still work very similarly to how they've been working in the past because some things haven't actually shifted.

[00:17:10] But assuming that the world kind of is moving in the direction that we're imagining, I think you start to have a lot more data and a lot more insights, a lot of the ability to transact faster, move at pace, et cetera. You're in like a slightly different driver's seat than you probably were before. And so I think it almost becomes a question of like, how do you want to show up in your role? Like, where do you think your value is? Right. Like we talk about the coaching part of that.

[00:17:34] But how much of that do you want to be involved in if you can also be empowering your business or your leaders to kind of do a lot of the work? So I think there's going to be this moment where there's a sort of a recontracting or renegotiating that's going to emerge because suddenly it's like getting certain aspects of the role done or getting the work done is becoming a bit easier. How much of it needs to stay within the sort of domain of the people partner or the business or HR as a function? How much of that do you start to push to managers or back to associates?

[00:18:02] So I think this rebalancing of like where work activity has to happen is probably going to be up for grabs. And I think that's what's going to probably be interesting. I don't know that I have the answer of where that's going to sit, because I think I've seen iterations where there's moments where the business is like, give it to me. I want to do it. And then you do. And then they go, I don't really enjoy doing this. This is a lot. So it just sounds like HR stuff. Why am I doing this? Right? This feels like heavy. And so I think that's the reality of it. Like it intellectually sometimes could sound cool to kind of go, oh, great.

[00:18:32] I can self-serve and do everything myself. And then it's like, oh, that's a lot of self-serve. I want someone else to do it. So I think you're going to get some of this dance, I suspect. I think that's right. And I would kind of like to take that theme and like push it up even higher.

[00:18:48] I would say, okay, my thesis is that the whole Taylorist design and the contract around work and the workforce, I think is broken. I think it's been broken for a while.

[00:19:05] But the whole Taylorist philosophy is based on the idea that work is pretty stable, that people are like widgets basically that you kind of buy and you fit them in to do the work. And then you kind of like fix it and forget it because the widget doesn't change. The work doesn't change. And if the widget kind of gets a little bit out of shape, you'll either like bang it back into place or you toss it out and you buy a new one.

[00:19:32] And like to me, that's like been the foundational kind of very stripped down, very like, you know, non-PC way to describe work. And that model's been kind of shifting out of place for a long time. Like the world has moved on. Work does move faster. Human variance is actually desirable. And I think that AI is about to, is going to break that mindset.

[00:20:03] Yeah. Not just mindset, the systems that kind of reinforce it. And HR is the primary system that reinforces this. Like this is the system that like HR is designed to make that thing go. And so assuming that human variance does matter and that jobs do change and that we even need jobs, I think we do.

[00:20:26] But let's just kind of just say, then I think the entire HR function gets rethought from the ground up. And exactly as Angela said, right, some things, there's going to be some sort of like back and forth. The business is like, obviously, I'm the right person to do that. And then they get the work and they're like, just kidding. I don't want to do any of that stuff. So I think that over time, the shape of HR will evolve.

[00:20:50] It will be an iterative evolution where the business takes some, gives some back and back and forth. But I think it is very, it's very interesting to think about what if you didn't have HR at all and you actually are designing an organization around who can do the work best? How do we know what the work is? How do we know who can do the best work?

[00:21:18] What does a system like that look like? That is where I think we're going to be going pretty soon. I think, I hope, because if we don't do that, then I don't know how companies are going to survive. Because AI demands an entirely new way of thinking about work and the workforce than we have in place today.

[00:21:40] Yeah, there's this notion of kind of zero-based budgeting, but thinking about it in terms of your workforce or even your org design. I will say just, you know, because we all live in reality that there's a huge countervailing force, which is just inertia. And the existence of current paradigms of how to think about these things. And frankly, just the work that needs to get done and people just doing that work and not really changing all that much.

[00:22:09] But I do think that organizations are transforming pretty rapidly. And you see it all the time in the news. You know, XYZ Corporation cuts 50% of its workforce, blah, blah, blah. I have been thinking about this a lot. I want to get you guys' reaction to it. The reason why this discussion about AI workforce transformation is challenging is because it's a multi-level problem. Meaning there's a way that executives talk about it. There's a way that teams and functions talk about it.

[00:22:37] There's a way of looking at it just based on roles. And then there's a way that individuals themselves think about it. And so executives to me seem to just care about like, how are we controlling costs? You know, how are we being profitable? What is the tokenomics versus the cost of labor? And like, do we have the right optimization there? Teams seem to be more focused on the process level. Like, can we use AI injected into the processes, become more productive?

[00:23:06] Roles seem to be like where this whole like task versus skills versus AI agents come in. And so do we have the skills? Do we have the right roles? Do we have the job architecture? All of that level of analysis. And the individuals themselves are just trying to figure out, am I productive enough? You know, am I adopting these new tools at the level and the pace that I need to do to kind of keep up with the Joneses of what's going on in the world? I don't know. Do you guys see things through that lens at all? Or have you experienced that?

[00:23:36] Or am I out in left field? I don't think AI transformation is happening that fast at all. No. Because I actually don't think that employees are sitting around thinking, where can I adopt AI? I think employees are sitting around thinking, what is the minimum I can do to keep my job? And like, pay for my mortgage and send my kids to summer camp.

[00:24:03] Like, I don't think that a typical employee is excited about AI. I think they actually feel very threatened by AI if they think about it at all, which the data suggests that most people don't, or they're not using it in their roles, or at least they're not using the tools their employers require them to use in their roles. So, I don't know. I think that the entire system is actually designed to reject AI.

[00:24:33] And in most other new things that come along. Exactly. Exactly. This is why I say, like, it could be total pie in the sky, like, ridiculous thinking. But we are trying to strap AI on top of a system designed for the industrial era over 100 years ago. Like, that doesn't go well.

[00:25:00] Like, either the system will reject it, or more likely, and I think what's actually happening, is the system will reject it except where AI reinforces existing incentive structures. And right now, the existing incentive structures for employers are to, like, for business leaders are to make things cheaper by eliminating humans, eliminating, like, junior hires. Like, right?

[00:25:29] But then you have individuals who are incentivized to do exactly what they did yesterday. So, you have this, like, weird misaligned incentive system at all the different levels. And as a result, I don't see AI transformation happening in any sort of meaningful way anytime soon. And that is not even factoring in the problem of data and the fact that it is, like, terrible and siloed and largely, like, exists in the heads of people.

[00:25:58] Like, I don't know. I have become increasingly negative, I guess. I would say radicalized. It's like, I used to be afraid that AI was going to blow everything up. Now, I'm actually more afraid that the system will, like, constrain AI so much in incumbent organizations that they're going to be, like, essentially, like, outcompeted by AI natives.

[00:26:28] And I don't think that that's good for society. Any thoughts, Angela? Yeah. I mean, I think what, you know, AI maturity in companies, right, is what kind of you're scratching at a little bit there, Sandra. The reality is really different when you're kind of on the ground, right? And so, the ambition and the way we talk about things and we tend to kind of operate in this, like, sort of, like, really odd mindset in terms of how it's taking over. And I think the reality is just not that.

[00:26:56] I think employees, for the most part, are just trying to do a good job or focusing on them themselves, right? And so, it's more a question of, like, okay, so if AI is a new thing, what does that mean for me? How does it impact my current reality? I think they're just thinking about those things. I don't think they're getting any further than that. And they're just watching to understand. And I think companies are also in a position where they're trying to figure out, like, okay, I've heard all these great things from the vendors, like you mentioned, all these promises. And so, in cases of some companies where they've invested, they're like, is it true?

[00:27:24] And the other companies who haven't yet invested, should I? So, I think it's more of that conversation that's probably happening. And so, I think where there is pressure is to show value. And so, it's almost like, okay, great. So, for the ones who have done it or who are further along the journey because they've gotten their data structures in order and they've had a longer time to sort of work with it, okay, what does that impact look like? And what does it take to actually, you know, get that impact? And is it something that's worth the journey for some businesses, right?

[00:27:51] So, if you don't have a clear use case for how AI is going to actually impact your business model in a way that's, like, meaningful and sustainable, the question is, like, well, then why am I going down this path? Or why am I going down this path in this way? There might be a version of AI you can explore, but I don't think it's one size fits all, especially because agentic AI is more experimental. And a lot of people are focusing on that. But not every company is going to benefit from those type of technologies. So, I do think there's a smart discussion to be had in terms of, like, anything.

[00:28:20] Where is the value play coming from? And making sure you're aligned to that as opposed to the, like, it's a shiny new object. Everyone's playing with it. I want to play with it. Because it gets expensive if you're trying to do this at scale, right? So, I think, if anything, what you're describing, yes, people are, some pockets may be sort of thinking about it. But I don't think it's this mass thing where everyone is sitting there going, oh, my goodness, AI is about to take over. What do I do? I just don't think the reality is quite as broad. I want to dig into that for just a second.

[00:28:51] Because I think a lot of companies, you know, they are chasing the new shiny object. But there's also behind the scenes a lot of investor pressure to make these changes to become, you know, more profitable, more efficient, more innovative, control costs, take costs out of the business, all those kind of things. And so, you said multiple times, Angela, about where the value is created. Where is the value created?

[00:29:17] Like, what are the kinds of things where value is created with adopting this type of new methodology of ways of work? I mean, AI is a really broad term that we tend to sort of lump all the different sort of capabilities into. So, I think part of the challenge is around that. I think for companies, depending on the problem you're solving, there are different opportunities to figure out, okay, well, where do I have a lot of messy data? And therefore, getting the system to sort of process that data adds value.

[00:29:43] So, I think, one, there'll be some companies where they've got a ton of data and that data is really important to how they drive value in general. So, they should probably be one of the first candidates to be thinking, hey, I got a lot of data I need to make sense of. And I struggle with the amount of humans involved in helping me make sense of it. So, that's probably a strong use case to say, yeah, you're probably ripe for some form of AI innovation because of the amount of data you have to process.

[00:30:05] So, I think that's one group and they'll have success in sort of structuring and making sense of different data sets that exist in the world. I think other groups will then kind of get value because they need humans to make sense of that or add at least some sort of context to that data, which is not always readily available from some of the technology solutions.

[00:30:24] And so, I think there'll be a second bucket of companies that will be able to say, hey, if I can get my teams to get some of my data structures right, then I have the best opportunity of saying, how do I get the right humans in my org to actually engage with the data that's emerging and therefore an opportunity can create.

[00:30:41] And so, I think some of the cases that you would have heard about that are the most popular are in the sort of like pharmaceutical and the science world where you've got scientists who are able to come up with new technologies or new modalities or identify new pathways because of the fact that they're able to process data in a way that they just could never do in the past. And so, I think in those cases, it becomes really meaningful because they're discovering new ways in order to like, you know, solve for problems, right?

[00:31:06] And so, I think you'll have groups where that type of opportunity will be ripe and it'll make sense for that investment. I don't know that everyone's probably in that same bucket where they have a ton of data that you need humans to make sense of and therefore they need to make all that investment. I think if you're not necessarily in a business environment where that's a key part of how your business thrives, that might not be the thing for you. And so, getting something off the shelf to experiment with might be sufficient if you're thinking about business impact.

[00:31:34] But I do think those are the types of questions or arenas where there's value in exploring. Absolutely. Lysandra, I want to come back to you just based on something you were saying earlier about the AI natives kind of taking over and that would be very disruptive. Just out of curiosity, in your revolutionized world, your radical world, in these AI native companies, does HR exist? And if so, what does it do?

[00:32:05] Boy, I have no idea. I mean, I would assume yes. So, okay, let me start back. An AI native company to me is an organization that it's just like a digital native, like, you know, like Amazon and Google and Uber and Facebook and whatever. But it's none of them are AI natives. I think AI natives are companies that are being built right now that have AI at the core. And they have the AI at the core of three different things.

[00:32:35] One is that it's at the core of all their products and services. The second is that it's at the core of how their business actually operates, like the work that everyone does all day long. And the third is that it is used as a mechanism for deeply understanding the business, like the actual function of the business, the customers, the employees, the secret sauce, everything.

[00:33:03] So, when we talk, a lot of people talk about digital natives or, sorry, AI natives just in terms of products and services. But I actually think the use cases are more universal and more interesting when you think about operations and knowledge, intelligence about the business itself.

[00:33:20] But in terms of like how is their HR structured, I think that it will be very different than how incumbents, HR, orgs are structured simply because HR exists as a function to do things that the business doesn't want to do. Like if the business wanted to do it, the business would do it. But they don't want to. It's inefficient or for whatever reason.

[00:33:44] But I think when you are redesigning the data architecture and you're building systems that are designed to do a lot of the processes that HR does today, I think it lowers the bar to entry for businesses to actually do some of that work themselves, which in many cases makes more sense because they're closer to the work. They actually have incentives aligned to putting the right people in the right positions to create the right value.

[00:34:11] So, I think at least at EPAM, which again is not an AI native company, our HR structure is very different because it's so distributed into the business and it's so built into the tools and data and systems that we have that it's just structurally a completely different kind of function in the organization.

[00:34:34] Yeah, that makes a lot of sense. I'm curious. Just one more question. I like playing in the future. I like playing with ideas. Do jobs exist in this organization? Yes, I think so. Jobs are an artifact, right? There are people out there that are saying like jobs won't exist. You're just going to be kind of an amalgam of work and all this kind of stuff.

[00:34:56] And, you know, you're going to get pulled in for gigs things. And I'm not doing it justice, but I'm just wondering, do jobs exist in these AI native organizations? I think in theory, jobs don't need to exist. But I think they will. And the reason is because humans want to be in groups. Humans want to say to somebody else, I do insert this thing.

[00:35:20] They want to be in a conversation or go to a conference or whatever and hear somebody else say, oh, I do that same thing. People want to some degree, they want structure. That structure isn't necessarily the most efficient or effective thing, but it is a psychological value to people.

[00:35:40] And I think there is also structural value in having to some degree definitions of things. The problem is when we get stuck in the definition and build systems around that stuck definition so that the system gets stuck by the definition.

[00:35:59] That's what I think jobs have done in organizations to date. So, like, I think jobs will exist, but I think that the boundaries will be more amorphous than they are now and the jobs will evolve over time. And that will be an expectation and like a feature, not a bug. Angela, any thoughts on any of this before I ask one more question? In terms of like jobs existing?

[00:36:28] Or just any of that. I mean, yeah, I mean, I don't know if it's going to be radical. I think in that way, I think jobs, you know, people are defined by that. I think there's a lot of jobs that you'd want existing. I don't know that there's some jobs that you want all going away because we live in a physical world. Right. And so a lot of those physical jobs are quite important. But I think where there's going to be meaningful impact is where can AI help accelerate those types of jobs in a way that's going to be helpful?

[00:36:53] I think that's kind of what I'd see as a big change if we can get our act together to do that properly. Yeah. Well, I've been talking a lot about in some prior episodes and I finally published a people intelligence manifesto on how I think people intelligence is kind of taking the place in some ways of people analytics and traditionally just kind of backwards looking data on organizations and dashboards and all that.

[00:37:19] So, the question I have for you, and this is sort of a provocative question, is what type of intelligence that you don't have today that you would need or like to have to be able to kind of make this transformation possible? Because I think both of you have said, you know, data kind of undergirds all of what we've been discussing thus far. But what's the level of intelligence from that data or maybe even data that you don't have yet that you would like to have that would help make this transformation more possible?

[00:37:51] I want to know what people do all day long, which I don't think anyone actually knows. Like we don't I mean, in like a non creepy way, like organizations have zero insights into what people do all day long. And as a result, they don't know where to stick AI. They don't know what skills they need, what skills they have. Like we have basically no information on people beyond the job description.

[00:38:19] That's how the system was designed and it's working great. I think in order to make progress with AI, we need to understand in a much, much more granular way what people do all day long and who's doing it and what do they want to do and what are they good at? Like I want to know all the stuff. She wants to know everything. Yeah, I mean, I think probably an opportunity is, you know, companies having to figure out like what is your role in society?

[00:38:46] Like, you know, do companies have to be on 24-7? You know, sort of like what is it that you need to get? What is it that you need from humans to sort of deliver your value and your business impact? I think it'll probably come down to something a bit philosophical in that regard, because I think once you sort of understand like what type of business am I running, what is the impact I'm trying to drive, what is the role of humans within that ecosystem,

[00:39:10] then I think how you negotiate and how you then contract with humans within that environment will probably shift and change in some form, right? Because I think that's what's kind of like hitting at the how many humans do I need in my company versus how many potential agents or whatever the case is. And so I think that's probably what companies are trying to figure out. It's like you can spend a lot of time trying to figure out what all the humans are doing. Equally, you can start to stand up all of these agents that do a thing, but is that the full job of what the humans are doing?

[00:39:36] And I think everyone's debating and realizing, oh, not really, because the tasks that these agents are doing are not the full scope of the work, right? And suddenly it's like light bulbs are going off. So I do think companies will probably have to stand back and just sort of start to really say, well, what is it exactly that I need to know to run my business? And then kind of like, is it worth the effort to then go after that? Because I think trying to get a handle on every movement that humans make, I mean, you go down and you get into the big brother aspect.

[00:40:02] I think there's a lot of evidence that shows when humans are in those environments, it's not the most productive long term. It causes other issues. And so I just don't know if that is the end goal, as opposed to you're really here to drive profits. And so you're probably going to just focus on what's needed for that. And then I think everything else will just become like, you know, icing on the cake. But it's like solve for the business issue first, I suspect. I'm calling my shot now.

[00:40:27] I think that employee surveillance is going to be the debate of our time. Right. Right. Because in the end state to kind of to Sandra's point that everybody just wants to know, what are people doing all day? Right. And not just what are they doing behaviorally, but kind of like what's between the six inches between their ears to what are people thinking all day? And this becomes a quite creepy proposition.

[00:40:54] And so, yeah, I think that we're early, but this is what's coming. And I think you're already seeing it creep up a little bit in certain organizations that I won't name. So, I mean, I OK, fine. I also won't name them. I was going to. You can name them if you want to. I'm not so bad. But I will say so I work in a company that is the closest thing to Big Brother that I think exists right now. And.

[00:41:24] You would think that it would be a huge problem for employees, but it's not. So I think there is a way to collect data. No, it's not how you collect it. It's how you use it that matters. It's the transparency around it, the value that you bring to employees. Or if you think about like the amount of information that we give to.

[00:41:49] Actually, probably evil organizations all day long on our phones, like people are actually willing to give up a lot of information. In exchange for stuff that's valuable to them.

[00:42:03] So the way that EPAM works is we collect a shocking amount of information about people, but we use that information to actually improve hiring and mobility and career progression and learning and staffing and and to make it more kind of meritocratic than other organizations. It's possible to do this. I don't know that it's going to be how many organizations actually use the data.

[00:42:34] But I think that will be to their long term detriment. I think if I guess what I said before, like, I think that AI is creating this maybe paradoxically an environment where companies have to do better by people. I think I think now if that's the case, then they can use data to improve the lives of employees and actually create competitive advantage.

[00:42:58] Or they can use data to squeeze people, make their lives terrible and get some incremental margin out of that and value out of that. It's possible. I just don't think I mean, it's definitely possible. It's likely. I just don't think that those companies are going to succeed long term. Yeah.

[00:43:18] Yeah, I'll say this like I think about like employee surveillance sort of through the same lens that I think about like national security surveillance, kind of the surveillance state you live in. And what we did is as we wanted more security, right? We didn't want bad things to happen to us. And so for every unit of increase in security, there was like a unit increase in surveillance to go along with it.

[00:43:45] And I think what is being proposed is we need to increase the units of like human autonomy or employee autonomy and employee benefit organizations for every unit increase of employee surveillance we put in. And the the benefit has to come first, not the surveillance. I think that that's really the only formula or just no surveillance, but maybe no employee benefit. Who knows? Hey, y'all, I'm Lee Cage Jr.

[00:44:13] And I'd like to invite you to listen into my podcast 15 minutes with 15 minutes with rising stars and seasoned disruptors, thought leaders and change agents who are sharing how they're reshaping work, rethinking worth and reimagining what's possible. This is fast paced, hard filled and unfiltered. These aren't just conversations. They're catalysts. So tune in, elevate, share the shift. 15 minutes with wherever you get your podcast. No thoughts.

[00:44:43] All right. This is why I have the podcast. Well, do you guys want to join me in Cole's Corner? Let's do it. Welcome to Cole's Corner. Let's do it. All right. I'm going to skip to what am I reading because both of you have guys have been on before and we can talk about we've talked about rapid fire and all that. So people can go back and listen to those episodes if they want to hear the funny stuff from those.

[00:45:11] But today, when I want to do instead of some research, let's do some me search. So the first article we have is from Angela and myself in the Direction Incorrect Substack newsletter called Skills Management and Capability Orchestration for Redesigning Work in the AI Era. And so the headline of the story, this came out not too long after Moderna merged its HR and IT functions together.

[00:45:39] IBM made some big waves. Goldman Sachs had in service now around the same time. And it was a lot about, you know, what role does skills inference model, skills data infrastructure play in making these decisions? And so essentially, I'm going to skip down a little bit. We talked about four requirements for capability orchestration. And so you have to split tasks between humans and AI agents. You need to create dynamic capability blueprints.

[00:46:08] You need to validate skills against real business outcomes and then orchestrate continuous optimization. So I'm just briefly covering a very dense article. It was very, very cool to write that with you. Angela, you're a very, very brilliant person. Anything you would add on this one before we open it up to Sandra? Yeah, I mean, I think a lot of those, that thinking is still true, given kind of what we've seen play out. So I think all of that is still very valid.

[00:46:35] I think whether or not companies are at the maturity level to get all those building blocks in place to kind of drive that, I think, is probably one question. But also, for some companies, getting to that end state, again, goes back to your business model. And so in that world, what does that mean for you? What is that going to do for you? So I think companies like a GSK or Pharma, very valid, right? Being able to process a lot of the data when it comes to drug discovery is super helpful.

[00:47:04] Probably the marketing world, similar dynamics where you're trying to think of new creatives in terms of advertising. So I think there's probably environments where that will still hold true. And then others will just have to kind of question, like, is that really needed? So that's kind of what I would add. Yeah. Any thoughts on that one, Sandra? I mean, I agree. I think I would extend that to say, in what parts of the organization is it worth it?

[00:47:28] So I think exactly to the point, Angela, you were just describing, it's like I think that some level of better data around work and workers is universally valid for every organization. But it's not universal for the entire organization. There are just certain roles that are closer to P&L, like roles where you make a lot of your money, you lose a lot of your money, or you have a lot of risk exposure.

[00:47:54] And in those roles and for that kind of work and for those people and agents, it's actually very, very helpful to have better data. Because right now, again, the assumption is, my assumption is that companies have very little or very poor data. And as a result, you can imagine there's just inefficiencies. When you don't have data, you're guessing. And if you're guessing, you're probably not guessing right all the time. So I don't have a lot of critiques, I think, of that piece.

[00:48:24] It was excellent. My only thing is, under what conditions does it make sense to pursue? Because it's a cost issue. It's a change issue. It's a problem. And it's only worth solving in certain contexts. So figuring out what those contexts are is, I think, the most critical thing. Absolutely. I love that feedback. And I will go ahead and move us on to our next one, which is some more me search.

[00:48:54] In the sense that Sandra came up with this new idea, which I suspect will show up in your book somewhere. And I think we're talking about writing an article together about it, which is the knowledge office. And so there's a visualization on the screen if you're only on audio. So I'll actually, Sandra, maybe I'll let you do your best job of explaining this visualization really quick. And then I'll give a little bit of oversight on it as well. But can you talk about what this circle chart is? Yes.

[00:49:23] So this is the concept of we're calling the knowledge office. So basically, if you think about a world where we've actually solved the data problems in organizations, this magical future, where we don't just have clean data and coordinated data. We also have continuously fed data by people explaining their context and tacit knowledge and whatever. So this like perfect, magical like data future for organizations.

[00:49:52] You still need to actually make sense of that data. Like you need people to look at that, all of the noise and identify the signals that show things that are going well that we should try to scale. Things that are not going well that we should try to fix. New opportunities, like to make sense of all of that information to actually improve the business on a continual basis.

[00:50:17] So the knowledge office is this kind of new function that we're envisioning. Actually, EPAM is doing this and a couple other companies are as well. where it is of humans plus AI that do four different things. Essentially, you sense stuff from the data. You make sense of the data. Then you envision what the future state should be for, again, not the whole organization, but like very specific things that you've learned about through the data.

[00:50:47] So if you see a problem in the data, you envision, okay, what is the way to solve for that problem? And then you kind of create a plan, like an intervention, if you will, for how to solve that problem or create that new product or whatever it is. And then the knowledge office activates. So they say, okay, team, go. And they kind of govern and make sure that the team is in fact moving forward, you know, toward that new vision.

[00:51:12] And then after some sure, some iterations and trial and error and fixing things, they get to the solution. And then the job of the knowledge office is to sustain that through changes in processes and tooling and incentive structures and whatever. So it's kind of like the brain of the organization. And I think my, you know, the initial plan that was like months ago that I wrote up, I think had five different elements of that knowledge office.

[00:51:39] So it would be a coordinating function that brings together people analytics, work analytics, org development, knowledge management, and L&D. And again, I don't know that this needs to be an actual formal office in an organization or an informal one. But basically, the bottom line is that one of the most important skills in the future is going to be org specific.

[00:52:08] What are the skills for your company that are special related to your processes, your special sauce, your whatever? And the knowledge office is, I think, well, I know, is a concept that is designed around that. How do you surface what those skills are, get people to develop them and then sustain the behaviors associated with them? Absolutely.

[00:52:35] And so I'm showing this post on the screen where you said, we talk a lot about durable skills, but in a workplace being transformed by AI, arguably the most important skills are org specific. Skills related to sensing, envisioning, and responding to your AI native workflows, your new AI enabled decision making templates, your operational friction points, your organization's IP and special sauce, your blind spots, your near misses and lucky breaks, your known unknowns and your unknown unknowns and so on.

[00:53:05] And so I'll pause there and say, first of all, I think it's a really neat concept. But second of all, Angela, did you have any thoughts on the knowledge office? Yeah, I mean, I think it makes sense. I think what companies are realizing is given all the foundational models that have scraped the Internet and that have sourced probably everything publicly available, they're looking inwards and kind of recognizing that the data that they have in-house is actually their secret sauce. And so how do we start to make sense of it, given that it is usually fragmented, disparate, siloed and all the things.

[00:53:32] And so finding a mechanism to sort of like centralize that and make sense of it from a point of view of I need an AI to help me then create value, I think makes total sense. So I think the concept is very sound. I don't know that it's the newest concept in the sense that I think we've always thought about like these organizations where you have like the CDO or the chief data officer and they had their orgs and they're trying to make sense of it. Probably for the purpose of maybe just management.

[00:53:59] But now I think because AI is kind of the end goal, which is how do I then take all of that and make sense of it? I do think it's almost like the evolution of that. So I do think it is very, very smart. I'm curious, Sandra, as you were kind of evolving that, what that looks like. So maybe we need to come back on the podcast and talk about the success of it. I think, yeah, no, it's definitely a smart direction of travel. I think you can't move past the fact that to get AI to work well, it's not just about having data. It's got to be data that's structured and that has a certain context tied to that.

[00:54:29] And so without that, you're not going to get the value. So I think it makes total sense of a knowledge office. I will hit us with the last article from the illustrious Angela Lehmathan, myself and Alexis Fink. She's amazing as well. And this article is titled What AI Actually Changes About Work and What It Doesn't. And so we go through, this is a very provocative one.

[00:54:56] It says, AI actually changes how work gets done, but it also exposes how poorly work itself is defined and understood inside most organization. More concretely, things like job descriptions are rarely reflect the reality of work. Most work isn't linear, digital, or cleanly task-based. Mediocrity and inertia. I think I wrote this one. And mediocrity and inertia are often structurally rewarded over excellence. Standardization is favored over experimentation.

[00:55:24] The talent shortage narrative is often more fiction than fact. Power, ego, and control frequently outweigh meritocracy and outcomes. And work is full of undocumented workarounds and tacit knowledge. And so we go through this and there's a few sections.

[00:55:38] First is what AI actually changes, how it gets worked, it's done, the task distribution changes, the speed and scale of decisions, the boundaries of roles, how feedback flows through the system, where accountability must be redesigned. And this concept that I think you introduced of adaptive labor, Angela, which I thought was very interesting. But then we talk about what AI doesn't change.

[00:56:04] And so it doesn't change judgment in ambiguous or novel situations. Frankly, the incentives in the culture, which we've talked a little bit about today of being kind of the inertia of organizations. Accountability for outcomes. AI cannot be held accountable. The need for clarity and purpose and values of your employees. And engagement in the physical world insofar as robots don't exist yet or aren't widely adopted. And then we talk a little bit about the new rules of work.

[00:56:33] So before doing anything, get your data about your work and workers in order, which I think aligns with a lot of what you've been saying, Sandra. Jobs will still exist, but will be in completely new and different ways. Humans will do human things and technology will do technology things. I feel like that's a tautology, but it's still interesting. And that collaboration is key. So any thoughts here, Angela, before I turn it over to Sandra and her thoughts on it? I mean, I think all those things are still relatively valid, right?

[00:57:03] I mean, I think, you know, AI. Are they super valid? We are so right. I think AI will change some things. And then I think the reality is that other things just won't change because, again, there's just constraints to it. So I think it's all still valid. And I think work is probably the thing we haven't spent enough time really getting under the hood of. So I think the Knowledge Office is an example of that. All of those artifacts, someone needs to look at them and start to make sense of it. And what is it really telling us? Is it really capturing the things that we think it is?

[00:57:29] And so I think now is the time for us to sort of, like I mentioned earlier, be a bit more honest in terms of what are these companies and these businesses that we're establishing? What are we really doing? What is the human component of that? What is the system actually that we're building, et cetera? So I think all of this is time or ripe for review. And I think that's effectively what the article is scratching at. Absolutely. Do you have any thoughts on it, Sandra? Oh, I agree with all of the things it said. And I would like add a couple other ones.

[00:57:55] But what I keep going back to is like AI is not going to do any of that stuff right now. Like when you talk about like what's not going to change incentives, I agree with that unless the system changes. And if the incentives right now are misaligned to AI, which I think that they are, then what AI can do is still very theoretical. Like in theory, yes, AI can do all of that stuff that you just said.

[00:58:24] In practice, I don't see it happening anytime soon unless incentives are realigned. Because incentives drive behavior. And the incentives for every organization and every individual and every team is to do exactly what they did yesterday. And so like you have a chicken and egg problem of like we want all of this cool AI stuff to be true. But yet your system is not designed for any of it to happen.

[00:58:54] You know what's so funny about hearing you say this, Sandra? So for those people who haven't listened to her prior episode, they don't know that she... Sandra's not like an HR person. You know, she didn't come up through her career. And so she's experiencing a lot of these things from the outside in rather than the inside out. And what you just described is the exact same debate that we've been having for like 15 years about when is HR going to adopt analytics? Right? The incentives aren't aligned to it adopting analytics.

[00:59:24] The incentives are aligned to HR staying and doing the same things that HR has always done. And why can't we get them to be more data driven in this decision making? And so it's funny that AI seems to be the old wine and the new bottle of the same problems that we've always had. So can I say I don't think so? As you said, I'm new. Like what the hell do I know? I don't think that AI is old wine and a new bottle.

[00:59:52] I think that AI has structurally changed the value of humans. Let me explain. So AI is a commodity. Like it's like electricity, right? Every company is going to have it. So companies that have it are going to be just like every other company that has it. So it cannot in and of itself be a competitive differentiator. AI is also eroding existing competitive modes that companies have now.

[01:00:21] Not eliminating them, but eroding them. So what it means is that now companies are like, what are they going to compete on? If everyone has the same AI, what can they possibly compete on? I think it's two different things. One of them is understanding your own organization. Angela, what you just called like your secret sauce. Completely agree. Like that's going to be huge.

[01:00:46] But let's just assume for a second that we could snap our fingers right now and every company will have perfect data today. Unless someone is adding new insights, like new edge cases, new ideas, new information, that data is going to get stale.

[01:01:07] And so I think that companies are going to have to actually compete on how effectively their people are thinking and learning and contributing and working in new ways. That means that the entire calculus around people is changed from being a widget to being a value proposition for the business. And if that's true, then HR can't possibly stand still.

[01:01:35] Because for the first time, people actually matter. Like how fast they learn, how well they think, how much they contribute. All of that now becomes for the first time a competitive differentiator where it hasn't before. And if that's the case, I don't think HR can stand still. I don't think the business will let them. To me, the question is not what is like, will AI or HR be different in the future? I think it will. It's who's going to be in HR?

[01:02:05] Is it the current HR people or not? I mean, I think every industry might be slightly different. I mean, I can only speak to if I look at Walmart right now, the associate humans, like we are very clear on where our value comes from and the value of humans. And so if anything, given the type of model that we have, you kind of understand the importance of humans when it comes to just everything about how we run ourselves.

[01:02:30] And so I think if anything, AI has only made us realize that more and start to realize like, oh, okay, because they are so important and because they drive literally how we do our business, how do we actually use AI to make their lives easier? So I think there's probably more of a push for us to figure out like, we want to understand the work so that we can understand what's the drag, what's the low value, what's the thing that's making it really difficult, the friction and all those things. I think there's probably a big pressure on that because you realize humans, you know, get tired, like agents don't get tired, but humans get tired.

[01:02:59] So how do we get smarter on that technology? I think there's probably a lot of pressure on that. But I think in companies like ours, where you've kind of understood really early on the value of the human itself, you probably are very practical in terms of really trying to understand where AI is going to unlock that value more. But I think it's really difficult for those companies where you can kind of see humans operating in the real world because, again, it's retail, right? Like you're operating in the real world to help real humans. You're like, yeah, AI has a role to play, but it's never going to be a replacement.

[01:03:27] It's never going to be more important or more value add. And even if you fast forward to the future where you've got all of the data components and all of the models and all the things working well, humans are still going to be very central to kind of how that part of the business, how that business is run. So I think AI will always be secondary. It will always be in service of the human. And I think humans just are the magical element and the secret sauce of what we do.

[01:03:49] So I think that's what also makes it quite difficult for us to take an approach or from what I'm seeing anyway, to take an approach where AI is anything but an enabler. I think it's kind of the sense I'm getting. So, yes, I think some businesses will, you know, they operate. You'll hear companies of the billion dollar company with one employee and a ton of agents and all the rest of it outsourcing heavily, of course, to humans, believe it or not.

[01:04:14] But I still think that's the point that I'm making is like the real value add and the value proposition is humans because we've got an X factor that agents just will never have. And you can try and program that. But that's the whole point. It's like you'd have to almost tell it a thing. So that's why I think the use cases and the value propositions with AI has always been the case of it does the hard part, but then the human gives it the meaning. And then when the human steps in to give it the meaning, oh, great, you've got magic.

[01:04:39] But I don't know that I've seen this sort of environment where it's just all AI running around doing all these things and everyone's like, oh, my God, it's amazing. I just don't see that because we give things meaning. And I just I don't know. I don't think that's going to change. So that's my take. I think that's that's excellent. Well, it's been fantastic having you both on as guests again on the podcast.

[01:05:01] If you again, as I said earlier, if you haven't gone and listened to their prior episodes, please go check out Angela and Sandra's prior episodes on the podcast. And I want to say again, congratulations to you on Angela on your new role. And Sandra, do you have a title for your book yet, by the way? No, well, we have like several. But I was told that we will not title it until the very, very end. So check out Sandra's amorphous blob of a book at some undisclosed time.

[01:05:30] But if people want to reach out, where can they find you guys? LinkedIn. Same. All right. Well, you've been listening to Directionally Correct, a People Analytics Podcast with your host, Cole Knapper. And today's guests, Angela LeMetton and Sandra Laughlin. Thanks for joining me today, guys. Thanks. Thanks for having us.