[MUSIC PLAYING]
Hi, everyone. Welcome to Mercer's podcast series on The New Shape of Work. My name is Gord Frost. I am Mercer's global reward solution leader. For those of you who are expecting to hear Kate Bravery, who is traditionally the host of this series, you may remember that I was a guest of hers on a recent podcast, and in that podcast she asked if I would guest host this one, which I I'm more than happy to do.
Today, we're going to talk about total rewards and the impact of artificial intelligence and generative AI in the total reward space. A great topic for discussion. I'm really happy to be joined by my colleague Sean McHale. Sean is our European rewards leader, and together we thought we would have a discussion about what we're seeing in the marketplace, what organizations are doing, what some challenges are that they're facing, and everything related to generative AI within the field of total rewards. So, Sean, thank you so much for being here today.
Thank you, Gord, for having me. Excited to be here.
Awesome. No. I think we'll have a great discussion. I thought we would start off by saying that, a lot of the clients that I talk to I find it's really interesting because their organizations are absolutely experimenting with AI, they're implementing generative from the perspective of marketing and client engagement, improving operations, improving workflow internally. They're experimenting with all of these different AI applications, yet they seem reluctant to actually start leveraging generative AI within total rewards or the rewards function specifically. And so I'm just wondering, are you seeing the same thing? And if so, why is that do you think?
Yeah, it's a good question, Gord, and I think I am seeing the same thing just from discussions with some bigger clients, and small clients, and medium-sized clients all around Europe and the UK. And I think first and foremost, many HR and total rewards functions or teams are kind of adopting a wait and see. So let's wait and see what our technology teams will come up with, and then we hang on to their coattails, and go with it, where I think that's probably a mistake as well because, I mean, remember what happened with these large HRS or HRM implementations where HR adopted that wait and see approach, and then got left behind because the CIO or CFO made a decision, and the technology then didn't work for a HR or for rewards, and it became a bit of a mess, and so I think that's part of it.
I also think there's challenges around data. And we're all in HR and reward, so we work with data all the time, and our access to vast volumes of data is growing really, really fast, and obviously our ability to extract knowledge from that data as well. I was looking at some stats earlier today around data and the volume of data globally. And in 2020, we had a global volume in zettabytes. So a zettabyte is 1 trillion gigabytes of about 50 to 60, so 50 to 60 zettabytes in 2020.
And by 2030, that's going to reach about 600 to 700 zettabytes, and by 2050 reach about 50,000 zettabytes. So that's the macro level. But look at every organization as a micro, and all of the different pieces of information that are attached to every single employee in those organizations, it gets pretty vast. So just I think part of the reluctance is really knowing to get the best from AI, we need to-- and organizations and our clients need to have the right data, the best data, and data really organized very, very well, otherwise it's not going to work. That's really the foundation that very baseline need for successful AI. Bad data in equals bad AI.
Yeah. And I think on that point, with employee level data, you're often dealing with confidential data as well. So you're dealing with compensation-related data, salary data, other personal data that HR is the custodian of. So there's this challenge of while the vastness of data is impacting all the functional areas that are leveraging AI, when it comes to HR and total rewards specifically, you're also dealing with confidential personal data, and so I think that's driving a bit of the reluctance as well.
Yeah. It's all very personal data. Obviously there's regulation GDPR in Europe and elsewhere. We now have the European AI Act as well, which is, I think, just adding to the risk averseness around how we use AI and how we can use it, and it's just giving us, I guess, more opportunity to push back from a rewards perspective. And that kind of lack of technical expertise as well from a rewards perspective, there's definitely a gap in expertise there, where--
In AI and technology, right?
Exactly. Yeah.
But there are others in the organizations that are well versed in AI and other skill sets across the business who know how to use data. So we bring them into the total rewards team to help out, to implement.
And I think from that perspective, while I understand the desire to hesitate or maybe wait and see, take the wait and see approach because of the risk involved, I think there's also a risk in waiting. A lot of executives, and CEOs, and CHROs are out there saying, hey, wait a minute, the potential benefits of the use of generative AI are so great. What are you doing? Why haven't you leveraged these? Because if other organizations are and your organization is not, are you falling behind? So similar to your analogy of the HRS implementations, if you let others take the lead and don't embrace it yourselves, you run the risk of being left to become obsolete.
Yeah, totally. And we are seeing that c-suite are pushing for more and more use of AI to increase productivity, to increase revenues, to increase profitability. And we'll see if it actually does that at the end of the day, but right now the belief is, yes, it will. And so companies that aren't embracing AI are running the risk of not achieving those financial goals and other goals. And so I think, it's here. It's here to stay. You better get on the train now one way or another, otherwise, you'll be left behind.
For sure. And so for organizations that understand than that, but they're not sure where to start because it can feel overwhelming. Where do you start to just dip your toes in the water? What are some good examples of things that you've seen other organizations do to start to leverage generative AI or the possibilities of AI within the total reward space that maybe not changing everything, but changing one piece of what they do, or one aspect of their job, or one element of total rewards?
Yeah. I think there's a few good areas where we can explore where there's not as much risk in the data that we're using. So thinking about writing or reviewing comp and benefits policies, and procedures and processes. So AI can use internal data to draft broader philosophies, or strategies, or even policies, or kind of align policy, streamline policies around the world where they already exist, ensuring that they're obviously fair and competitive. They align to the right markets in which they sit, and they're standardized, and can really more fairly recognize employee performance.
And then obviously employees can access them once they're built into AI as well. It could be loaded into a chat bot once we tell employees that it is being loaded into a chatbot because that's part of the EU AI Act as well. Also job descriptions, we're seeing this quite a bit actually. And so GenAI can help organizations to really write job descriptions. You're not starting with a blank piece of paper and there are obviously technology firms out there that can help do this already, but now it's totally democratized.
And so any rewards, colleague, or team member can actually write a job description leveraging the career stream descriptors from a Mercer or our competitors, leveraging the family descriptors, the subfamily descriptors, anything we need that can be loaded into the AI to inform a job description, and then set out in the right kind of format using the company tone culture, et cetera, to build into it and make things very, very consistent across the board. That's really where job descriptions, as we know, fall down right now because we've got many, many people across the organization writing job descriptions on their own. AI should be able to do this very, very consistently.
Thinking about job descriptions, job architectures as well. And so existing job architectures can be enhanced using AI. Again, looking at the grades that feed into them, the job families, subfamilies, career pathing, and even just creating job architecture, maybe even mapping skills to those jobs in the architecture as well. That's where AI can help a lot, and speed things up.
And even thinking about that at Mercer, we've obviously leveraged AI for quite some time now in our survey data collection process. So we use the Mercer Data Collection tool, or MDC, or GDAP for some organizations around Europe which looks at job titles submitted across the board by organizations around the world, figures out what the pattern is, and then we'll recommend a match to the Mercer Job Library, so speeding up the whole process.
Now, all of these are obviously it's streamlining a lot of these more basic processes that take a lot of time for total rewards teams, or did take a lot of time in the past. But they're still not perfect, they're only getting us probably 80%, maybe 85% of the way there, so there's still a need for that human interaction, that human review, peer review to make sure that the output is absolutely right at the end of the day, don't trust the AI yet.
Yeah, but I think that's interesting because to your point earlier, even though some of these things like standardizing and rewriting job descriptions, or you reviewing job architecture and mapping to job levels and stuff like that, a lot of that seems like it's kind of foundational, and it is. But to your point earlier, it's so time consuming. If you're a large multinational organization with thousands of employees, across thousands of jobs, with thousands of job descriptions, you almost would never tackle this work just because it's so time consuming and so mundane.
But with AI, if you could get 80% to 85% of the work done through automation, and then your total rewards team only needs to get it that last 15% of the way across the finish line, it's a huge time savings. And there's a huge benefit to the organization because now you have that consistency in nomenclature, in wording, in culture, in tone. You've got the consistent inclusion, as you said, of skills, or competencies, or career leveling. So that then opens up the door to leveraging much more AI because as you said earlier, if you've got garbage that you're putting in, or the data quality is not that good going in, the possibility of leveraging it is only so good. But once you've got higher levels of quality across all of your data sets, then you can actually, open the door to much more sophisticated analysis and additional benefits, I bet.
Exactly. That's exactly the point. Get the base right, get the foundation right, and then you should see the benefit downstream. Again, more sophisticated outputs, more sophisticated analyzes that we're able to garner from right information going in.
So what do you think-- so for the organization that's done some of this foundational work, and has their house in order, and all of that stuff is in good shape, what are some of the say, more sophisticated types of analysis that you think organizations could start to tackle, and/or are you even seeing a few more advanced organizations that are venturing there already?
Yeah, there's a few that are there already. There's many that aren't. So if you're not there yet, or if the listeners aren't there yet, don't worry, there's still lots of time, but there are a few that are definitely exploring a little bit more and how I can help. And so thinking first of all, first and foremost, pay transparency, we know is obviously very big around the world at the moment. But the EU pay transparency directive coming through and pay transparency, more prevalent in many US states.
And so from a pay equity perspective, AI can actually run a pretty decent pay gap analysis. If we have the right job architecture feeding into it, and if that our employees are sitting in the right jobs in that architecture, we should be able to use AI to run a fairly basic pay gap analysis. And then also, maybe create some fair pay algorithms that look at qualifications, education, experience levels, performance metrics, and to ensure broader compliance and equitable pay going forward as well. And so that's definitely one area that I think we should be seeing a lot more traction right now, and I think we are in certain organizations.
Also in some organizations, in broader rewards processes, AI can recommend appropriate rewards based on performance metrics, based on other feedback from colleagues, and then it can automate the reward allocation process. So think about annual merit process, pretty time consuming, lots of people involved, and potentially lots of people involved, and those people also have bias built in. Maybe unconscious bias obviously as well, but that could mess up all of the hard work we've done from a pay equity and pay transparency perspective.
AI if trained properly, can remove that bias and can streamline the process once we're feeding it the right information. Again, and it should make it a lot easier and much quicker to do.
Yeah. Interesting.
So much data feeding in there as well though, Gord. You've obviously got performance ratings, potentially. You've got current pay levels, comp ratios, pay ranges, actual feedback, maybe from colleagues, the funding available. So there's lots of different data points coming from so many different systems as well that make it complex but doable if we've got it ingesting the right information.
Yeah, and it's interesting because you raised a point about performance ratings and colleague feedback, and it reminds me of a situation where I was speaking with another organization recently. And they had started using AI, and so this was a technology organization, and so they had built this themselves. But it was an AI-enabled performance rating and performance feedback system where at a very regular interval-- because a lot of organizations only do that once or twice a year because it's very time consuming to collect the feedback, and synthesize the feedback, and request the feedback, and all of that sort of stuff.
What this organization had done was they were using AI to be able to do an analysis of the network of individuals that a person works with regularly through who they correspond with most in email, and in Teams messages, and stuff like that. And then in an automated fashion, it would send an automated message to all of the people that the AI system identified as being the people that you collaborate with the most, with a simple set of questions to ask, that only took five or 10 minutes for people to respond to both with a choose the right response from an ABCD, and then adding in some wording as well.
And then the AI tool was able to summarize, and also summarize the responses, and summarize the verbal responses like that people had entered in to really quickly summarize an overall performance review for people. And so it allowed them to do it much more easily, much more frequently, much less time consuming, and get you real time feedback than you could act upon over the course of the year. So again, just one example of how people were using automation, some elements of the generative AI, other elements of technology to really improve a process that already exists, but make it much more seamless, and much more frictionless within the organization.
Exactly. And that way as well, it's a really good example, but it should allow that organization to identify those more high potential employees as well, and make decisions about how we might need to act to retain them or do something additional for rewards perspective.
Or potentially recommend developmental opportunities for them, or different career paths. I think once you then link them to the other HR processes you have in your organization, the potential really becomes limitless, which is pretty exciting.
Exactly. And the potential for proactivity rather than reactivity, which is where HR normally gets caught up.
Yeah. So if we think about that a little bit more like the exciting future applications. So stuff that maybe organizations aren't doing today, but that they could start to do in two, or three, or five years assuming that they get the foundations in place, and get their house in order, and all of that kind of good stuff. What are some of the more exciting potentials that you see in the future as organizations begin to really leverage the potential benefits of generative AI?
So I mentioned that the chatbot example, and I think we will see improved virtual assistants across the board. I mean, we're all using ChatGPT bots all the time now. So I think as I mentioned, can take a benefits handbook or whatever, 200 plus pages of it, and probably many versions from around the world as well, ingest that and then we can actually train it to answer questions based on the policy information therein, so newly acquired knowledge for the AI.
And then rather than employees calling somebody in HR, or somebody on the benefits team, or somebody in a call center to answer a question, AI can give them the answer, so they're just interacting with the computer. Obviously, there might be a need to escalate in some instances for more pressing needs from the employee side, but hopefully they should be fewer and easier to manage from that perspective.
And just an example there, I was talking with a colleague, and they were saying, what if the AI chat bot already knows when you call your gender, your relationship status, whether or not you have kids, what your prior-- what your benefits usage is, or what benefits you have access to. It would already all of that about when you call, and you would just have to call and say, I'm having a baby. And it would know based on your gender if it's you physically having the baby or if it's your partner, it would if you already have other children that are covered under your medical plan. It would already know all of this stuff and could just give you the answers that you need rather than having to go through several levels of policy to get to that.
Exactly. And likely several phone calls or several other interactions with somebody else, which is obviously not the employee experience that we want to have for our employees, and that's [AUDIO OUT] anymore either because they see in their personal lives.
Also speaking about personal personalization of total rewards, so I can ingest vendor policy information to answer questions for employees, as I mentioned, and can personalize recommendations based on all of the prior behavior data giving those really personal outputs to employees, and really understanding who you are, where you're coming from, and exactly what you need, and very, very different to current state offering where likely that's very first question that's going to be asked.
Well, and talk about the current state. I was at another conference recently where somebody gave it the statistic that the average employee will spend more time selecting the outfit that they're going to wear that morning to work than they'll spend selecting the benefits for themselves and their families.
So this fact that if rather than having them read through all the material, and understand what the benefits offerings are, and then have to select all the different options that are available to them, if the AI tool or assistant can already everything about them and then propose the best total rewards package for them without them having to get educated on everything, without them having to read through everything, without them having to make the choices, you're going to get better outcomes at the end of the day, right?
So I think not only then are you getting a better outcome for the employee because you're really personalizing the total rewards to exactly what they need, or will benefit from most, you're actually ensuring that the company's money is getting the best ROI. So you're getting the best return on your investment then and you're spending your dollars in the most optimal way possible because you're getting the best outcome for each employee.
Yeah. Optimizing it for every single employee. And obviously, we know from a comp perspective and benefits perspective, we can align it to every single job, and understand what the market is. This is kind of going to the next level and understanding what each employee wants as well, which is phenomenal there. And then also thinking about mobility, global mobility, and that process, and what we pay expats.
Usually, multiple data inputs needed there, but AI should be able to take all of those disparate data sources and employee information to suggest the appropriate compensation needed. So looking at cost of living, family status, and tax legislation, other legislation in the country that the person is moving to, take all of that information and make the right recommendation for the expat to give them the right experience as well. I think that's probably maybe a forgotten grouping of employees for many organizations.
Well, and I think it's-- they're forgotten because it's usually small. Your expat population is not big. But if you think about the amount of time and effort spent on expats or the risk of if you get an expat assignment wrong, I think, it takes a disproportionate amount of time and resource to manage expats. And I'm even I'm thinking beyond the tax issues, and the comp issues, and that sort of stuff.
But again, if the AI tool knows about your family status, knows if you've been on expat assignments before, knows what languages you speak, it can help to do selection of appropriate housing, or schools for an expat kids, or cultural and language training like all of the other elements of the expat experience as well, if those were fully automated and curated, imagine what a better experience you could get at the end of the day.
And that's what it ties back to. That's what we're all aiming for, is the best employee experience for our colleagues, and AI should enable us to get there. But, again, thinking about everything we're talking about, there's vast amounts of data, and data points that are feeding into what's needed here. So back to our initial discussion around having that baseline, having that foundation of data, that's absolutely key. We got to get that right first, otherwise, we're going to struggle.
And I think to what you just said, that's stuff that a total rewards team needs to tackle right now, right? That's stuff that they can be doing right now today because if they're not, they're never going to be able to realize the benefits of what generative AI can present in the future. So all of these issues around a more tailored employee experience, more personalization, more optimal spend of company dollars, all of that stuff is a possibility, but only if you start getting the foundations in place right now.
And I think that because this is going to move so quickly, total rewards seems that are not working on that right now, like we said earlier, at the start, are going to get left behind. So what are some of the things that you think that they act like a total rewards leader or a total professional should actually be working on today to start to move the needle on this?
So again, I think, it's the basics. And so thinking about obviously getting involved to start with. And let's not leave it to our CIO, and CFO, and others to make decisions. We need to be involved. We need to be at that top table, having discussions around how we're going to govern AI, and the impact it's going to have on our employees, and our jobs, and how we get work done in our organization, and that's more strategically for sure, but then more basics. Are our policies up to date? Do we have formal written policies in every country? Are they globally consistent or not? What can we do to get them there before we then feed them into the technology and for it to ingest them and then hopefully spit them back out as a chat bot or something else?
Our job models or job architectures of the days. What about our job coding? When was the last time we reviewed a job architecture? When was the last time we reviewed our employee mapping? And whether they were sitting in the right jobs or not? Do we have the right number of families or subfamilies in our architecture? Is our reporting hierarchy up to date? Our job descriptions adequate? When did we last evaluate our jobs? When did we look at our organization sizing to evaluate our jobs? Our pay range is up to date.
I mean, the list goes on and on and on because we always deal with so much data, but we don't really think about it as we're kind of caught up in the weeds, and in our day-to-day jobs. But we're dealing with vast amounts of data all of the time, and just getting, ensuring that data is as up to date as possible before we think about leveraging AI, I think that's the key here, and so doing it all.
And I think, Sean, to that point, the total rewards team with an HR often is the custodian of that employee level data. The data around rewards, the data around what job they're in, how that job maps to the job architecture, all of that stuff often resides within the total rewards team. And so I think it's a double edged sword in that the rewards team has the ability to impact that right now. They can start working on that today, making sure that they've cleaned up their cleaned house, that the foundations are strong, that they're ready to start leveraging AI.
At the same time, as I said earlier, because this is moving so fast, I think, there's going to be an expectation from leadership, whether it's the CEO, or CHRO, or whoever, that the organization is going to start to move in that-- to move to leverage AI. And if they get stuck because they realize that at the end of the day their underlying data is not of the required quality, they're going to come knocking on the door of the total rewards leader and say, hey, what's going on here? Haven't you been doing your job? Why haven't you-- why are we not ready to move ahead?
So I think it's both of benefit, but it's also an imperative that the total rewards teams are addressing these issues now proactively because if they don't, they're going to get-- they're going to be stuck.
Exactly. Yeah, absolutely. So having a voice, and making yourself heard now is key, and learning the technology, or at least learning about it, and how we could deploy it within total rewards. There's lots of different out there at the moment. There's 300, maybe even 400 now, different platforms. Now, a lot of them are still in their infancy, but that means they're not ready yet for prime time so let's explore, figure out what could be best fit for our organization, or what kind of tools we can actually take from each of them to implement in our organization.
Yeah. I've even seen a couple of organizations where they've been growing their HR analytics teams or ensuring that within HR and within total rewards, they've got some people that have that analytics skill set because, again, it's a skill set that you may not have in HR that often, but you probably have it in other parts of the organization. So again, I've got one client that I've worked with where they've actually moved someone from their IT team into HR as their HR analytics specialist to help build out that capability within HR and total rewards specifically.
Yeah. Makes a lot of sense. And I think diversifying the skill set that we have within total rewards, within HR makes absolute sense. How can we get the best AI? We need additional skill sets to figure that out. And then finally also, Gord, thinking about jobs, jobs themselves. So as comp professionals, as rewards professionals, we're used to market pricing and determining our benchmarking jobs, determining the market rate for jobs.
As AIs democratized more and more, and everybody, every single job starts to use it, how do we then understand what to pay those jobs? Are we paying the AI or are we paying the person sitting in the job? And there probably are going to be some jobs that aren't impacted by AI, I think, more maintenance jobs, electricians, plumbers, those types of trades jobs, excuse me, where AI won't be as leveraged or potentially mightn't be as leveraged. It probably will be. We're not the experts, but their market rate then increase because they could be in more demand. And so, again, just another perspective to think of from a rewards perspective.
Yeah. No, I agree. As jobs change, the market value of those jobs will also change. In total rewards, teams are going to be expected to be on top of that. Those are all really good points.
Yeah.
So maybe just to wrap up, again, I think a lot of this is exciting, really exciting. It can be daunting or overwhelming to organizations. We've given a few areas where if they're not sure where to start, what they might do first. But are there areas where you think Mercer can help, or where you've seen or others of your colleagues can dive in and help organizations to get started with this stuff?
Absolutely. Well, we're compensation consulting firm and benefits consulting firm, and so we know what best practice is from let me call it an organization perspective. So ensuring your data, your company data, and your employee data is in the best shape possible, we can help with that for sure. So job architecture, job leveling, ensuring global consistency, and more is definitely a space that Mercer can help.
Obviously also I mentioned pay transparency, I'll mention it again. Pay gap analysis, we can identify obviously any existing pay gaps, and then also any bias that currently exists in the process. And so they don't get perpetuated when we start to build AI into that process through bias or otherwise.
And then as you mentioned, I think bringing in data scientists, our own data scientists, to help get the best from the system or systems. And so developing a roadmap of how we might want to leverage AI and HR in our total rewards processes going forward, what's right for our organization or for your organization? And might not be right for others, and so let's talk it through and see where exactly it should slot in, and where we can get the best from it, and we have a dedicated team that are doing that right now as well, so many, many places that Mercer can help.
Nice. Yeah. So look, Sean, thank you so much for joining me today, and sharing your thoughts and experiences with our listeners. This is a topic that I'm sure we could go on and on. And again, six or 12 months from now, they'll probably be even more to talk about a new use cases and new evolutions, so this is definitely a topic that will be evolving quickly.
But I'm sure our listeners have benefited from just hearing your thoughts, and experiences, and our conversation today, so thank you for that. Clearly, GenAI is impacting the world of work in a really dramatic way, and total rewards is no exceptions. And so I think total rewards teams really need to, as we said earlier, get on this now and start to take action now so that they can play an active role, in the process within their organization.
Yeah, totally agree, Gord. And it might just be baby steps now, but I think this AI will continue to gain traction and snowball and snowball. And so jumping on now is critical, and needs to be done, so why not now?
All right. It has to be now. It has to be now.
Yeah.
So listeners, thank you for tuning in. If you're interested in this or other topics associated with the new shape of work series, please do visit our interview series on mercer.com. If you want to learn more about any of the ideas that we talked about today, feel free to connect with your local Mercer consultant, and thank you for listening.
Thanks, Gord.
All right. Bye, Sean.
[MUSIC PLAYING]