AI in HR: Dayforce on Why Governance Helps Companies Move Faster

[00:00:03] - [Speaker 0]
What if the very rules that were designed to control AI are actually what allow businesses to move faster? Well, HR may be one of the highest stakes places to use AI because these systems can ultimately influence who gets hired, who gets promoted, and who gets paid, and how much. But today, I'm thrilled to be bringing on David Lloyd, chief AI officer at Dayforce. And together, we're gonna talk about why companies cannot outsource accountability to tech vendors and how leaders can create safe spaces for employees so they can use AI and also bust a few myths. I wanna talk about why good governance can turn speed, trust, and innovation into allies rather than competing priorities.

[00:00:55] - [Speaker 0]
I don't wanna reveal any spoilers. We got a lot we're gonna be talking about today. So enough from me. Let me introduce you to my guest so we can get this conversation started. So thank you for joining me on the show today, David.

[00:01:08] - [Speaker 0]
Can you tell everyone listening a little about who you are and what you do?

[00:01:12] - [Speaker 1]
Neil, thank you for having me. My name is David Lloyd. As we've already determined, probably not the David Lloyd who's one of the well known tennis players in in The UK, but David Lloyd nonetheless from Albuquerque, New Mexico. So we'll go from there. Neil, I love building and growing companies, especially with AI.

[00:01:32] - [Speaker 1]
It's been probably almost twenty years of my past life that I've spent in the using and applying artificial intelligence in the enterprise solutions that we've built out. And I think know, when I think of myself in all that time, I've had an amazing partner now through all of life's trials and tribulations who supported that craziness and two adult kids that actually let me see the world through their eyes, which is a very different place than it was through my eyes at their point in their careers. So that's a bit about me. I'm pleased to be part of Dayforce. I was fortunate to come here about five years ago.

[00:02:13] - [Speaker 1]
And I love data and AI, and that's exactly where they want to drive. They really want to build on the lead that they already had, the momentum in such an important space being human capital management. And so for me, I lead over 500 people in our platform organization and engineering, product, and really drive the development of new AI solutions for our data science teams and others. And it's an exciting place to be.

[00:02:43] - [Speaker 0]
Wow. And I just love the fact that you've been working in AI for more than twenty years, long before it was considered cool in the mainstream area. But, obviously, fast forward to present day, we often hear around AI as a tool for drafting emails or taking meeting notes, but HR could be one of the most powerful and high stakes applications of AI today. But as someone right in the the eye of the storm storm, so to speak, why do you think that is?

[00:03:12] - [Speaker 1]
I think it's partially because of the complexity. I think when you look at I use the term human capital management or h HR right off the top. When you think of payroll, you think of workforce management, you think of engagement, performance, compensation, benefits, all of these very complex topics and complex workflows and things of that nature, you realize that for so many of the HR organizations, it's a very daunting task. I mean, most of their budgets have not gotten larger over the last number of years. They've stayed the same.

[00:03:46] - [Speaker 1]
Yet the expectations for what they have to achieve and what they have to get from those systems is huge. So to me, I think it's the complexity. But also the thing that I think adds to that complexity almost no other way except maybe health care is the nature of the data itself. This is compensation data. This is data about my kids as my beneficiaries, my spouse.

[00:04:12] - [Speaker 1]
So we have, within these HR systems, some of the most trusted data in the world, data that is very private to us. So we have both the desire to do exciting things and the whole side of compliance, security, and trust. And there's a lot of questions around AI right now and what that looks like.

[00:04:34] - [Speaker 0]
There really is. And a lot of AI conversations right now are focusing on productivity gains, but we are starting to see AI change what jobs actually look like day to day. And I think one of the things we often forget about is it's real people with families and and loyal to enterprises that are right in the heart of this pace of technological change that we've never seen before at this kind of scale. So as responsibilities shift, how can HR leaders better help their employees feel like they they actually have a voice in that transition instead of feeling like it's a change that's just then thrust upon them or or happening to them and they have no say in that? Because we'll talk about the tech a lot today, but I think this people element is so important too.

[00:05:19] - [Speaker 0]
Right?

[00:05:20] - [Speaker 1]
I think it's such a great point, Neil. I'm reminded of a colleague of mine when we're out speaking at events. Loris Apps talks about this. He heads up product for our AI area. And he uses he uses the term put your mask on first.

[00:05:35] - [Speaker 1]
So whenever you get on an airplane, they always talk about putting your mask on before helping others. And I think it's such a it's it's a great metaphor because what we need for our HR leaders to do is first have comfort or be comfortable with AI to build up some of their own literacy around that. Because, arguably, I think one of the most important roles we have in the organizations today is HR. They actually have to help an organization, to your point, through one of the most massive change cycles we've seen from a technology impact. And so we really need our HR leaders to feel comfortable with the technology.

[00:06:17] - [Speaker 1]
They don't have to be experts, but they have had to understand it, be part of it, work with it. And so that they can better understand the impacts that are going to happen, to your point, to the workforce in general because I believe roles will change and evolve over the next number of years. It's a process I call compression, which basically means there's individuals doing roles in which what we really want to do is they're actually absorbing more and more parts of other roles into theirs, which means there's smaller numbers of those roles around, which is the compression part. But more importantly, people start using AI to actually augment the way they work, which actually extends the job that they used to do to almost a broader job from that standpoint. So I think it can have a lot of different impacts on people today that makes them uncomfortable with how fast it is actually shifting.

[00:07:15] - [Speaker 0]
And one of the reasons I really wanted to ask you that is I think many career ladders and performance frameworks that you see in corporate America and indeed over here in The UK, they were all built long before AI entered the workplace. So what does it actually look like for maybe a a chief human resource officer listening who is challenged with rethinking how performance is measured, how it's rewarded in a in an AI enabled organization? It feels like there's so many so many things need to change there.

[00:07:43] - [Speaker 1]
There there are some. I would say some things are absolutely consistent. The ones that are consistent are one of the the conversations I got into about two weeks ago was with a group of VPs of HR. And I said, today, if you were hiring someone who didn't know Word, Excel, PowerPoint, Microsoft Mail, or or Gmail, would you actually hire that person? And the answer was absolutely not.

[00:08:09] - [Speaker 1]
Yeah. So although the acceleration of AI has been much more pronounced, the question is, would you hire someone who didn't know how to use that set of tools within their environment? When you talk about that, how do I actually help those individuals from a literacy point of view and things of that nature accelerate their learning so that they feel that they're part of the change, not sitting on the sideline of that change? So yes, there's certain expectations of technology that get thrust on us. Absolutely, we've compressed time with this, and it's been happening over such a short period of time.

[00:08:47] - [Speaker 1]
But we really need to take that kind of approach, I believe, in thinking about how we help others. And that all comes down to literacy, which is still very, very low.

[00:08:57] - [Speaker 0]
Yeah. It's such an important point. And I think one of the problems that people listening may have encountered is that the AI strategy in their organization is discussed at the leadership level, but them as employees feel its impact in their everyday decisions that it shapes. So how do you ensure AI is improving that whole employee experience, not just and feel like they're not just treated like numbers on a spreadsheet?

[00:09:23] - [Speaker 1]
I think one of the things that we've seen a lot of recently is actually how you bring groups of individuals that are using AI together. One of the things that organizations have to recognize today is that everybody's carrying it around in their pocket, in their mobile device. And so if you're not providing an environment that's embracing it to help people learn safely in, let's say, a safe sandbox, they're going to be using it on that device anyway. So what you wanna do is two things. You want to actually find people that are using it today and bring them more or less in the tent with you and understand how they're using it, how they're thinking about it.

[00:10:01] - [Speaker 1]
Because if everybody's using AI capabilities differently and the organization has no process to harness a consistent way of applying it, you're probably losing a lot of really great lessons learned that people have adopted on their own. So I think there's an opportunity to learn from the crowd being the group of employees in that organization, how they're using it. I would say too, not to punish individuals for using it. I I have a lot of concerns with the sensitivity of data that people may be doing improper things with the data, not because they wanna do improper things, but they're actually trying to do their work better. People are saying you have to use AI.

[00:10:38] - [Speaker 1]
You have to begin applying this. And arguably, they may not have been given the guardrails about what is appropriate, what content can we use with these models. Do we have a place where I can use it safely? So I think the HR professionals really have to help with the CIO, the CFO, and others in the organization to create these safe spaces to not only bring everybody into so they can learn, drive the literacy rate, but also take the lessons learned out of that and allow them then to apply those organizationally on across the board basis. And that's when you're going to get truly better outcomes from the use of AI.

[00:11:18] - [Speaker 0]
And if we zoom out for the moment, I think AI is being used in so many different organizations now to help decide who gets hired and who doesn't, who gets promoted, and what somebody actually gets paid or compensated there. So when you when you put all all these things into one box or one melt import, what what responsibility is that placing on organizations? Do you think most companies fully appreciate exactly what's at stake and and what AI is managing and and their role in this too? What what are you seeing?

[00:11:48] - [Speaker 1]
A couple of things. I think one of the things that's most important about that is we believe firmly around the human in the loop. Yeah. If you are using AI as an individual on behalf of your organization, you are accountable for the outcome of that AI. So you made an earlier comment about it's it's doing an edit of my draft or something of that nature of my email.

[00:12:13] - [Speaker 1]
Although I think those simple simple examples are long gone for us. They they they're they're almost become the norm. The point is if I have AI write an email and I send that email without reading it, that's on me. I own that. So I think the responsibility doesn't leave us simply because there's this new capability called artificial intelligence.

[00:12:35] - [Speaker 1]
Now that will change over time with agents and things of that nature taking on certain responsibilities for different processes. But as individuals that are using these tools in the day to day work that we do, we are still accountable as the reviewer, the manager, the supervisor of the AI that we're using in the work that we're actually promoting and actually providing to the organization. That doesn't disappear.

[00:13:01] - [Speaker 0]
And for any business leader listening to our conversation today, they wanna follow in your footsteps there. They wanna make a difference and do things the right way. How should their organization build confidence that their employee data is being handled responsibly, especially when AI can influence every decision that can impact their career? A big responsibility again, isn't it?

[00:13:24] - [Speaker 1]
It is. I would say that most organizations need to step back. I think there's been this mad rush for every organization to claim that they're using AI. It doesn't mean you're using AI effectively. You may be wasting a lot of the money.

[00:13:38] - [Speaker 1]
It doesn't mean that you're using AI thoughtfully, which I think is the other problem, and that is the one that I'll probably hit on here more than anything else. Organizations today, especially leaders, if you look at the leader literacy for AI, it's another scary figure that I won't draw too much attention to. But the leaders who are telling everybody to work smarter, not harder with AI are themselves not necessarily working smarter, not harder with AI. So I think we have to make sure that literacy is across the organization, and they understand the tools and the capabilities and the expectations they have and what the employees would be using in their jobs. But I think the most important thing we can do is two things.

[00:14:18] - [Speaker 1]
If I if I said two things to the HR leaders and business leaders, it would be this. First of all, provide a safe environment for which everybody in your organization can actually use AI. So they're not using dark AI where they're using it on their own and providing things into it that maybe they shouldn't. Secondly, define. This is all gonna come back to data.

[00:14:41] - [Speaker 1]
Define what data is acceptable to use in that sandbox, in that safe environment for those individuals? I if I have a spreadsheet of all my employee information and their salaries, can I use that? And you might be able to in that sandbox safely and expect that in the end, I'm not necessarily providing it to a model that's training on it somewhere else. So define your your classification of your data, what data is permissible in what models. Because you may be able to use one that's outside of that that safe sandbox, but you can only put material that's public to the company in it.

[00:15:19] - [Speaker 1]
That would be okay. It's probably already trained on it. But things that you are doing internally, then you want tight definitions or guardrails around what is acceptable data that can be used in that environment and label it accordingly so people know what to use it for and how best to use it. And that would be the easiest way to start.

[00:15:42] - [Speaker 0]
It's not your fault that your agentic AI systems are acting outside of compliance. There are just too many datasets to guardrail them all. But with Denodo, you can now organize your hundreds of data sources into one layer and govern your agents with a single approach. Try it now with Denodo by visiting denodo.com to learn more. And I think we should also highlight, at the same time of everything we're talking about here, we're also seeing new AI related regulations emerging, especially here in The UK and Europe.

[00:16:20] - [Speaker 0]
On the other side of the pond there, in in The US, we've got states like Illinois, Texas, and Colorado leading the way, but who should ultimately be responsible for overseeing the AI use within an organization, and what should leaders be doing today to maybe stay ahead of some of those potential risks?

[00:16:36] - [Speaker 1]
Of the things we do and we spend a lot of time with our customers on is actually how we apply governance to the AI that we build for our customers. And we come from a fairly strong place. We're ISO 42,001 certified, one of the few organizations globally of our type that actually is certified for our AI governance by ISO and we're Nesta tested as well. And we actually do third party external audits of our higher risk AI. You just mentioned a couple of use cases there in recruiting.

[00:17:09] - [Speaker 1]
So when I think of things like recruiting and I think of different ways that AI is used, what we do and what we recommend as well is you hold yourself to the higher standard that you can for the way in which you think about artificial intelligence. For example, the EU AI Act is very specific on how models are evaluated, what are low risk models, risk models, and high risk models. If they're high risk models, then how do you treat the output from those and hiring decisions and things of that nature in the future? Just to set a context for setting that high watermark for how you evaluate, not can you, but should you. And I think those are two really important parts of that.

[00:17:55] - [Speaker 1]
And I think that's the starting point for organizations around how they're using artificial intelligence. One of the things that we believe is really important is it should be always up to the individual. We call it AI by choice. So the individual organization should always decide when AI is going to be enabled through whatever platforms they're using in their environments. You should never be surprised with the CIO to see a new AI feature in your software that all of a sudden, 600 of your employees have, and you didn't know what's going to get turned on tomorrow.

[00:18:30] - [Speaker 1]
That should never happen in our view. And so we believe very much in AI by choice. You need to opt in, not opt out so that employees know that what they're provided with is safe to use, again, those guardrails.

[00:18:44] - [Speaker 0]
And you mentioned ISO and this a few moments ago, and I know you were recently accredited with these. So tell me more about what that means to you. And and for leaders that who aren't compliance experts listening to our conversation today, what do these milestones actually prove about how you build software and and why it matters so much to your customers?

[00:19:06] - [Speaker 1]
The compliance was a result of a a process we've been on for the last five or six years in ensuring that the AI that we're actually providing to our customers is safe, reliable, and trustworthy. And I think so very proudly, at the end of last year when we were received the certifications, it was a great thing. But what it does is it actually sets the right discipline for the organization. Now you'll hear people say, oh, David, AI governance. David governance, it takes so much time.

[00:19:36] - [Speaker 1]
It slows our innovation down. That's BS. To be very, very factual about that, we actually turn any new AI idea over in less than a week. Wow. But that's because we bring the right people together.

[00:19:51] - [Speaker 1]
And the case you're talking about, the CIO, the CHRO, perhaps a chief privacy officer, or someone with a responsibility for privacy in the organization come together, and they can actually very simply have a process that brings new ideas in from any employee in the organization. We've take in any idea in from over 9,500 of our employees. And, actually, it's six business questions. That's all we ask. That's all we need to get the process started.

[00:20:21] - [Speaker 1]
And there's really three questions every leader should ask themselves when they're looking at these new ideas. Does it need AI to solve the problem? That's a big one because AI adds lots of complexity to your point. You have biasness and all these other things, transparency that come into play. So do I need AI?

[00:20:43] - [Speaker 1]
If I don't need it, I've simplified my life in some ways. If I do need it, fantastic. We understand that. Then the next big question is, do we even have the data necessary to actually bring that idea to life? So I may want to build this brand new system that predicts home prices in Downtown London.

[00:21:03] - [Speaker 1]
But if I don't have any data on prices for homes, it's a great idea, but I can't execute on it. So you may be able to buy that data or do other things with it, but you have to really define, do you have the data you need in your organization to put it to work? And then the third most important question our our chief privacy officer and legal team are used and leveraged heavily is, are there any existing regulatory frameworks that really prevent us from doing this and going forward with it? So those are your three on ramp or off ramps for any idea that's just come through. And I think they're really important because those get you through and cut through the ideas so quickly to try to get you to the ones that may have the most value for the organization.

[00:21:50] - [Speaker 0]
Wow. One of the things I love doing on this show is busting a few myths. And there will be some leaders that still view governance as something that slows innovation. So it was great that you busted that myth and to hear a story that governance is actually what enables organizations to move faster and operate more effectively over the long term, not slow.

[00:22:09] - [Speaker 1]
Made such a good point about Neil that with the regulatory frameworks changing so rapidly. So for example, and this is why the AI by choice comes into play again. What you could do in New York City and what you could do in Boulder, Colorado in the advertising of a job could be completely different. So you need to put the thought behind the processes as well as the platforms you're using that give you that level of agility. But on a global basis, to your earlier point, in operating anywhere.

[00:22:40] - [Speaker 1]
And because many of our customers are operating in multitudes of countries, we have to think forward that way. And I think it just makes the outcome that much better for the customer.

[00:22:50] - [Speaker 0]
And we will have a few people listening that could be nervous, especially around the fact that their company could, at some point, face scrutiny from employees, regulators, the public, their customers. So what is it that separates a truly defensible AI program for one that that could be vulnerable to criticism and end up on somebody's newsfeed somewhere? Because I think there could be people listening that get nervous over that happening.

[00:23:13] - [Speaker 1]
Well, we're not gonna say the g word again, governance. But the the truth of the matter is governance looks at not just can we, but should we. And so you have to think about things such as transparency, biasness, and other areas like that that are really key when you're building these models or using or applying these models. Can you describe transparently how this works to another individual? Do they understand how the data is used to formulate a prediction, especially with one that could be a discriminative prediction where it's actually predicting something versus generative where it's making up new content?

[00:23:57] - [Speaker 1]
So I think those kind of processes you really have to think a lot about as it relates to them biasness. Is the data that you're using in some way biased one way or another? The fact of the matter is, as humans, we a lot of us are biased and we create this data. So you really have to be clear that the data that you're using actually removes bias from the models that you're actually building. And then from there, it's monitoring because some of the models you may build will change in behavior over time.

[00:24:29] - [Speaker 1]
They learn. And so what you originally thought they were to make as a prediction at one point in time may change over months as it takes in more data and adapts the way in which it's looking at those patterns and thinking about what the representative prediction's going to be. It's called drift. So you drifted from where you started to where you are today. So having those systems in place is important.

[00:24:53] - [Speaker 1]
But you can buy platforms that do this and help you go through the transparency, the biasness, the deployment and drift analysis of your models and things of that nature. And you should expect that from any vendor that you're dealing with to be able to do the same.

[00:25:10] - [Speaker 0]
And I suspect every enterprise in every country in the world is currently being pitched or bombarded with pictures of AI solutions that solve all the problems, cut costs, and boost revenue, etcetera. But if a CEO or HR executive is listening to us today and they're thinking about AI tools, what are the one or two questions they should ask vendors before making a purchase to try and determine some of the good solutions from some of the not so good, shall we diplomatically say?

[00:25:40] - [Speaker 1]
Yes. And I've seen them as we assist our chief, digital officer, Carrie Rasmussen, in the analysis of software that comes into the business. So the one thing that I always look at, and it's I think it's become a very common one, which is do you train your models on our data? And it shouldn't just be a yes or a no. It should be a contractual thing that you engage with with that organization.

[00:26:06] - [Speaker 1]
And I believe many organizations will be surprised at the amount of pushback they will get when they actually have to put pen to paper and contractually say, we do not use any of your data to train our general models. Now it's okay if we're training a model just for our company and if we ever leave that company model is destroyed and never used. But training a general model on our data could be a very dangerous thing for organizations from that perspective. That's probably one of the biggest callouts that I see. Also, especially in international sense, one of the things that we constantly look for is something as simple as consent.

[00:26:46] - [Speaker 1]
Now in The US, for example, a lot of the software companies, especially the startups, they don't think about this in the EU context. But if you actually wanna have a conversation with somebody that you're recording, for example, you need consent. And I can tell you up to about six months ago, I would talk to companies that were like, well, what do you mean consent? Well, you actually have to engage the individual, get their actual consent that you're actually going to record their conversation and use that for some purpose after that point in time. And so I think that that's another powerful piece that comes into play as well.

[00:27:20] - [Speaker 1]
And I think we we have a number of different ones that we help customers with. I think there's a list of about 10 questions that you can ask every vendor about their particular process to assert whether or not they're doing the right things. And us, the certifications help. They basically say we've been audited, looked at by a third party, and we are doing the right things. And I think that's a big thing I would look for selfishly out of those organizations as well, which is proved to us that you actually have the processes in place to do this properly.

[00:27:55] - [Speaker 0]
And if we go back, what, only six years ago, working from home was just for the privileged few, not the many. All that almost changed overnight. Now we have hybrid working. We have AI everywhere, and I think that shows it's almost impossible to predict the future. But with that disclaimer in place, if I was to pull a virtual crystal ball out now, what would your vision be for the future of use of AI in HR over the next five to ten years?

[00:28:23] - [Speaker 1]
Wow. I I won't even try to guess ten years out. Thank you very much. And my five year one will probably be wildly off. But I think what you're going to see is massively increasing level of autonomy and trust in the AI systems for complex tasks.

[00:28:44] - [Speaker 1]
I believe that the the ability to have the human in the loop for those core areas that touch all of us, whether you talked about performance, you talked about compensation, you talked about hiring decisions. While I think there should be a human in the loop or on the loop as a minimum for those kind of decisions, I think one of the largest things you're going to see is really an automation of a lot of these core capabilities across HR, which is simply going to allow them to focus more on the change management that is constantly going to be affecting these organizations as they change over time and in a very compressed time frame. So massive autonomy, massive autonomous capabilities in that regard. And I think you're going the one other thing I really like is you're going to see the introduction of these digital assistants that come into conversations that are experts. So not large language models, but very specialized models that may be experts in collective bargaining agreements, in actually hiring contracts, in different aspects of HR, where the HR practitioner could bring in one of these specialists working on a collective bargaining agreement, collaborating with AI in that sense to create a better outcome for that content.

[00:30:11] - [Speaker 1]
And that gets back to, remember I talked about in the beginning how people start consuming other parts of other jobs into their job. That's what's going to enable that kind of capability.

[00:30:21] - [Speaker 0]
Wow. And I think that is a perfect moment to end on today. But before I let you go, we've covered so much in a short amount of time today, and I'm sure there's gonna be a lot of people listening wanting to carry this conversation on you start today. For all things Dayforce and follow you online and and some of the things that you're posting out there, where would you like me to point everyone listening?

[00:30:40] - [Speaker 1]
Oh, dayforce.com has always got great content on what we're doing on the AI side. And anyone who would like to, contact me on LinkedIn, I'm, fairly prolific out there, in terms of trying to get the message out and talk to people about it, just reach me on LinkedIn. I'm happy to continue the conversation there as well.

[00:31:00] - [Speaker 0]
Well, so many big takeaways from our conversation. In particular, busting the myth that governance is something that slows innovation. It actually enables any organization to move faster and operate more effectively over the long term. If there's one thing people remember today, I hope it is that. But more than anything, I will add links to everything that we mentioned today.

[00:31:21] - [Speaker 0]
I encourage people listening to go check that out. There'll be links in the show notes. But thank you for sharing this story today, and thank you for your time. Appreciate it.

[00:31:29] - [Speaker 1]
Neil, it's been a pleasure. I hope you have a great day, and I look forward to hearing your podcast shortly.

[00:31:36] - [Speaker 0]
I think today's conversation boosted one of the most persistent myths surrounding enterprise AI, and that is AI governance slows innovation. Because David showed today how clear guardrails, accountable leaders, trusted data, and strong review processes can all collectively help companies move from an AI idea to a decision in days, but while protecting employees and the entire business. And when technology is influencing careers, compensation, and hiring decisions, being able to explain and defend how all these systems work, I think that matters enormously. But I'd love to hear your thoughts. Is your business investing enough in AI governance before these systems begin making decisions that really impact and affect people's lives?

[00:32:31] - [Speaker 0]
Let me know. I'd love to hear more about the journey that you've been on, and you can reach me at tech talks network dot com. Socials, I'm just at neil c hues. If you want me on LinkedIn or Instagram, all the usual places, hit me up there, and I'd love to hear from you. And while you have a think about that, I'm gonna prepare for tomorrow's guest.

[00:32:50] - [Speaker 0]
So thank you for listening today. I'll meet you here same time, same place tomorrow, but thank you for listening. Bye for now.