How AI Will Disrupt The HR Tech Market. And Also The HR Function Itself.

April 08, 2023 00:26:22
How AI Will Disrupt The HR Tech Market. And Also The HR Function Itself.
The Josh Bersin Company
How AI Will Disrupt The HR Tech Market. And Also The HR Function Itself.

Apr 08 2023 | 00:26:22

/

Show Notes

This week I summarize the results of 20+ interviews with HR Tech and AI vendors and talk about how AI and all its various manifestations can teach us a lot about how to approach Human Resources in a much more systemic way. Not only are the HR tools and Copilots starting to roll out in the market, but we are also seeing how AI-centric platforms (Talent Intelligence systems, for example) will really disrupt and ultimately obsolete many traditional HR tools. I also introduce the idea of "Falling In Love With The Problem," which in many ways defines what we call the new Systemic operating model for HR.  (Uri Levine, the founder of Waze, really first coined this phrase.) Resources I Recommend What Is A Neural Network? Fantastic Overview Of How AI Systems Work. Redesigning HR: An Operating System, Not An Operating Model. Why Is The World Afraid Of AI? The Fears Are Unfounded, And Here’s Why. Irresistible: The Seven Secrets Of The World's Most Enduring, Employee-Focused Organizations SeekOut Brings GPT4 To Recruiters. Eightfold Launches Copilots For HR. Workday’s Response To AI and Machine Learning: Moving Faster Than Ever Yann LeCun and Andrew Ng: Why the 6-month AI Pause is a Bad Idea
View Full Transcript

Episode Transcript

Speaker 1 00:00:09 Hey everyone. Today I want to give you some really thoughtful ideas about some things that have occurred to me this week. And as most of you know, I spend much of my time during the week talking to HR executives in HR departments, and also a tremendous amount of time with vendors, technology providers, consulting firms and solution providers. And today I want to talk about the interconnection and the lessons we can learn between the innovations in AI and everything we do in hr. And I know that sounds a little bit strange, but there's a really interesting story here. So early in the week, I ran across an incredible YouTube video on neural networking. And I really recommend you watch it. It will teach you a lot about AI and it will demystify it. If you're not a math major, you might find it a tiny bit intimidating, but it's actually, it's very, very easy to understand. Speaker 1 00:01:02 And the graphics are incredible. And basically what AI is all about is using mathematical models, prediction algorithms, cost minimization, algorithms, what are called back propagation algorithms. You take a whole bunch of data from a whole bunch of sources and identify patterns, classify objects, and make recommendations and identify the statistical relationships between these objects. And so when you use chat G P T, you're really using something that's not that intelligent, but something that's very, very well trained that understands the relationship between hins, which is groups of letters and other tokens in a very sophisticated way because it's been trained on lots and lots of language. And so when you watch the video and you take a look at how neural networks work, you'll begin to understand that the mathematics behind this are actually fairly well understood that the innovations and the research that's taking place is advancing the speed of the math, the quality of the math, and the algorithms that are used to train the models. Speaker 1 00:02:11 Because a trained model is essentially a data system or an algorithmic system that has captured trials and errors of billions and billions of observations to figure out through this, you know, multidimensional model, what is most likely to be what, and what are the relationships that are most likely to happen in the future. And that's what language models do, is they do that with words. And when you go to image systems that does it with images that recognize your face, AI used in medical research can identify cancers and on and on and on. So I'm not gonna spend a lot of time on that. We are going to build a new AI course. We have an AI course in our academy, by the way, that's pretty good, but we're gonna update it. And so, so if you join the J B A, you'll get it and it'll be out, you know, probably in the next month or so. Speaker 1 00:03:03 But we're gonna do a lot more on that. But in the meantime, let's think about hr. So all of the discussions we have with you about various HR issues tend to revolve around complicated problems. Friday, we talked with one of the HR leaders at a very large insurance company and she reflected that this is a company that just went through a massive merger with another company. They're moving from the insurance business into the wellbeing and health business. You can imagine that scenario. Most of the insurance companies are doing this and they're having all sorts of questions about operating models, leadership, hiring, recruiting, training skills, et cetera. And she said, I'd like you guys to come in and take a look. Well talk about a multi-dimensional problem. As you know, merger and acquisition and a business strategy change affects operating models, skills, roles, reward systems, leadership models, customer focus. Speaker 1 00:03:55 You know, really interesting problems. I consider this to be maybe the most interesting work we do. We also had the big reset group, which is about 400 companies that work with us every Friday on a whole bunch of topics. And I joined the group that was working on engagement and retention, and we had long conversation about the role of AI and the different models that companies use. And at the end of it, Noah Reitz, who's been leading that group, said, you know, what we've been basically talking about for the last five weeks is that even if we look at every individual element, retention, whether it be pay, pay, equity, work-life balance, flexibility, workload, job fit skills, education, career, et cetera, he said it always came down to culture, it always came down to leadership and management. And I'm sitting listening to this and I'm thinking, you know, this is a multi-dimensional problem. Speaker 1 00:04:42 You've all read books and articles and studies that have said the number one solution to employee engagement is having a best friend at work, or it's belonging or it's trust, or whatever it is. And the reason we keep coming to these simplistic solutions is that we're not using holistic analysis when we solve these problems. And I know there's a lot of statisticians and smart mathematicians in hr, but what I've observed is most of the time we do AB tests, we look at the impact of a few things and we come to conclusions that, aha, if we do this, that outcome will change. And yes, that works, but we don't know what we didn't test. So that is what systemic HR is all about. Systemic HR is about thinking about your job as an HR professional or leader as a problem solver, and all of the domains of hr, whether they be recruiting skills development, pay culture, leadership development, management tips, recruiting, et cetera, all those things, all the, all the 92 practices that we develop in our academy are interrelated and connected to every problem. Speaker 1 00:05:53 So if the problem is retention or performance or productivity or harassment or poor customer service or poor quality or misalignment, we have to bring all of those domains together in a holistic way. In other words, what we talk about a lot in the big reset is what is called falling in love with the problem, not falling in love with the solution. That seems to be a buzzword phrase that has come out of product management by the way, and marketing that we need to second and third order understand the problem before we come up with a solution. Let me give you a very, very simple example that I think everybody can relate to. Most companies have a problem of onboarding when they hire a lot of people look at what's been going on in the tech industry with all these people that were hired and didn't have enough to do. Speaker 1 00:06:41 And so you could, you know, sort of say to yourself, oh great, let's put together a team and build a new onboarding solution. Let's look at all the mechanical stuff they need to do when they join the company, the people they meet to meet, we'll give 'em a little bit of training, we'll introduce them to a peer, we'll give them some projects to work on their first couple of months and boom, we'll have the onboarding problem solved. And I've seen this a lot. I mean, I've seen this many, many times. So you can buy off the shelf onboarding tools, which presumably you do this for you. Well guess what? That isn't the problem. The problem of onboarding and most companies is not getting people up to speed the first couple of weeks. It's teaching them what they need to know to get to a steep ramp of productivity in their job. Speaker 1 00:07:27 And it's not the first day or the first week, it's the first year or two that really needs to be addressed. And so if you really look at the problem of onboarding as a problem first before you look at a solution, you would, you know, interview a lot of people that have joined this particular, by the way, it's also different by job role. As you know, onboarding a software engineer is not the same as onboarding a salesperson or a customer service person or you know, somebody who's a janitor or an operations person. You would, you would study the first year or two of these people that are highly successful. You would look at the people who were unsuccessful and you would then back up and say, what are all the things that drove success for the people that really thrived and the people who didn't thrive? Speaker 1 00:08:12 And you might find that it isn't the onboarding, it's really the selection and the background of the people or the personality type or the education or so forth did that. You might say to yourself, wow, this onboarding program is important, but we really need to go back and talk to the talent acquisition group first and see how they're hiring people. Because I can't build a generic onboarding program unless we know exactly the, you know, type of people and background of people we're gonna hire. And then you're gonna get into pay, and then you're gonna get into diversity, and then you're gonna get into management development. And you see my point that something as simple as onboarding is a systemic HR problem. And you can replicate that idea into every single thing that goes on in hr. And we are getting very close to publishing some really good information on how to do this. Speaker 1 00:08:58 We had a fascinating conversation with Tom. Tom, we've talked to Nestle, we've talked to some amazing companies about this. Come to our conference, you'll get some previews on all this, but you understand the idea. So all of this is going on in our HR domain and in a parallel universe, there's a bunch of mathematicians working on ai. And as I've gotten to know ai, and I've talked to more than 20 vendors now in the last couple of weeks about it, I realized it's a very similar analogy. What happens in an AI system as you, the difference between AI system and a traditional software system is the AI system analyzes the data and from the analysis of the data understands and develops applications and solutions driven by the data. You don't program. You don't tell the AI how to identify a cat or a dog. You teach the AI how to identify a cat or a dog. Speaker 1 00:09:51 In other words, you don't tell the AI how to select the right candidate. The AI learns how to select the right candidate. And so in many ways what's going on in AI is very, very similar to the systemic problem we have in hr. We can collect and start to manipulate and understand a huge amount of complex, sometimes ill-defined data, but, and the system will identify patterns and give us good recommendations, good characterizations and good solutions. And this is a lot of what we try to do in hr. I mean, if I think about all of my observations of HR for the last 25 years, the reason I find it so fascinating is we're dealing with a very, very complex problem. Human beings are complicated animals, take something as simple as the word skill, okay? Everybody's gone nuts about building the skills based organization. I am kind of tired of hearing that because it doesn't mean anything. Speaker 1 00:10:49 When I got outta college in 1978, every company I interviewed wanted to know my skills. This is not a new idea. And we've oversimplified the word word skill is a very complicated idea. For example, if you're a salesperson and I teach you how to enter an opportunity into salesforce.com, is that a skill? Yeah, it is. Is that going to make you a better salesperson? No, it's not. It might help the company track what you're doing, but it may or may not have any impact on your ability to identify a problem, propose a solution, handle the objections, and close the deal. It might have nothing to do with that whatsoever. So the word skill, as I've talked about many times, is a complex idea. We as humans consider around and we can categorize the difference between a skill and a capability. We can look at granular skills versus complex skills. Speaker 1 00:11:42 We can look at hard skills versus soft skills. Other guys have done a lot of this. I do a lot of this with our clients too, but reality, it's kind of amorphous. So if we used ai, maybe the AI could do this for us. So let me talk to you about a conversation I had this week with the founder of one of the most successful AI companies in the industry. I'm not gonna mention their name because you may guess who they are. And he is a PhD researcher. And one of the things I am discovering is that the AI systems in HR that are the most powerful and the most important are oftentimes developed by PhD researchers, not by product managers. And so there's an interesting change going on where product management is not necessarily the right discipline. To build a great AI system, you really need a researcher. Speaker 1 00:12:35 And so what he has done in their company is they have over many years, and they started maybe a decade ago, built models that look at the patterns of high performers. First of all, you have to identify and decide what a high performer is. People that whose careers progress at an above average rate, who move up levels more quickly than others who have broader spans of control and so forth. There's, there's obviously data elements that would define what a high performer is. And their system has done that. They built a model to go through billions of employee records profiles and figure out who a high performer is. And then they can look back at that model and they can say, let's look at the characteristics or skills or experiences that those high performers used or consumed or developed relative to others. And guess what? Their model actually generates a skills graph. Speaker 1 00:13:31 And it isn't things like entering opportunities into Salesforce, it's more complex than that. And because they are falling in love with the problem, not the solution, they are getting orders of magnitude better answers from their system than the typical E R P or somebody threw an AI engine in there and just said, Hey, let's just figure out how to recommend courses because he's interested in Java. And these courses all use the word Java in the title, right? I mean, that's kind of like generation zero of this stuff. And that is a systemic approach. Another example that I think is fascinating is another vendor who is also led by a PhD, by the way, who did a lot of academic work on ai. They are doing the same thing for leadership. They have selected a particular company. This is a test, they're doing it as a, as a development of a product. Speaker 1 00:14:18 And they've said, let's look at all of the people above a certain level in this company. Let's look at their job profiles, job histories, credentials, information we can capture from them. And let's try to figure out what are the leadership skills and capabilities and experiences they've had relative to other people in that particular industry. And for this particular case, it was the insurance industry. And I looked at the model and I thought, my god, this is ground breakly important thinking here that has come out of ai. And my point is two here, one, this idea of falling in love with the problem is the most valuable thing we could do in hr. Because if we understand the problem well, we can bring the resources to bear to solve the problem. And by the way, creating a new HR operating model to save money is actually kind of easy. Speaker 1 00:15:06 But designing the HR function so that you can go after problems in a multidisciplinary way that's a little bit harder. And what you find, by the way, in the domain of HR when you do that, is the roadblock you run into like a ton of bricks is the skills of the HR team. If you don't understand the domain or the complexity of the domain, you will provide a simplistic solution. Just like if the AI model is trained in a limited domain, it will not necessarily be as smart as you want it to be. So it's very, very analogous. These ideas. And let me go back to HR one more minute here before I talk about ai. Again, you know, we are doing, I'm going to New York this week, I'm gonna be visiting with some clients, we're gonna be launching our academy for a couple big companies. Speaker 1 00:15:53 And what occurred to me as I was looking through the data of our capability model is we are living in, and you are all working in a profession that is every bit as complex as it or software engineering or genetic engineering or any other domain you may pick. If you look at the 92 capabilities that we've identified in HR that matter, and by the way, that seems like too many, but it's not. When you look through it and you look at things like change management or candidate selection or employment brand, candidate marketing or employee experience, those are complicated topics. So your future as an HR professional has to be grounded in this idea that you're gonna learn a lot of things and you're gonna have a T-shaped career. You're gonna be very, very deep in some things and you're gonna be broad in other things. Speaker 1 00:16:43 And, and really this is the reason I bring this up, is this is our mission. This is what our academy is all about. This is what our research is all about, is letting you build out that T in your career so you can add more and more value, you can get promoted, you can do bigger things in companies, and we can just make the economy and the workforce better for everybody. So let me go back to ai. So what I believe is happening in the AI tech area is two or three things I would say of the vendors I've talked to. They've fall into three categories. There are vendors who are oftentimes led by PhDs who are building talent intelligence systems that are very, very intelligent. And these systems are open data systems. They capture data from many, many sources because as I said, if the model is limited, the answer is limited. Speaker 1 00:17:33 So if you're a hospital and you wanna put all your nursing credentials into the AI system that you should put them in there. If you're a bank and you want to put the bank balances of all of the branches in there to see which branches are performing the best, you should put that in there. They're building really advanced predictive and utility based systems that I think are gonna transform the way we work, the way our companies operate, the way we do talent management and everything. And those are what I call the deep level two talent intelligence vendors. I call 'em the second generation talent intelligence vendors. Then there are a set of companies, vendors who, who understand the potential for AI and they're trying to product manage it into their products. And they're usually running into problems along the lines of, geez, we just spent a lot of time building this system without thinking about the AI implications. Speaker 1 00:18:24 I wonder if we have to rebuild this. And they're afraid of doing that and they're frightened, by the way, the first category of vendors are willing to replace algorithms with new ones. Very quickly, one of the, the heads of one of these companies said to me, I gotta be willing to throw away a lot of my code pretty quickly here as the AI advances. Second category of vendors are not as comfortable doing that. They tend to be slightly bigger companies, they tend to be slightly older companies, and they tend to look at AI as a feature for selling purposes. They try to put demos of how cool it's gonna be, but they haven't necessarily looked at the deep data implications. They're not thinking about the learning that's gonna take place from the AI yet. And they haven't really investigated the use cases enough. And I've been sort of surprised how many vendors fall into that second category. Speaker 1 00:19:12 I'm gonna do my best and we're gonna do our best as an analyst firm to coach those vendors how to be more forward looking in their products because I do believe the AI implementations in HR are going to revolutionize what we do and perhaps obsolete some things that we take pretty seriously today. The third category of vendors are software companies that just don't seem to get it yet. And these are companies that see AI as a marketing opportunity and they are talking about it fairly incessantly, but not making a whole amount of sense. And, and you know, the problem for those of you that are HR people is you may not be able to tell the difference between these different kinds of solutions because they're gonna show you something and it's gonna look really cool and you're thinking, oh, that's great, I want to use that. Speaker 1 00:19:57 And to some degree, the vendors are going down their own learning curve of discovering that the more they use the AI models, the more expansive and different their products may become and the way they have to change them. And so that's why I really want those of you that are HR people to just spend a little time getting to know how this stuff works. I'm gonna put some links into the podcast for you to read through and talk about a couple of more things relative to education. I mentioned we're gonna be putting a lot of things in the J B A about this at the irresistible conference in June. We're going to have some senior executives from Microsoft and I think we can get somebody from open AI to come to talk to you about the implementation of AI in HR within Microsoft and what they're doing and what this means to you from a Microsoft perspective. Speaker 1 00:20:46 And they, they are definitely down the learning curve on what generative AI is and how to use it. And I am doing lots of work with our team on understanding the implications relative to different vendors. If you are a software company or a vendor, you know, we'd love to talk to you, we'd love to see what you're working on and give you whatever advice you'd like. And for those of you that are out there buying systems, it's not a bad time to slow down a little bit and talk to your own team about the implications of these new technologies. I believe, and I have a pretty good sixth sense about this, that the AI implementations of enterprise software in the HR domain are going to, to some degree obsolete. A lot of the old systems we have, we're not gonna throw the old ones away because they're doing a lot of hard work that we still have to do, but they're very different in their applications and their potential. Speaker 1 00:21:43 And the core sort of transactional e r p management systems that we've bought in the FA past and some of the examples, the the most profound examples are now happening in re in recruiting. But we have co-pilots coming out from companies like Eightfold and DSE out and beamery and others where they're now recommending interview questions for recruiters, they're recommending candidates, they're giving selection advice based on data in the system. Think about the implications of that and how powerful it could be when you're doing an interview and an interview intelligence system recommends a certain question that has proven to be very predictive of the success of this role. I may have mentioned this in another podcast, but when Laslow Bach was running HR at Google, he told me they did a study of the predictability of the off the shelf interview questions that they were using in the success of candidates who were joining Google. Speaker 1 00:22:39 And they're very scientific about it. And he said, you know, none of our interviews were very predictive of how well people were gonna perform at Google. Our AI did a much better job. And so just in recruiting alone, we're starting to see these tools come out in ways that for sourcing, selection, interviewing, candidate analysis, we can do things a lot better. Imagine how that's going to work as a coaching tool for leaders, as a coaching tool for new managers, as a coaching tool for teammates, as a productivity tool to figure out why meetings are not performing well, et cetera, et cetera, et cetera. And this is only going to get better because as one of the vendors I talked to this week, a really interesting learning vendor by the way, that's going to disrupt the learning industry. I believe he had met with Sam Altman and looked at G P T five and he told me what's going on at G P D five and I've, I, I assume he is correct, is they're taking the learning that has happened in generative AI for text and language and they're combining it with the learning that can take place in image recognition, video and audio. Speaker 1 00:23:46 So now imagine you're having a teams meeting, everybody's talking about some problem, there's a big debate, maybe there's an argument, a meeting complete concludes, there's some, you know, conclusions that are drawn, some questions that are left open. The AI analyzes that meeting and tells you, I don't know what it's gonna tell you. I mean we'll know in another year. That's where this is going. And you compare that to a sort of a dumb meeting software, you compare an intelligent recruiting system to an applicant tracking system. It's like comparing a rock to living breathing animal. They're just not the same. There's no point in doing a comparison and you can't take the rock and add a few features to it and turn it into this intelligence system. It's not that simple. So, so I think this is a fundamental change in the tech architecture and the way these systems are gonna work and the way you think about data and reflecting that back to hr, again, that is exactly what systemic HR is about. Speaker 1 00:24:45 I do believe that the companies we're talking to now that are doing systemic HR that are very consultative and cross-disciplinary in their strategies are just doing amazingly different things for their workers, for their companies, for their executives, for their customers than the traditional approach. So I just think this is such an interesting relationship between these two things going on in our world right now that are so similar, the multi-dimensional complexity of our jobs as HR professionals, as recruiters, as developers, as business partners, and the multi-dimensional, highly refined AI that's taking place in data and tech. So take that as something to think about over the weekend and let's talk next week. I'll tell you all about some cool stuff we're doing in New York at the end of next week and we look forward to hearing from you with your feedback and your comments. Thank you.

Other Episodes

Episode 0

October 22, 2023 00:20:01
Episode Cover

Trailblazer HR Technology Vendors: STRIVR Labs, Seekout, Sana Labs, HiBob, Rippling, ServiceNow

In this podcast I briefly discuss the Trailblazer HR Technology vendors STRIVR, Seekout, Sana Labs, HiBob, Rippling, and ServiceNow. These are vendors who are...

Listen

Episode 0

September 24, 2023 00:17:44
Episode Cover

A Brief History of Human Resources And Why Systemic HR Is What's Next.

In this podcast I go through the last 100+ years in the evolution of Human Resources. As you'll see, this profession has exploded with...

Listen

Episode

March 02, 2024 00:21:16
Episode Cover

M&A In HR Tech and Implications Of The Workday-HiredScore Deal

This week Workday announced plans to acquire HiredScore and while it may seem tactical, this deal could have bigger implications. In this podcast I...

Listen