The Role Of Generative AI In HR Is Now Becoming Clear
The Role Of Generative AI In HR Is Now Becoming Clear
HR Is An Integrated Operating Function
Let’s remind ourselves that HR, like Finance, IT, and other internal functions, is a design, support, and integration function. HR partners with the business and deals with a myriad of complex issues: hiring, onboarding, training, leadership development, performance management, pay, rewards, benefits, hybrid work, organization design, diversity strategy, culture, and more. And prior to the emergence of what we call Systemic HR, most of these operating functions were done somewhat independently.
Today companies are dealing with a competitive labor market, high levels of turnover and workforce stress, and the need to reskill, upskill, and intelligently move people internally. Problems like diversity and inclusion, culture, and leadership development remain paramount, and HR teams are also worried about employee experience, productivity, and internal efficiency.
Data within HR is spread all over the place. The average large company has more than 80 employee facing systems, and each one stores large volumes of important data to help manage its own area. When a business leader or executive wants to make a change, look at a business scenario, or fix an underperforming group, they need all this data in one integrated place. AI promises to bring this dream to life (more below).
As HR teams build new programs and solutions, we are also dealing with an overwhelmed workforce. Employees are largely tapped out (87% believe they are operating at full capacity) so we have to simplify work, reduce the number of systems, and save people time on administrative functions (enabling them to operate at the “top of their license”). This means HR teams are continuously dealing with this problem of expanding the number of services yet shrinking their footprint and making them easier to use. AI helps with this.
Finally, HR teams are turning into creators, developers, and consultants. As our Systemic HR research points out, the future of HR is fewer “support agents” and more “consultants, product managers, designers, and advisors.” This means more and more HR teams are “building things” and “analyzing things,” which is essentially a core part of what Generative AI does.
So in a sense, Generative AI is the perfect new solution for almost every challenge HR teams face.
How Will We Get There: Real Use Cases
As we’ve talked with dozens of companies and vendors, let me summarize some of the big, high-ROI, real-world use-cases we see.
1/ Talent Intelligence for Recruiting, Mobility, Development, Pay Equity
Talent Intelligence is now a reality. Companies can use LLM-based systems (Eightfold, Gloat, Beamery, Seekout, Phenom, Skyhive) to identify hundreds of characteristics (ie. skills) in their people, enabling companies to intelligently source candidates, decide who is ready for promotion, move people to new opportunities (talent marketplace), and identify pay inequities.
This domain, which we’ve studied for several years, is now available “off the shelf” from many vendors and using data from providers like Lightcast a company can relatively easily start to identify capability gaps, look into the outside market for trends, and build a strategic and operational solution for many HR practices using AI. Our Talent Intelligence Primer explains this in detail – I believe this market is still young and will eventually disrupt many of the core HCM players. (SAP introduced their Talent Intelligence Hub this year, by the way.)
In recruiting there are now plugins to generate job descriptions, tune them for different roles, create personalized candidate emails, and enrich your own resume. These tools are getting smarter by the minute: they can now personalize every part of the recruiting process, saving recruiters time in outreach and writing. I just saw Eightfold’s latest AI job description builder, for example, and it lets you tweak the description based on skills, technologies, and many other factors (Seekout has a similar offering).
2/ Employee Experience Apps (Onboarding, job transition, administration)
The second growing space is the “intelligent employee chatbot” that brings together documents, support materials, and transactional systems into an easy to use experience. Several of our clients are experimenting with this and our own JBC HR Copilot offers this type of solution to HR professionals themselves. These are really enterprise applications, where companies put together their own content, develop a data security strategy (we don’t want every employee seeing every document or process), and then use “orchestration” tools to connect the chatbot to enterprise systems.
There are many ways to do this. OpenAI has a feature called “function calling” that lets a develop take any input (“I want to log my vacation.”) and turn it into a simple call to an API like the vacation page in Workday or SAP. IBM Watson Orchestrate is designed for this (now in use by SAP), and there will be many such tools available from platform vendors and the HCM providers. The Workday Assistant is a first generation attempt at this.
Once you marry the knowledge of the various HR systems with process documentation, the chatbot can be an eventual replacement for many or most employee portals. I just talked with a large media company and the HR leader told me it took her three hours to find the right information on executive comp to do her analysis. This type of situation is common everywhere.
So far what we’re discovering is that these should be focused on narrow use cases first, then they can expand. A large hotel chain, for example, just build a chatbot designed to help front office workers understand precisely how to serve high net-worth customers. It connects to the reservation system and helps the employees know how to customize services for that client. Imagine an onboarding tool, leadership transition system, etc. like this.
Every EX vendor is going to want to be part of this. Providers like Firstup use AI to customize employee communications to each person individually. This will become a core set of features we use for many of our employee experience apps.
3/ Employee Training and Compliance Apps
The $350 billion employee training industry is hungry for Generative AI. We’ve seen tools that generate training from documents, automatically create quizzes, and take existing content and turn it into a “teaching assistant.” Just yesterday I talked with a client who has a new leadership development program they just built with a vendor. We discussed taking that content and putting it into our Copilot to make it available “on-demand” with a conversational interface for managers. That is not a difficult project once you have the AI platform in place.
But there’s more. Cornerstone, Docebo, Degreed and others are now using AI to intelligently recommend content (based on Talent Intelligence, not just clickstreams), produce and recommend micro-learning based on role, team, location, and employee activity, and even use AI as a game to “prompt” the employee to learn more.
To give you an example: we just launched a micro-learning program in our Academy to teach HR people about AI. That course, which consists of a series of interactive questions and small notes and interactions on your phone, could be imported into our copilot, for example, and offered when someone asks a question. These are not quite out of the box solutions but we’re close.
Remember that much of an L&D teams job revolves around content creation. These new Gen AI apps that build characters, images, scenarios, and videos are going to be widely used by L&D teams. I just found a tool that takes long videos (ie. instructor-led courses) and quickly finds the “most interesting” or “most dense” content to create mini-snippets. Imagine the opportunity you’ll have to take long videos and turn them into chapters, on-demand learning, and promotions for new things to learn.
4/ Employee Development and Growth Apps
Next there’s the massive new area of tools and platforms to help employees with their careers. Thanks to Talent Intelligence platforms, we now have “career pathways” being generated by AI (not your boss). These systems look at your skills and your experience and show you (graphically) all the options you have for growth, all based on the experience of millions of others.
Did you know, for example, that a marketing manager who does analytics could move into data science, cyber security, or even financial analysis? Or that a person who works as an hourly “transportation support” person in a hospital can join a career pathway to become an X-ray tech or clinical nurse?
These pathways are all exposed and explained by AI, and these new systems show you precisely what you need to learn, what certifications you must acquire, and even who you can talk to about this path. We are actually working on this type of solution for HR professionals (coming soon) and you’ll be amazed at how helpful these tools can be.
Why is AI so important? Because this is fundamentally a big-data problem. I cannot possibly guess all the career options an individual may have in our company, but if I plug their profile and history into a system like the Eightfold Career Navigator or others, we can both see many options we never even considered. Products like Gloat, Fuel50, Eightfold, Beamery, and Workday all offer this out of the box.
And think about how this will help non-degreed workers move up in their careers. No more shopping around on websites to guess where to apply for a job – these career navigation systems are going to transform the lives of many many people.
5/ Performance Management and Operational Improvement
Should AI be used for performance management? Well I don’t expect these systems to write performance reviews, but yes, they will help a lot. Consider the typical problem we have in every company: a team, a workgroup, or an individual is simply underperforming. This group or person’s numbers are behind, their projects are late, or their quality is not up to snuff. Do we wait for the manager to decipher what’s wrong and let them figure out what to do?
That’s how it works today: each manager has to guess, figure out, and decide “what to do” about a low performing individual, team, or project. Why not let the AI do some of this for us? We have seen apps, for example, that show you the integrated “view” of performance in a company. This is, in many ways, a data problem.
What if we find, for example, that the project teams that are over a certain size simply don’t get things done? What if we look at the skills composition of a team and see that an important one is missing? Maybe tenure is the problem (it often is, by the way). Maybe diversity is holding teams back.
While the line manager may not do this kind of analysis, I can guarantee you that the HR consultant would love to help here. These kinds of broader organizational design and performance projects are everywhere, and once we have all the data in an AI system we can simply ask it questions.
I asked Bard, “please compare the financial growth, returns, and margins of Chevron and Exxon.” It did a pretty good job in about ten seconds. Imagine if you did that in your own company across teams? Once we get our internal data into the right AI system this is going to be a regular and common thing to do.
6/ Retention, Hybrid-Work, Wellbeing, Engagement Analysis
And that leads me to my final big area: studying, analyzing, and improving employee retention, wellbeing, and engagement.
Every company I talk with is now dealing with employee burnout, wellbeing, and other engagement issues. For decades we relied on surveys and various benchmarks to try to figure out what to do. And yes, good feedback systems give us lots of information that helps.
But what if we simply put this data into our big AI platform and asked it some questions. “What are the top factors contributing to turnover in the sales department?” It may be manager. It may be compensation. It may be tenure. It may be something else.
Yes we can always use surveys, town halls, and other listening methods to do this. But what if we just look at the data? The Bank of America Academy, which we’ve written about many times, is a story of a company that “discovered” what its talent problems were by detailed analysis of the data. They found out, for example, that bank balances are very correlated to the tenure of employees in the branch. And tenure is driven by many other factors: how people where hired, onboarded, and supported in their career journey. By doing that analysis they were able to dramatically improve their business performance and retention. Their engagement surveys would never have pointed this out.
How Do You Get Started?
And that leaves us with the big question: How do you get started? Let me share what we’ve learned.
First, rather than “chase the technology” it’s much better to “fall in love with the problem.”
In other words, what problem would you like to focus on? Is it employee onboarding? HR self-service? Hourly worker scheduling and shift management? This means getting your team together to prioritize your investment, because building an AI-based solution won’t be as simple as you think.
Second, once you’ve decided where to start it’s time to get the IT team involved. Each one of these use-cases turns into a set of issues with data quality, data management, data dictionaries, and then security, business rules, and confidentiality.
Remember that “throwing information into an LLM” may sound like fun, but even if it works you’ve just given all sorts of people access to information they may not need, want, or even be allowed to see. So a chatbot implementation means focusing on user experience, data management, search, and orchestration all at once.
Our work on our own copilot has already given us this experience. Once you get the data together (and in most cases it’s not clear who owns what), you have to start testing Gen AI use cases, define security rules, and decide what, if any, back-end orchestration you want. These aren’t as exciting as “throwing a bunch of spreadsheets into OpenAI and starting to ask questions,” but this is what real solutions need to do.
Third, you have to realize that AI systems, unlike transactional systems, take care and feeding. “Prompt Engineering” means tuning the system to answer questions correctly, finding gaps in your data or documentation, and continuously trying to keep the user experience simple. And once the chatbot or other system is operational, I can guarantee there will be demands for more (and new) data.
In many ways a new AI system is like a new baby. It has to learn how to walk, talk, behave, and stay out of trouble. The off-the-shelf tools won’t do this until you’ve really used it, so you’ll need IT’s help making sure your system is sustainable and supportable as it grows.
How Will AI Impact HR Itself?
And then there’s the big question about your role. Will these new systems make you obsolete?
The answer is clearly no. These intelligent systems are data hungry fiends. Once you build them and add the right information, you’re going to turn into an analyst, a chatbot trainer, a product manager, and a designer. A lot of the mundane work of finding information and analyzing it may go away, but your higher level job of knowing what information to use will remain. And there will be many new jobs taking care of the AI systems, tuning them, and continuously improving them as new applications arrive.
Let me leave you with this: despite the explosion of excitement in this space, AI implementations in HR are technology projects. They have many of the same issues and challenges of any transactional system, with the added element that the system itself is “learning” along the way. SAP doesn’t change behavior as you use it: the AI system will.
I can assure you that this entire domain is both over-hyped and under-estimated. If you start small, get your hands dirty, and bring your IT team with you, you’ll start to see astounding business benefits in any of the areas I discuss. And in the next article I’ll explain how to build an AI-Hackathon for your team.