ADP is using AI to enhance knowledge management and keep workers productive.

Artificial intelligence is quickly becoming a workplace fixture, and many thought leaders expect the technology to disrupt jobs and current business models as large language models such as GPT-4 advance in capability. According to a Pew Research Center study released in April, 62 percent of people surveyed think the use of AI in the workplace will have a major impact on employees during the next 20 years, and a greater share say it will hurt more than help workers.

But while AI implementation is cause for concern among some business leaders, Andrea Elkin, vice president of strategic enablement, enterprise learning, and knowledge management for ADP, sees opportunity. In particular, she sees potential in applying generative AI technology to ADP’s knowledge management efforts.

“ADP has a self-service knowledge acquisition culture, and knowledge management is an area of huge growth and investment for ADP’s \[talent development\] efforts.” says Elkin. “Our work in knowledge management is a shining spot in what we're doing today.”

Although the company’s TD function has been an early adopter of embracing knowledge management as part of a broad talent strategy, Elkin sometimes felt alone in her ardent pursuit of such solutions among peers in the field.

She shares an anecdote from several years ago at an industry event, when she asked a talent analyst about his thoughts on knowledge management as well as the future possibilities of AI and machine learning. His reply was less than positive, proclaiming that “It never works. You’re wasting your time,” Elkin recalls.

But recently, she attended another event with that same thought leader, and his perspective had changed significantly. She says he now thinks “Knowledge management is one of the most exciting places to be in our field.”

He’s not alone. And finally, neither is Elkin. With the release of ChatGPT in November 2022 and other recent advancements from Microsoft and Google, more businesses are interested in exploring applications of AI.

While other organizations are starting to invest in infrastructure and tech, ADP is firmly situated to leverage emerging AI technologies. “The sheer fact that ADP already invested in knowledge management as part of our talent strategy means we’re positioned to use generative AI to accelerate our existing initiatives,” she states.

Training plus knowledge management

In business for more than 70 years, ADP provides its global clients with cloud-based human capital management solutions that unite HR, payroll, and benefits administration, as well as business outsourcing services, talent analytics, and compliance expertise.

When Elkin joined the company seven years ago, a major priority was to centralize functional learning. Today, she and her team are responsible for strategic enablement and enterprise learning for 60,000 staff members worldwide, which includes internal support functions and client-facing employees. Additionally, her function trains the company’s more than 1 million clients on how to deploy and run ADP’s various products and services.

According to Elkin, an effective talent development strategy for a large workforce with continuously evolving skills demands more than traditional training solutions alone. Knowledge management, she asserts, must be a significant element of ADP’s approach to enabling its workforce for success.

A critical function of her team is determining which skills and information require training versus when support from a knowledge management system is a better fit. Elkin explains that “You need training when you have massive amounts of new information that require context—not just recall.”

For example, when someone is new to the company, structured learning helps create context about where they fit in the organization and what’s expected of them.

Meanwhile, knowledge management is more appropriate when employees simply need to be able to retrieve information easily and quickly rather than relying on rote memorization. Such recall information constitutes what Elkin calls the “Search and Find and Solve Platform.”

For instance, she details that when a worker has an issue with a client that needs immediate resolution, they likely aren’t going to remember everything they learned in training, nor are they going to have time to look though their email or a slew of printed job aids to find the guidance they need.

Instead, “everything they need must be accessible quickly in a knowledge management portal,” Elkin says, adding that “if you can't find answers or support when you need them, they’re not useful.”

Knowledge management plus AI

Employees’ queries to ADP’s knowledge management system trigger the delivery of articles describing how to perform specific tasks or providing solutions to common issues.

The job of the enterprise learning team is to build the knowledge management technology system and its infrastructure. Meanwhile, the team relies on subject matter experts within the business lines to write articles—based on guidelines and templates the team creates and provides to SMEs—that live in the system.

The team also ensures that content is accurate through a centralized validation process. “We review content, and we work with subject matter experts to verify content before it’s added to the system,” confirms Elkin.

It’s a deliberate decision to create article-type assets for the knowledge management system. Although people find videos, screenshots, and other types of visuals more engaging, they’re also harder to maintain and update. “For now, it’s more important to have useful and up-to-date content. It’s more important to be accurate,” she says.

Further, Elkin’s team analyzes how employees are accessing and using assets on the system. A dashboard enables any knowledge owner to see what workers are searching for and the keywords they're using.

“The functionality is akin to Google,” describes Elkin. “Every time an employee conducts a search, the system knows who you are, what role you're in, and what topics you seek information on. It also knows whether the employee received what they needed from a first-click position or second-click position, or if they received null results.”

She adds that in “about 93 percent of our searches, people click on the first two articles. That means we have huge relevancy in our knowledge management platform.”

The system also enables users to write feedback about the information they receive.

“We have a long way to go,” says Elkin. “But with generative AI, I think eventually we’ll be able to directly link specific content with the intent of a client call. We can get closer to analyzing ‘What is this call about? What’s the problem? Can the knowledge management system retrieve a smart answer for the employee rather than a long article?’ That’s where we’re headed.”

Elkin predicts that as the AI embedded in other company-wide systems progresses in functionality and accessibility, the learning enterprise will be able to use that information too. The TD function will be able to sort through employee interactions on customer relationship management software, chats on Microsoft Teams, email, and so on to see what types of tasks staff are working on and seeking information about most frequently. She anticipates how the enterprise learning team will be able to use those insights to develop better training and create valuable assets on the knowledge management system.

“That’s where generative AI’s going to really come into play. Eventually, it’s going to be able to write its own articles. For example, by listening in on client calls, it can determine which ones are resolved most efficiently and write a case study that we can add to the knowledge management platform.”

Elkin also believes that as generative AI improves, there will be greater capacity for the knowledge management system to dynamically produce shorter, fine-tuned, on-the-fly content when employees need immediate answers to questions and on-the-job support.

What’s more, as AI progresses, she expects to integrate more knowledge management solutions into the flow of work.

“Knowledge management becomes performance support … when it’s embedded in Teams or some other tool people use to do their work like a CRM, when it becomes a part of the work environment,” she says.

Skills multiplied by environment

As developments move technology capabilities forward, Elkin maintains that her team will continue to focus on not only teaching employees what they need to know to perform their roles, but also how to learn and find information as they need it.

“To continue to advance employees’ knowledge so they are able to problem solve, we have to teach them how to use these tools that deliver them information and resources,” she says.

To that end, training and onboarding programs that the enterprise learning team develops will present scenarios that ask people to use the knowledge management tools to solve problems so that “they’re building regular use of knowledge management into their work habits,” states Elkin.

It all boils down to nurturing an environment that enables employees to grow and use their skills, according to Elkin.

“We can do a lot of things to build your skills, and we can do a lot of things to build your environment,” she says. That is what she calls the productivity formula: “Skills times environment equals productivity.”

Elkin explains it this way: “If you have zero skills, I can put a lot of things around you—like knowledge management systems, support technology, and coaching—but you will still have low productivity. Or I can have somebody with great skills, but if the environment of performance support around them is lacking, their overall productivity will still be really low.”

She is quick to remind that “It’s a multiplier, not a plus. If one of those factors—skills or environment—is zero, your productivity is zero.” As a talent development leader, she says “I work on building skills, but I also work on building the environment. Our workforce needs both.”

Read more from Talent Development Leader.