The Next Frontier?
Natural language processing is beginning to demonstrate its potential as a transformative technology in a variety of HR areas.
By Terena Bell
No one wants to replace your HR staff with robots. At least that's what Rob May says. He's CEO of Talla, a Boston-area company that developed a "chatbot" that interacts with employees via chat programs such as Slack and Microsoft Teams.
May's company is one of many artificial-intelligence start-ups that seeks to shift some tasks from workers to technology. Talla promises to take over routine HR chores such as helping employees change their tax withholding or retrieve vacation balances. Talla, May says, doesn't want to eliminate your people -- just their busywork.
Whether your team consists of 10 people or 10,000, busywork is very often a big part of the HR job. Inevitably, those in the profession get bogged down in people asking about vacation accruals and 401(k) plans, and "all these things that are just . . . repetitive . . . ," says May.
As a result, HR isn't able to spend its time on the issues that really matter to the business.
Talla isn't alone in the effort to eradicate HR busywork. Several new HR tech products now use "natural language processing" to understand -- and communicate -- as people do. Commonly referred to as NLP, the technology uses machine learning to, among other things, recognize different phrases that express the same meaning.
For example, one employee may ask: "How do I get my partner on the health plan?" Another says, "My husband needs insurance." NLP enables Talla's platform to know these sentences mean much the same thing.
Making It Easy to Use
Of course, there's nothing new about using technology to streamline routine work that's performed by HR, points out Josh Bersin, principal at Bersin by Deloitte. But NLP can streamline it even more.
"Some of these chatbots are doing relatively traditional stuff," Bersin says. "But they're doing it in a very easy-to-use way."
Among Talla's customers is rLoop, a Silicon Valley nonprofit based in Menlo Park, Calif., that's striving to build a futuristic "hyperloop" transportation system.
Talla "just took away all those mundane repetitive tasks," says Edward Chen, head of HR at rLoop. At any given time, rLoop has 200 to 1,000 volunteers in more than 20 countries. That's a lot for one person to onboard. So Talla welcomes new team members, suggests training, then says, "[I]f you have any questions, ask the bot." Over time, the bot will be trained to answer the same kind of questions.
"If someone asks a [new] question [and] the bot doesn't know how to answer that," Chen says, "I will respond. But the next time the bot will be trained to answer that question, so it really helps me to focus on something more strategic."
Another HR tech tool that uses NLP has been developed by Scout, a New York startup that promises to find job candidates who may be overlooked by traditional recruiting software. Older systems simply search resumes for keywords from the job description. But if your resume is written in a way that doesn't fit the job, you're not going to get a call -- even though you may be the smartest person on the planet.
It's easy for resumes to be written differently. Say that "smartest guy" is a gal -- or an ethnic minority. Not everyone uses language the same way. And not every overlooked resume is a case of linguistic discrimination; sometimes the problem is a simple synonym.
Scout CEO Andres Blank gives an example: "We say 'sales development representative,' " he says. " Many times, it's called an 'inside salesperson.' Sometimes it's called a 'sales representative.' Sometimes it's called an 'SDR.' And sometimes, salespeople don't have 'sales' in their title at all: They're 'customer success gurus.' "
Scout brainstorms new keywords, creates Boolean search queries and scores the candidates it finds online. Blank says this differs from traditional technology because Scout collects feedback from "the client and improves its queries accordingly."
This is where the "intelligence" part of the technology comes in. It's one thing for software to simply regurgitate words you've given it, searching for them over and over again. But artificial intelligence with NLP can recognize phrases you haven't provided and even generate words on its own.
Learning Company Values
Then there are new NLP tools that go beyond simply streamlining human resource work. One is Growbot, which was developed by a San Francisco-based start-up of the same name and aims to encourage -- and measure -- peer recognition within an organization.
Operating through company Slack channels, Growbot learns "to listen to certain praise cues -- kudos, cheers, props," says Growbot CEO and co-founder Jeremy Vandehey. The tool then reports which employees receive compliments from colleagues; some clients integrate results into employee reviews.
Among the company's clients is Intuit, the personal-finance technology giant. Dana Sednek, who works in talent development at Intuit, says adding Growbot means having "more visibility in the way that we are recognizing and demonstrating our values in our everyday work."
Sednek loves it. "We've got a lot of words that mean one thing internally that . . . might not mean the same things outside of our culture," she says, referring to how Intuit builds vocabulary around company values. "Oftentimes, when we recognize each other, we use our values as an extension of that recognition to say, 'Wow, I was just in a meeting with you and the decision that we made really represents [the value] to be decisive.' "
Intuit set Growbot so it could pick up values such as ''be decisive" or "learn fast."
By relying on Growbot to track adherence to company values, Intuit uses it differently than Vandehey had anticipated -- another illustration that AI can solve old HR problems in new ways.
Intuit also uses Growbot to keep remote workers engaged, says Sednek, who herself works remotely. Together with the tool, the company uses a product called Spotlight to give workers rewards as part of the recognition process.
Spotlight is about recognizing those who go above and beyond their work, Sednek says, not just giving everyday recognition and reinforcement that people normally give each other.
"When you're remote or you're not in the same physical location with people, there's risk of being disconnected from the culture," she says.
On the Horizon
Where will natural language processing take HR next?
Bersin views NLP as "one of the biggest innovations" in HR technology and believes the possibilities are many. While NLP may have only a small foothold in the market today, he points out that at one time "candidate relationship management was a niche market, referral tools that refer candidates to their peers on the internet was a niche market" and "video interviewing was a niche market -- and now everybody is doing it."
Bersin believes this wave of technology is "as big as many of the things we've seen."
Why? Because "the biggest challenge most people have with HR tools is that they're just too hard to use," Bersin says. "It's too hard to find the job, it's too hard to find the candidates, it's too hard to find the transaction ... . If these things allow you to have both a natural language interface and a smarter set of recommendations, I think they're going to take off like wildfire."
Talla's May offers a mechanical comparison to show what he thinks of the technology's potential.
"Think about it like a construction worker who wears an exoskeleton robot and can now lift 400 pounds instead of 100, and do so more safely," he says. "Think about the cognitive equivalent of that, and that's what you're going to see in the workplace . . . . [I]t's going to have a really explosive impact on the ability of all kinds of knowledge workers to ... perform at their best."