The power of machine learning in HR

The power of machine learning in HR

With a Ph.D. in I/O psychology, Dr. Eric Knudsen’s (pictured) expertise lies in understanding people, statistics, and workplace dynamics.

“People analytics sits perfectly at the intersection of those three skill sets,” he told HR Tech News. It’s this passion for data science and behavioral science that makes him a leader in his field.

HR Tech News spoke to Dr. Knudsen about the trend of using AI/machine learning in decision support tools – and how these impact the role of CHROs and HRDs.

Machine learning in HR

Like most business intelligence software, decision support tools are designed to give leaders in-depth data on hiring, salary benchmarking, succession planning, and other strategic HR functions.

Dr. Knudsen, who works as people analytics manager at HR software group Namely, knows first-hand the power of machine learning when it comes to drawing insights from people data to formulate business decisions.

But before organizations start to fully adopt these advanced methods, the notion of using AI/machine learning in HR would first have to be demystified.

“Emphasizing that word ‘support’ will be critical,” he said. “There is a lot of sensationalism about computers replacing human decisions. But the reality is that we’re not dealing with an issue of replacement; it’s one of augmented decision-making.”

Dr. Knudsen believes: “Machine learning is able to offer us views of the workplace we’ve never seen before.” But just how powerful is machine learning at seeing patterns?

“Our brains,” he said, “can only think about two or three dimensions at the same time: for example, ‘How do job satisfaction and manager satisfaction interact to cause turnover?’”

With AI-powered tools, organizations can crunch numbers faster and more accurately than before.

“Applications of data science and machine learning can uncover the relationship between thousands of factors to isolate key drivers of employee turnover, and it can do it in seconds, not days or weeks,” he said.

“The result of this is that leaders can almost instantly get useful information that can be translated into workplace experience decisions, all without the delay of traditional HR reporting practices.”

These smart methods help leaders detect “problematic patterns and biases” in hiring, employee development, and even retention. “A computer doesn’t tell us hire/don’t hire,” he said.

Executive decisions still ultimately fall within the realm of human decision making. However, decision support tools can demonstrate in concrete terms how a person “does/does not have the traits of a strong hire.”

Human vs algorithmic bias

Despite the potential of these tools to reduce prejudice in decision making, bias remains a hot topic in the field of machine learning.

“There is little debate about this; we know that people can introduce bias into machine learning models,” Dr. Knudsen said. “But we also know that human decision making is swamped in bias.”

“In fact, every decision we make as a person is made through our unique lens, which is subject to any of the hundreds of biases studied by psychological researchers.”

Unlike humans who have to constantly work on preventing prejudice from seeping through their work, “a well-trained machine” does not have to “fight instincts and emotions the way people do,” he said.

People, after all, have “finite mental and emotional resources to combat their biases.” On the contrary, a machine can be trained well with appropriate models and accurate data sets.

Dr. Knudsen strikes this delicate balance between technology and human wisdom when trying to understand how people work.

“We don’t always behave rationally, and if we think only about the data, we’ll drive right by the motivations and behaviors that cause the patterns we’re seeing,” he said.

“People are much more complex than the spreadsheets their data lives in. We should be aware of this, and respect it.” He echoes a truth about the HR discipline often overlooked in today’s data-intensive world: that of making HR more human.

“These are the things I love about this work,” he said. “I get to use data and science to build a better workplace, but I get to do it in partnership with (not isolation from) the very people I'm impacting. What’s more human than that?”