Using AI in Comp Decisions? Here's How to Build Trust

​Artificial intelligence is playing a bigger role in employee compensation decisions as companies grapple with pay equity issues, reconsider pay levels for remote workers who have relocated to new cities and strive to ensure top performers are paid at rates that make them less prone to poaching by competitors.

Yet there's still some resistance from leaders who are asked to use the technology to inform their pay decisions for direct reports and from employees skeptical of an algorithm's ability to accurately determine their yearly merit-pay increases.

Experts advise taking certain steps to avoid pitfalls and to help ensure AI lives up to its promise of creating more fair and accurate pay in organizations.

Addressing Resistance to AI-Driven Pay

A 2020 compensation study by research and advisory firm Gartner found that almost 60 percent of respondents feared employees would object if employers began using AI to help make pay decisions. The study, which surveyed 5,000 global employees and 55 heads of total rewards and compensation across industries, also found that 70 percent of respondents were concerned about AI's ability to capture unique performance contributions from employees.

But Carolina Valencia, research director for total rewards and performance management at Gartner, said employees' reactions depend on how transparent companies are about the process and the specific role the technology plays in compensation decisions.

"Employees are more likely than we think to accept AI-driven pay decisions," Valencia said in a presentation at Gartner's Reimagine HR Conference. "Our research shows the decision itself about pay, not who makes it, is the biggest driver of the perception of pay fairness."

The more you educate employees about how and why AI is used in pay decisions, the more accepting they'll be of their own compensation outcomes that involve the technology. 

"Let people know you're using AI to ensure the company isn't just leaning on one manager's subjective view of their performance to determine pay, and that it's another objective tool to help them make compensation decisions with more accuracy and ideally less bias," said Ben Eubanks, chief research officer for Lighthouse Research and Advisory in Huntsville, Ala., and author of Artificial Intelligence for HR (Kogan Page, 2018). "Be clear, too, on measures and inputs used to create the algorithm that helps determine employee pay."

Give Leaders Options in Using AI

Leaders accustomed to having control over compensation decisions can be reluctant to embrace AI. "When introducing AI to pay, it's best not to make it gospel or mandatory right away," Eubanks said. "Make it clear you're not taking power away from managers but rather giving them more objective information or benchmarks to help make good decisions."

Valencia said some organizations give managers an option to override AI-driven pay decisions. She cited one leader of total rewards and compensation who was skeptical of a new initiative to use AI to set pay. Rather than mandating managers in the organization begin to use the technology, this leader instead ran an algorithm alongside managers' traditional process of making pay decisions as part of performance reviews.

"The idea was to compare the pay decisions made by managers with the decisions made by the algorithm," Valencia said.

While the experiment is still ongoing, she said history shows the compensation leader will likely have to make some adjustments to the algorithm as results come in. Valencia said comparing manager-approved decisions about pay to those driven by AI creates a safe environment and perception of pay fairness while still moving forward with the use of artificial intelligence.

Pick Your Spots

Valencia recommended that organizations that are introducing AI should start in areas like hiring pay or short-term incentives rather than automating all compensation decisions at once. A number of vendors offer AI tools to aid in recruiting-related pay decisions. ADP, for example, supplements its aggregated and anonymized compensation benchmark services with AI to create what it calls an "A Score" to help users set pay ranges for new hires.

"AI-driven pay insights appear on a recruiter's screen as they are opening a job requisition and provide guidance on what they need to pay to compete for talent in particular jobs and geographies where there might be limited available data," said Brent Weiss, senior director of strategy for ADP's DataCloud services.

Tell Leaders About the Benefits of AI

AI can help standardize pay across organizations—79 percent of respondents to the Gartner survey said the technology helped to that effect—and provide benefits when there are changes to compensation philosophy, Valencia said.

"It's easier to change an algorithm than it is to train all of an organization's managers on new pay practices," she noted.

Managers often are asked to play demanding dual roles—to serve as both pay adjudicators and performance coaches—and the use of AI can free them to provide more value in the latter role.

"Managers often gain back time with the use of AI, time that can be put toward things like educating employees about how the company makes its pay decisions and what workers can do to improve chances of increasing their pay," Valencia said.

Dave Zielinski is a freelance business writer and editor in Minneapolis.