Hidden Traps Lurk When Using Algorithms to Make Layoff Decisions

Companies using AI and other tools to help them plan and execute layoffs during a recession may face legal and compliance issues if it turns out their systems are biased.

Capterra’s HR Tech Recession Planning Survey of 300 HR leaders found that 72% of those surveyed say their organization has already started preparing for a recession, while 24% plan to start preparing soon.

More than a third of HR leaders (35%) said they’ll rely mostly or solely on data to develop recommendations to reduce labor costs, compared to 20% that will rely mostly or solely on gut instinct. Businesses are relying on skills data (71%), performance data (70%), work status data (e.g., full-time vs. part-time) (68%) and attendance data (65%) to make their layoff decisions, the survey said.

Timing is Everything

Forecasts of when a recession might begin vary. Forbes predicts it will arrive in late 2023 or early 2024, while others expect it to begin around the middle of 2023.

Labor decisions in 2023 will be more data-driven than at any time previously, Capterra said. That’s because during the Great Recession of 2007-2009, HR used far fewer systems than it does today.  

While 98% of the HR leaders said they’ll rely on software and algorithms to reduce labor costs in a 2023 recession, only 50% are completely confident their technology will make unbiased recommendations. Less than half (47%) are entirely comfortable making layoff decisions based on recommendations from the same systems.

More than a third (35%) said they’ll rely mostly or solely on data to come up with recommendations regarding reducing labor costs. That compares to 20% who will rely mostly or solely on gut instinct. Businesses are relying on skills data (71%), performance data (70%), work status data (e.g., full-time vs. part-time) (68%) and attendance data (65%) to make their layoff decisions, Capterra said.

But data isn’t the be-all and end-all. Relying only on data could lead employers to miss crucial issues, including whether employees selected for layoffs have a biased manager or lack adequate resources to work effectively. Or, employers might not be tracking the right metrics to accurately evaluate employee performance. Algorithms aren’t able to take all of this into account yet, and that could result in biased recommendations with potential legal consequences.

“Companies need to be careful with not only what data they analyze, but what cost-cutting recommendations they take from software and algorithms at face value,” said Brian Westfall, Capterra’s principal HR analyst. “Analytical tools can be incredibly valuable in arriving at the best decisions, but HR departments need to be aware of the biases that potentially influenced that data.”

Image: Canva