Should AI use be tracked in performance reviews?

Companies seek returns on their hefty investments in artificial intelligence and many big names have turned to tracking employee usage of it as one way to do so.

Still, while almost every Fortune 500 company is now tracking overall AI use — even tying performance review metrics to AI use — “they can’t necessarily tie it to business outcomes quite yet,” Leslie Caputo, organizational psychologist and SVP of global solutions at coaching platform EZRA, told HR Dive via email.

So should companies implement AI use as part of their performance management process? While it can be “a great first step” to making AI part of broader workforce conversations, employers will want to consider a number of nuances inherent to AI adoption before making it part of any review process, Kate Jensen, a director analyst at Gartner focused on HR technology, told HR Dive.

Looking at AI use alone “doesn’t encourage the right behaviors,” Jensen said.

“Employers mostly gain the appearance of control over a massive bet they’ve made,” Caputo said. “Companies have invested enormous amounts into AI, and they are under pressure to deliver ROI. Usage numbers are an easily-achieved metric that leadership can show the board and say, ‘look we’re adopting it.’”

Tracking use alone could allow for the proliferation of AI “workslop” — poor quality outputs from AI tools — and obscure insight into what value AI brings to the table, both Jensen and Caputo noted.

Employers also can’t ignore the potentially trust-damaging messages sent to employees when AI use is flagged in performance reviews.

“We see a lot of concern about AI replacing jobs,” Jensen said, “and I think the use of AI in performance metrics can increase that hesitancy around role replacement.”

It may also “come across as a form of surveillance,” Caputo said, which can make employees doubt that their employers trust their judgment on when and how to use AI.

“This erodes the human capabilities that actually create value in the AI era, like curiosity, discernment and the willingness to push back on bad AI outputs,” Caputo said.

Studies have pointed to heavy reliance on AI leading to skill atrophy, including more traditionally “human” skills such as critical thinking — trends that blanket expectations for AI use could worsen, Jensen said.

AI use can still play an important role in guiding performance discussions, however.

When AI use is discussed with intention and nuance, “it can really encourage innovation and skill development,” Jensen said. Companies intent on improving AI use, for example, may wish to set AI-related goals that aren’t evaluated punitively to encourage innovation, she added.

From there, as leaders see how workers use AI in their roles, they can establish baselines and build expectations based on what they see. “It helps maintain that foundational skill level you need in those roles,” Jensen said.

“The differentiators that turned AI into real performance included discernment, curiosity, connection and humility,” Caputo said. “Smart companies will ultimately shift to measuring how employees execute strong judgement in AI usage, instead of usage or tokens alone.”