JPMorgan Chase is giving its managers the green light to use AI when drafting performance reviews, a shift that may ease one of the most dreaded chores of corporate life while stirring a whole new set of questions. For employees, the real worry isn’t whether the write up lands on time, but whether it still feels human.
Companies and management experts say AI can save hours and produce clearer, more consistent feedback than the fragmented notes many bosses rely on.
Still, they warn that leaning too hard on a chatbot risks turning reviews into soulless, AI flavored filler. And the truth is, plenty of managers already use generative tools quietly, bending the rules however they want.
“The boundaries are going to be redrawn with this technology shift,” said Benjamin Levick, who runs AI and operations at Ramp.
He said the process could feel awkward or even dehumanizing if companies apply the tech clumsily. But he also thinks managers can strike a balance if they treat AI as support rather than a replacement.
JPMorgan’s guidelines reflect exactly that. Supervisors can use an internal chatbot to help compose reviews, but the bank warns that AI can’t assign scores or make decisions about pay or promotions.
Human judgment still runs the show, at least on paper.
Levick argues that skipping AI altogether might shortchange employees. A manager forced to evaluate a year’s worth of work in a few rushed hours will miss things. Yet the opposite extreme – a bot reading every artifact and spitting out a grade – feels just as wrong.
Human memory also gets messy. Peter Cappelli, a management professor at Wharton, said bosses overvalue the recent past and bring their own biases to the process.
In some cases AI can deliver a more objective assessment than a human. Still, there’s a catch: employees may distrust praise or criticism that doesn’t feel authentically written by their manager.
Research on AI assisted decision making adds another wrinkle. Give people an automated tool and many offload the hard thinking. In performance reviews, that can erode credibility further – employees already treat the annual ritual as corporate theater.
Now they may feel processed instead of evaluated. And because chatbots tend to flatter, AI assisted reviews might skew too positive and shy away from real performance problems. Managers still have to check the tone and accuracy, which some skip.
Companies wrestling with these tensions won’t figure it out this season. Some have already made AI proficiency part of the review criteria – KPMG and Shopify among them – which makes the question even more tangled.
Levick said he uses AI at nearly every stage of his evaluation workflow, mostly to summarize peer feedback and volumes of work submitted over the year. But he insists the final review reflects his own judgment. The machine informs, he says, but doesn’t decide.
Others are testing systems that take a bigger swing. Rippling’s Talent Signal evaluates employees’ first 90 days by measuring their output – resolving customer issues, writing code, or handling tasks of varying complexity – and labels them “high potential,” “typical,” or “pay attention.” Managers can approve or reject the assessment with a thumbs up or down.
According to our analysts, the fight ahead won’t be over whether companies use AI, but how far they let it drive. Some employees may appreciate faster, clearer feedback.
Others may see the tech as one more sign that performance management drifted away from real conversation and into automated triage. Either way, the line between assistance and overreach keeps getting thinner.
