AI for performance appraisals – and why your people deserve better
According to an article in the FT in October 2025, JP Morgan Chase and Boston Consulting Group (BCG) are now allowing employees to use in-house AI tools to help draft their year-end performance appraisals.
Both organisations have positioned this as a time-saving tool rather than a decision-making one, but as someone who works in people development, there’s something about this that doesn’t sit quite right with me. And at this early, imperfect stage of AI, it already feels like we are crossing a line.
AI doesn’t know your employees. It might be able to polish words, but it can’t understand nuance, context, or the human side of performance.
The narrative still needs to be shaped before AI gets involved, but here’s the reality: In a world moving at lightning speed, where everyone is stretched thinner than ever, how much time will truly be spent on the thinking required BEFORE hitting the AI button?
And that’s exactly my worry. Performance reviews may soon become just another quick AI output…which raises some initial questions for me:
❓Will everyone start sounding the same?
“Tom/Mary consistently exceeded expectations in delivering against KPIs…”
“Showed strong business acumen through reliable achievement of revenue objectives…”
❓Will review reports adopt that AI synthetic text feel?
– Capitalised section headers
– Lists of three generic ideas
– Short, clipped sentences
– Over-formatted bold phrases
– Catchy empty bullet points
– Lots of ‘em’ dashes (—)
❓As a manager, will people really trust your feedback when it’s obvious that you didn’t even write it yourself?
At the end of the day (or year as is normally the case for appraisals) the purpose of a performance appraisal is to provide meaningful and personal feedback – supported by concrete examples and actionable suggestions. AI can’t generate that; only a manager can.
If AI helps you polish your words after you’ve already decided what you need to say, fine. But even then, I would question whether the text needs to be polished much anyway, if you are going to have a direct human conversation about it. Afterall, it’s not an external communication piece so simple, direct and human is probably best.
And yes. Although a conversation is a key part of the appraisal process, it’s the written review that lives in print for the next 11 months – and that can impact employee perceptions, engagement, motivation, and ultimately career trajectories, for both good and bad.
AI-generated text may be efficient, but it often strips out meaning and feeling. People deserve better.