Your company said AI would make performance reviews faster, fairer, and “more objective.” Unfortunately, the model appears to have trained exclusively on passive-aggressive Slack threads, rejected calendar invites, and your manager’s least generous interpretation of “growth mindset.”
The result is a performance review that reads less like professional feedback and more like a legally reviewed burn book.
The Algorithm Has Concerns
It starts gently. “Meets expectations.” “Collaborates cross-functionally.” “Demonstrates ownership.” Then, halfway through paragraph two, the model pivots into something weirdly personal.
While the employee completes assigned work, their overall vibe suggests they may be waiting for everyone else to stop talking.
You ask HR if this is normal. HR says the model is “still learning.” From what, exactly? Your archived eye rolls?
- “Shows initiative, but only after being perceived.”
- “Could benefit from appearing less correct in meetings.”
- “Frequently contributes ideas that create work for leadership.”
Is this publishing?
Wow 👆 Peep that experiment!!!
Anything you added in your rich text field should be displayed right above here! How's it looking? :-)