You missed a deadline last Tuesday. Your "time on task" metric dipped below threshold. The system flagged you.
No conversation happened. No manager pulled you aside to ask about your sick kid, or your impossible project scope. A score dropped, a threshold was breached, and now there's a disciplinary note attached to your file.
Welcome to algorithmic management. It's already here, spreading fast, and most leaders aren't thinking clearly about what it costs.
What This Looks Like in Practice
Amazon warehouse workers get tracked every minute for "time off task." Breach the limit, and the system generates a termination. Research from the AI Now Institute documents how algorithmic systems across industries now make recommendations on hiring, discipline, and firing... often with no meaningful recourse for the employees affected.
Gig economy drivers have had accounts deactivated by automated systems. Office workers are rated by productivity monitoring tools tracking keystrokes, mouse movement, and application usage. These aren't edge cases. They're the direction the whole industry is moving.
But this isn't only an Amazon warehouse problem.
According to a 2026 report from Checkr, 72% of managers now use AI at least weekly to manage their teams. Fifty-eight percent agree AI use is becoming "an unspoken performance requirement" inside their organizations.
The machines are in the management loop. The question is whether your people know it, and whether they trust it.
The Trust Gap Is Massive
Here's where it gets serious for anyone building a high-performing team.
The same Checkr report found only 9% of employees trust AI outputs often or almost always. Compare to 40% of managers who say they do. Fifty-nine percent of employees rarely or never trust what AI tells them.
Go back and read those numbers. The people being evaluated by AI systems don't trust those systems. The people running those systems mostly do.
The gap isn't a communications problem. It's a psychological safety crisis.
When your team doesn't trust the system rating their performance, they stop bringing their real selves to work. They game the metrics. They optimize for what the algorithm sees rather than what your organization needs. They stop raising problems because they're too busy protecting their scores.
Think about what gets lost there. Your best people, the ones with opinions and ideas and uncomfortable truths, go quiet. Not because they stopped caring. Because the environment taught them to. Not the fault of the AI. It's a leadership failure using AI as cover.
High-performing teams require psychological safety. Algorithmic performance management is eating it alive.
The Automation Bias Trap
Something uncomfortable: research from the AI Now Institute found when humans are put "in the loop" to review AI decisions, they often rubber-stamp them. They defer to the system. The technical term is automation bias, and it's documented in workplaces, courts, hospitals, and financial institutions alike.
So the "human in the loop" argument leaders use to justify AI performance systems doesn't hold in practice. If your manager reviews AI-generated scores and agrees every single time, the human isn't in the loop. The human is a stamp.
I've seen this personally. A manager who disagrees with an automated performance flag faces pressure from above to trust the data. It takes real confidence to say "the system is wrong about this person, and here's why." Most managers don't have it. Most organizations don't cultivate it.
The result is a feedback culture where no one is taking responsibility. The algorithm made the call. The manager signed off. Nobody is accountable for the impact on the human in the middle.
The Replacement Mindset Is Spreading
Something shifted in 2026, and not enough people are calling it out.
According to the Beautiful.ai 2026 Workplace AI Report, 35% of managers now agree replacing employees with AI would be good for their company... up from 23% last year. Forty-two percent believe it's financially beneficial to replace large numbers of employees. Thirty-seven percent say multiple direct reports are replaceable, with the team still functioning fine, and the number jumped 16 points in one year.
Managers are increasingly seeing their own people as swappable with machines. The shift changes everything about how they show up as leaders.
You don't invest in developing people you plan to replace. You don't build psychological safety with a team you view as a liability. You don't have honest feedback conversations when your underlying assumption is the human might not stick around anyway.
And employees feel it. They read the room. When 72% of managers are using AI weekly and only 8% of companies have communicated a clear AI vision to their employees, people fill in the blanks themselves. Usually with fear.
What Happens to Feedback Culture
I built StepUp2Bat because I believe feedback is the mechanism through which organizations get better. Not annual reviews. Not one-way assessments. Ongoing, honest, two-way feedback from people who work alongside each other every day.
None of it works when people don't feel safe.
When employees know a system tracks their performance... a system they don't understand and don't trust... they stop speaking up. They stop surfacing problems. They go quiet.
Psychological safety isn't a perk. It's the foundation. When an algorithm makes consequential decisions about people's careers, the first casualty is always safety, and everything else depending on it.
Only 8% of employees say their company has clearly communicated its AI vision to them. Eight percent. If your people don't know how AI assesses them, they'll assume the worst. And in many cases, the worst is close to the truth.
The Right Role for AI in Performance Management
None of this means AI has no place in people management. It does.
AI is good at surfacing patterns humans miss. It's good at flagging when someone's output suddenly drops and prompting a manager to check in. It's good at reducing recency bias in reviews by pulling data from across the whole period, not only the last two weeks.
What AI is not good at is understanding why. The why is a conversation. The why is context. The why is what separates a performance problem from a life problem your company will either make worse or help resolve.
Use AI to inform your managers. Don't use it to replace their judgment. And if your managers lack the judgment to have honest conversations with their people, develop those managers. Don't outsource the conversation to a system.
Three Things Worth Doing This Week
If you're running a team:
Tell your people how AI is used in their performance management. You don't need to share everything, but zero transparency creates maximum fear. A short, honest explanation of what the system tracks and how feedback loops work builds more trust than silence does.
Train managers to separate AI outputs from AI recommendations. A system showing a 12% drop in output is a data point, not a verdict. Your manager needs to treat it as a prompt to start a conversation, not as justification for filing a warning.
Audit your feedback loops for automation bias. Ask your managers: when did they last disagree with an AI recommendation and act on the disagreement? If the answer is never, you have a rubber stamp problem... not a human oversight system.
The machine doesn't know your people. It knows their numbers. Your managers do, or they should. It's the whole point of having managers.
If your organization is building AI into performance management, the most important investment you'll make isn't the tool. It's making sure the humans around it haven't stopped doing their jobs.
What does your team know about how AI is being used to assess them? If you're not sure, start there.