The Invisible AI Manager - How Algorithms Quietly Decide Who Looks Valuable, Productive, or Expendable

The Algorithm Did Not Fire You. It Merely Made You Fireable.

The first generation of workplace AI was sold as an assistant.

It would summarize meetings, draft documents, answer questions, and remove tedious tasks. Workers were told to imagine a clever colleague living inside the software.

The next generation may look less like a colleague and more like an invisible manager.

Twenty-six Meta employees have filed a lawsuit alleging that AI-assisted productivity tools and performance systems helped identify them for layoffs. They say the process disproportionately harmed employees who had taken medical, parental, or family leave. Meta denies that AI made the decisions and says human beings selected the workers.

That distinction will become one of the defining arguments of the AI workplace.

The algorithm did not fire anyone. It produced a score.

The algorithm did not discriminate. It measured activity.

The algorithm did not punish caregiving or illness. It simply noticed that certain workers generated fewer visible signals during certain periods.

Then a human reviewed the information and clicked the final button.

This is how automated power often enters an institution. Not by formally replacing the decision-maker, but by arranging the evidence that the decision-maker sees.

A productivity system decides which actions count. A ranking system determines who appears exceptional, average, or expendable. A dashboard turns incomplete measurements into tidy comparisons. By the time a manager is asked to make a judgment, the software has already framed the available reality.

That does not make every algorithmic recommendation wrong. Organizations have always used records, metrics, and performance reviews. Human managers are inconsistent, political, biased, and sometimes cruel. Data can expose favoritism as easily as it can reproduce it.

But workplace AI creates a new imbalance.

The company can see the worker in extraordinary detail: messages sent, tasks closed, documents edited, meetings attended, response times, office presence, sales activity, customer ratings, perhaps even patterns of language and collaboration.

The worker often cannot see the system evaluating them.

They may not know which data were used, how different activities were weighted, whether leave was correctly excluded, which comparisons were made, or whether a manager overruled the software. They may not even be able to prove that an algorithm influenced the outcome.

This asymmetry matters beyond Silicon Valley.

Hospitals can rank nurses by throughput. Warehouses can measure every pause. Call centers can score tone and emotional compliance. Delivery platforms can quietly reduce access to work. Office employees can be compared through digital exhaust that was never designed to represent the full value of a human contribution.

The people most exposed will not necessarily be the least capable.

They may be the worker who takes time to mentor colleagues instead of maximizing visible output. The parent who uses protected leave. The disabled employee whose work pattern differs from the statistical norm. The person assigned to a difficult project whose progress cannot be reduced to completed tickets.

AI may make management more informed. It may also make management more confident in bad abstractions.

That is the larger transition now unfolding. We have spent years debating whether AI will do our jobs. We have paid less attention to whether it will become the institution through which our jobs are judged.

The central question is not whether a human remains somewhere in the loop.

It is whether that human still possesses meaningful judgment, or merely provides legal authorship for a conclusion the system has already made.

The future workplace may not announce that an algorithm has become your manager.

It may simply become impossible to find the manager who can explain why you were chosen.