Stay Human · The Boss Is an Algorithm Now

When the Algorithm Becomes the Boss

The short version: Most of this site is about the robot that takes your job. This page is about the one that keeps you and makes the job worse. You don't have to be replaced to be managed by a machine. Right now, the majority of US employers track their workers digitally, and most use AI to score them: keystrokes, idle time, "anomalies," a productivity number attached to your name. It's called algorithmic management. If your job feels like it's being run by a stopwatch you can't see, this is why.

The Scale · How It Works · What It Costs · The Trap · What to Do

The Scale

This already happened. It's not coming, it's installed.

That means the surveillance is no longer a warehouse problem or a call-center problem. It's in the laptop of the developer, the designer, the analyst, the person who thought a desk job kept them clear of all this. The camera, the keystroke logger, and the scoring engine now sit inside the same machine you use to do the work.

How It Works

Algorithmic management is simple to describe and brutal to live under. The software establishes a behavioral baseline for each worker. It collects data continuously, across many dimensions. It compares your current behavior to your baseline, flags any deviation as an "anomaly," and generates an automated alert or a score. A human manager used to watch a team. Now the system watches every person, all the time, and hands the manager a dashboard of red flags.

In warehouses, this runs on wearables that track your location and movements down to the second, building a high-resolution map of every pause. Amazon pioneered the model. But the most telling example is white-collar: Meta rolled out software on work laptops that captures keystrokes, mouse movement, clicks, and periodic screenshots, in part to train its own AI on what employees do all day. Lay off thousands, then surveil the survivors to teach the machine their jobs. That's the loop in one company.

What It Costs

The companies sell it as productivity. The data says otherwise, in two directions at once.

It doesn't even work on its own terms. 72% of monitored employees say the surveillance does not improve their productivity. And it drives away the people it watches: 42% of monitored employees plan to leave within a year, compared to 23% of their unmonitored peers. You spend money to install a system that workers say doesn't help, and it doubles the rate at which they walk out the door.

The human cost is the part the spreadsheet misses. Constant measurement turns a job into a test you can never stop taking. It removes the small human slack that every real workplace runs on: the moment to think, to help a coworker, to be a person for thirty seconds. The "anomaly" the system flags is often just a human being, being human.

The Trap

Here's why this belongs in the replacement fight, even though nobody gets fired in it. Algorithmic management is how the human gets squeezed into something a machine can eventually copy. The more your work is reduced to measurable keystrokes and timed motions, the easier it is to argue that a model could do those keystrokes and those motions. Surveillance isn't just control. It's the training data and the justification for the next round of cuts.

So the worker who says "at least the robot didn't take my job" can still end up managed into a version of the job that's stripped of everything human, scored against a baseline, and primed for replacement the moment the math gets close enough. The boss being an algorithm and the worker being replaced by one are the same project at two different stages.

What to Do About It

First, name it. The vocabulary is half the fight, and companies count on you not having the words. "Productivity insights," "workforce analytics," "engagement scoring" are all the same thing wearing a nicer coat. We keep a running translation in the Lexicon.

Second, this is the rare AI fight with a clear organizing target, because it's a workplace issue, not a sci-fi one. The strongest pushback is collective: the unions and guilds already bargaining over what can be tracked, who sees the data, and whether a score can fire you. The New York Times Guild is fighting exactly this. So is the labor movement this brand stands with, named over at Shawn Fain and the UAW.

Third, stay countable. A worker who's on the record refusing to be reduced to a number is the one thing the dashboard can't optimize away.