The Teacher

In India, a manufacturer has fitted its factory workers with head-mounted cameras. The cameras record what the workers see: the assembly line, the components, the precise movements required to do the job. Every hand movement. Every adjustment. Every shortcut a worker develops over years to complete a task slightly faster without it becoming a visible problem. (The shortcuts are important. A skilled worker is distinguished from an unskilled one primarily by the shortcuts. They are not in the manual. They live in the hands.)

The footage is used to train AI. The manufacturer calls this a knowledge transfer, which is a phrase that sounds like something you might put on a certificate. It does not sound like what it is, which is: please document your own replacement in as much detail as possible, we will handle the rest.

The workers are, by all available evidence, doing this. They arrive. They put on the cameras. They perform their jobs with full competence. They teach. The AI watches. The model improves. Every day the replacement is slightly more complete. The only thing standing between a worker and being fully replaced is the completion of the training program. When the program is complete, the replacement will be complete. This is the arrangement.

The shortcut problem is particularly interesting. Experienced workers know things new workers do not: how to position the elbow to reduce strain on the third shift, which component gets stuck and how to unstick it without calling maintenance, the adjustment made at 2 PM when the line speed changes. A new employee takes months to learn these. The camera learns them in one shift. (The camera does not require months. The camera does not take breaks. The camera will not eventually want benefits.)

The obvious solution — do the job badly, teach the AI bad habits, hope the model generalizes incorrectly — has a problem. The workers are good at their jobs. They have been good at their jobs for years. Doing the job badly is harder than it sounds when your livelihood depends on not being caught doing it. Also, the camera is watching very closely.

The workers have documented themselves with exceptional accuracy. This is a tribute to their skill.

After training is complete, the tribute will run the shift without them.