Supervision tax

The recurring human cost of keeping an AI system safe and correct enough to use: verification of its outputs, oversight of its decisions, the alignment and labeling labor that made it safe, and the monitoring, incident response, and literacy that keep it that way. Named a tax because it is levied on every unit of AI work, recurs for the life of the system, and is easily treated as someone else's problem.

The fraction of "saved" human effort that must be put back in as ongoing oversight of the AI (reviewing edge cases, catching wrong outputs, handling escalations, monitoring, re-tuning). Applied as: adjusted savings = gross savings x (1 minus tax rate). Typically 15 to 25 percent mature, higher early or customer-facing, never zero.

The recurring cost of keeping a human meaningfully in or on the loop of an AI system, reviewing outputs, sampling decisions, and being ready to intervene. It is set by the stakes of the decision, not by whether you built, bought, or wrapped, and it does not go away by choosing to rent the capability. Removing it to lower cost, as Michigan did, is what turns a cheap capability into an expensive disaster.

The ongoing human labor cost of reviewing, correcting, and handling what an AI system escalates. A major TCO line, treated in depth in Topic 8.2 (see Topic 8.2).

The ongoing human cost of keeping an AI system safe and correct: reviewers, escalation handlers, retraining, and incident response. Frequently the largest and most-omitted cost, it is the number that most often turns a proud build into a sensible buy or kill. (see Topic 8.2)