I’m super skeptical that it’s actually a financial cost saver. The data infrastructure and sunk costs are quite high for systems like this, and the non-zero defect or failure rates mean you still generally have to employ most of the same staff. They just do something else on the computer or deal with other gate check issues, stuff they generally still had to do already while scanning people onto the plane.

Face recognition systems are notoriously difficult to implement in a way that actually saves money.

Even if boarding huge flights happens faster, the limiting factors are usually runway traffic delays and other problems anyway.

All this infrastructure only so sometimes on a few international flights, when no special issues pop up, they can possibly save a small amount of money (relative to other costs involved) by boarding slightly faster (a matter of a few dozen minutes, max) and potentially depart sooner if there’s no other type of hold up.

It doesn’t add up.

  The data infrastructure and sunk costs are quite high for systems like this, and the non-zero defect or failure rates mean you still generally have to employ most of the same staff.

Right, which is why one could reasonably speculate that CBP is picking up the cost of the system. Without the capital costs it could easily make sense.

The clincher would be how edge cases are handled. Is CBP paying for a human behind the scenes to remotely override rejections? Is it acceptable customer service for falsely rejected passengers to wait at the gate until shortly before departure when an airline employee can show up to double-check their documents?