Governing the machine

This series is deliberately unexcited. It takes the position that the interesting question about artificial intelligence in government is not capability but governance: who decides, on what information, accountable to whom.

The practical essays are written for the people who actually have to do the adoption, and they argue that technology is the easy part. The harder work is process, deciding what a service is supposed to accomplish, removing the steps that exist only because they always have, and building the capacity to sustain the new way of working after the consultant leaves. Governments that skip that step buy software and keep their old problems.

The more uncomfortable essays are about judgment. When a system recommends, the recommendation acquires authority that nobody voted to give it, and the public official who overrides it needs a reason they can defend. That is a capacity question before it is an ethics question.

The conclusion the series has reached is that adoption without institutional capacity is not modernization. It is delegation.

What is unresolved is procurement. Nothing here has yet answered how a small municipality with no technical staff evaluates a vendor claim it has no way to test.