Governing the machine

Artificial intelligence is not primarily a technology procurement problem for government. It is a governance problem. The important questions are who decides where these systems may be used, what evidence supports those decisions, how their performance is evaluated, who remains accountable for the outcome, and what happens when human judgment and machine recommendation diverge.

Technology can increase public capacity, but only when institutions understand the work they are changing. Automating a poorly designed process preserves the process at greater speed. Buying technology without the people, expertise and governance needed to evaluate and sustain it transfers judgment from the public institution to vendors and systems the institution may not fully understand.

Government therefore needs enough technical and organizational capacity to remain the principal rather than become the client of the machine. AI adoption becomes modernization only when public institutions retain the ability to decide what the technology should do, measure whether it works, and remain accountable for the consequences.