The Question Is Not What AI Can Do — It Is Who Controls It

Every few weeks, a new AI capability makes headlines. A model that writes better than most graduates. A system that diagnoses disease from a scan.

The capability conversation is impressive. But it is also, in an important sense, a distraction.

The question that matters more is not what AI can do. It is who controls what AI does.

When a healthcare system uses AI to triage patients, someone chose what the system optimises for. When a hiring platform filters applications, someone decided what counts as a good candidate.

These are not neutral technical decisions. They are decisions about values, priorities, and power — made by organisations that are rarely accountable to the people affected.

The public conversation needs to shift. Not from optimism to pessimism. But from asking what AI can do to asking who gets to decide what it does — and on whose behalf.