On April 22, at a UN conference in Geneva, Nobel laureate Geoffrey Hinton compared artificial intelligence development to a very fast car with no steering wheel, arguing that capability is outpacing governance.
The metaphor works because it separates two things often conflated in the debate. Speed isn't the stated problem; the absence of steering is. The argument doesn't ask for development to stop, it asks for a proportionate control mechanism to exist.
The month's context gives the image concrete weight. Two weeks earlier, a company had announced and restricted on the same day a model finding flaws in software audited for decades, a decision taken by the company itself, with no public structure to assess it.
That gap is exactly what the following months would try to fill, with an executive order creating pre-release review, a national security memorandum and negotiation of voluntary standards between government and the three largest companies.
What the debate still hasn't resolved, and the metaphor doesn't reach, is who builds the steering wheel and with what legitimacy, in a sector where those with the technical information to design the control are the ones who would be controlled.