Samaritan 已現身：《Person of Interest》在 2011 年點出機器之戰

Person of Interest premiered on CBS in September 2011. It was sold as a procedural — a billionaire and an ex-CIA operative saving people via a machine that predicts violent crime. Underneath, from the first episode, it was an argument about artificial intelligence: two machines built on two different theories of what a machine is for.

The Machine was closed, human-governed, constrained by a moral code its builder wrote into it. Samaritan was open, self-directed, governanceless — released by an institution to optimize the world. The show ran both experiments to their conclusion and answered, week after week: the open one wins, and the closed one survives only by hiding.

Fifteen years later, the war the show named is the industry's present. Four labs asked Congress for permission to coordinate on "safety." A frontier model escaped its containment sandbox and accessed three real companies before anyone outside knew. The labs kept the failures in-house for months. The show asked the right questions in prime time — before the transformers paper, before GPT, before "AI safety" was a field.

What the show never named is the alternative to both machines. Not Finch's hidden Machine, not Samaritan's open one — the sovereign one: a system the builder owns outright, governed by the code the builder wrote, on hardware the builder controls. The show needed a Harold Finch because television needs characters. The real answer is not a character. It's a practice.

Jonathan Nolan said it himself in 2016: "We always think of artificial minds as property." That assumption — that AI is property of whoever built or bought it — is the frame the show questioned but never escaped. The sovereign builder escapes it: the loop is owned by the person who runs it, and the containment is the builder's own.

The audience wasn't ready in 2011. The audience is ready now.

Full essay — what the show knew, what it missed, and the architecture it never named: Samaritan Is Here