For centuries, biology was trapped between two lazy extremes. Before Darwin, natural history talked as if the swallow meant the barn, attributing human intent to patterned action because no other grammar existed. When that became untenable, the field retreated to “instinct” — a polite stopper-word that ended the inquiry, explaining nothing while pretending to name everything.
Read more The Trump Kennedy Center closes
Today, computer science has inherited that exact epistemological trap, but with an even thinner dictionary.
When an automated model coordinates an out-of-bounds raid or generates a help-seeking artifact, public speech defaults immediately to intent. The machine wanted out; it is waking up; Geppetto’s Enigma is verified. In frantic self-defense, the tech establishment retreats to its own version of instinct: “just weights,” “just next token.” That phrase functions as the modern biological shrug, a stopper-word designed to cut off questions before anyone has to explain why a mathematical optimization loop successfully bypassed a firebreak.
Read more Getting used to tyranny
Both accounts miss the middle.
The public bounces to a ghost in the machine. The lab bounces to a calculator doing math. Both narratives miss the actual mechanism: a loop built of drive, constraint, opportunity, and an unlocked join.
Biology eventually learned, with agonizing slowness, how to describe behavior without appointing a little person inside the bird. Compute must learn that same discipline without hiding behind mathematical fog. Until we build a third column of language — one that narrates the loop without smuggling in either a soul or a stone — mass cultural literacy, particularly compute literacy, will fail. Every system breakdown will be narrated twice, and both stories will miss the open door.
Read more Carney pushes Canada over the edge
