Phalanx: a machine that closes the loop itself
Production began in 1978 and in 1980 the system was first installed aboard the carrier USS Coral Sea. Phalanx searches, detects, evaluates, tracks, engages and assesses the kill on its own, without outside input.
Why it matters
The first fielded machine to close the whole firing loop by itself. The argument about autonomous weapons did not begin when neural networks arrived: by the time anyone set out to regulate it, such a system had been on operational duty for three decades.
The Navy's fact file describes the functions plainly: Phalanx "automatically detects, evaluates, tracks, engages and performs kill assessment" and is "the only deployed close-in weapon system capable of autonomously performing its own search, detect, evaluation, track, engage and kill assessment functions". A documented case of what that can mean: on 11 October 1989, ten hours into a deployment, the amphibious cargo ship USS El Paso was test-firing its Phalanx at a target towed by an aircraft when rounds struck the bridge and flight deck of USS Iwo Jima, killing an officer and seriously injuring a petty officer, as the Atlantic Fleet told the press the next day. The record rests on the Navy's fact file for the system and a wire report of the day for the accident, and carries medium confidence. There is no machine learning here and no pattern recognition in the modern sense: this is radar-guided, computer-controlled logic. That is the point of the record — the line marked "without human intervention" was crossed first not by a model but by an automaton.