DARPA's Strategic Computing Initiative

In October 1983 DARPA announced the Strategic Computing Initiative, a plan to spend about a billion dollars over ten years on the technology of machine intelligence and the applications that would justify it: an autonomous land vehicle, a pilot's associate that would help fly a fighter, and a battle-management system for the Navy. The document cited Japan's Fifth Generation programme explicitly. The money that had left AI in 1974 came back, larger and with military objectives attached.

The programme paid for a great deal that lasted. Carnegie Mellon's autonomous vehicles, including the neural-network-driven ALVINN, were funded by it, as were advances in speech recognition, parallel hardware and expert systems. It also paid for the university laboratories through the years when industry was collapsing around them.

By 1987 DARPA's own management was concluding that the applications were not going to arrive on schedule, and the programme was wound down as the Cold War ended. Its historians Alex Roland and Philip Shiman judged that it had "failed to achieve machine intelligence" and succeeded at almost everything else. The self-driving cars of the 2010s were, in a direct line, Strategic Computing's autonomous land vehicle finally working.