Cyc (from “encyclopedia”) was started in 1984 by Douglas Lenat at the Microelectronics and Computer Technology Corporation (MCC), a research consortium in Austin, Texas. The project’s aims were laid out by Lenat and colleagues in the paper “CYC: Using Common Sense Knowledge to Overcome Brittleness and Knowledge Acquisition Bottlenecks,” published in AI Magazine, volume 6, issue 4, in 1985. This entry dates the milestone to the project’s 1984 start.
Cyc was a response to a frustrating weakness of expert systems: they were “brittle.” A medical or configuration system could perform expertly inside its narrow domain but had no grasp of the obvious background facts every person knows, such as that water is wet, that people cannot be in two places at once, or that buying something means you then own it. Lenat’s bet was that this missing common sense could be written down explicitly, as a vast hand-built knowledge base of facts and rules, and that doing so would take a decade-long, large-scale effort rather than a clever algorithm.
The project ran far longer than its original plan and continued for decades through a spin-off company, Cycorp. It accumulated millions of hand-curated assertions about the everyday world. Cyc is admired as an audacious attempt to tackle common-sense reasoning head-on, but it never became the foundation for general AI that its founders hoped, and the field largely moved toward learning knowledge from data instead of hand-coding it.
Why business readers should care: Cyc represents the high-water mark of the “knowledge must be entered by hand” philosophy. Its limited payoff after enormous effort helps explain why modern AI bet instead on machine learning, where systems absorb knowledge from data rather than from human-written rules.