NELL, the Learning Computer, Teaches Itself from the Web
From a lonely basement home, NELL constantly surfs the Web, and classifies everything it scans into specific categories. It even makes accurate associations and decisions based on its information database. As one example of the adaptive system's capabilities, the Times reports, "Peyton Manning is a football player (category). The Indianapolis Colts is a football team (category). By scanning text patterns, NELL can infer with a high probability that Peyton Manning plays for the Indianapolis Colts -- even if it has never read that Mr. Manning plays for the Colts."
The researchers allowed NELL to scan the Internet unfettered for six months, during which time the program encountered difficulties with recognizing homophones and differentiating between similar terms and usages. The "knowledge base" apparently sorted "Internet cookies" as "baked goods," which one researcher believes "started this whole avalanche of problems." The scientists offered a helping human hand, though, and corrected the mistake and its aftermath.
The Carnegie Mellon group apparently aspires to create an autonomous system, but NELL is "not there yet." A little human guidance might be necessary for this incredible and evolutionary technological development, anyway, because -- left to its own devices -- NELL might just become a sullen, spiteful and porn-addicted cyber-bully. So, please, as awesome as NELL is, just don't give it legs.