The A-Z of AI

Knowledge

AI systems build knowledge in a fundamentally different way from humans.

The knowledge an AI system attains is very different from the intricate knowledge human beings develop over the course of a lifetime.

Humans make sense of the world by automatically drawing on previous life experience to contextualize everything around them. But AI systems don’t have the same kind of background knowledge to call on, or the instinct to do so.

AI systems must be explicitly taught how to cross-reference and analyze a broad spectrum of information, in the form of organized data, in order to build up their experience. There are ceilings and boundaries to what they can learn, determined by the parameters put in place by AI design teams.

Knowledge is as far from information as wisdom is from knowledge.

A lineup of birds in an array of colors, shapes and sizes that have a brightly colored parrot sitting conspicuously among them.

Take an AI system trained to spot a parrot in a set of bird photos. It does not "understand" what a parrot is. It can learn to spot identifying features, such as the colorful feathers and curved beak, but a system’s knowledge is constrained by the kind of information it is given to learn from.