WHAT ARTIFICIAL INTELLIGENCE (AI) IS AND IS NOT

DEFINITION OF ARTIFICIAL INTELLIGENCE IN PLAIN LANGUAGE AI is the practice of building computer systems that perform tasks which, when a person does them, we would say require intelligence. Examples of such tasks include understanding a spoken sentence, recognizing a face in a photograph, translating between languages, and deciding the next move in a game. The word artificial simply means made by people rather than occurring in nature. The word intelligence is harder, because even scientists disagree about exactly what it means. For this book, a practical meaning is enough. An AI system is a computer program that handles tasks where the correct behavior is not written out step by step in advance, but is instead worked out from patterns, examples, or goals. To see why this matters, think about how ordinary software works. An ordinary program is like a recipe. A recipe lists exact steps, and if you follow the steps you get the dish. A programmer writes every step, and the computer follows them exactly. This works wonderfully for tasks like adding up a shopping cart, because a person can describe the exact steps ahead of time. But some tasks are very hard to describe as steps. Try writing exact steps for recognizing a cat in any photograph, whether the cat is sleeping, leaping, half hidden behind a sofa, or photographed in the dark. Nobody can write all the rules. AI offers another route. Instead of writing the rules, we show the computer many examples and let it find the patterns for itself. That shift, from writing the rules to learning the patterns, is the heart of modern AI. It is also the source of both its power and its unpredictability, because a system that learned its own patterns may behave in ways that its builders did not directly write down. It is important to say clearly what AI is not. AI is not magic. AI is not a person, and it does not have feelings, wishes, or a life of its own, even when it writes in a friendly voice. AI is not automatically correct. AI is not one single thing either. The word covers many different techniques, from very simple to very advanced, and part of becoming a harness engineer is learning to tell them apart.