What You Will Learn
By the end of this chapter, you should be able to:
Think about the last time your phone finished a sentence for you, or a map app steered you around a traffic jam. Nobody stood beside you giving instructions. Software looked at patterns, made a sensible guess and got on with it. That everyday experience is artificial intelligence at work, and it is the subject of this book.
Artificial intelligence, usually shortened to AI, is the simulation of human intelligence in machines. These machines are programmed to learn from experience, reason about problems and make decisions, much as people do.
AI is not a single invention. It is a family of related ideas and tools, and writers use many different names for its parts. Figure 1.1 shows how three of the most important terms fit together. AI is the broad goal of making machines behave intelligently. Machine learning is one way of reaching that goal, by letting systems learn patterns from data instead of following rules written by hand. Deep learning is a particular kind of machine learning that uses neural networks with many layers.
The terms in the box below appear again and again in later chapters, so it is worth getting comfortable with them now.
Key Terms
The wish to build thinking machines is much older than the computer. Ancient myths and folktales describe animated statues and mechanical beings. In ancient Greece, stories about automatons fired the imagination of writers and philosophers. These were legends, not engineering, but they show that people have long wondered whether intelligence could be made rather than born.
Philosophers began to treat the question more seriously in the seventeenth century. René Descartes described thinking in mechanical terms, and later advances in mathematics and logic suggested that reasoning could be written down as a set of rules. Together these ideas gave the first AI researchers a theoretical starting point.
The twentieth century turned the old dream into a research program. Electronic digital computers arrived in the 1940s and early 1950s, with machines such as ENIAC and UNIVAC, and researchers finally had a tool powerful enough to test their ideas.
One of the most influential figures of this period was Alan Turing, a British mathematician and codebreaker who is widely regarded as a founder of both computer science and artificial intelligence. In 1950 he published a landmark paper, “Computing Machinery and Intelligence,” in which he proposed what became known as the Turing Test. The idea is simple. If a person cannot reliably tell a machine’s answers from a human’s, the machine can be said to behave intelligently. The test gave researchers a way to talk about machine intelligence in practical terms.
The 1950s and 1960s were years of great optimism. Scientists explored logic, problem-solving and symbolic reasoning as the basis for intelligent machines. One early success was the Logic Theorist, written by Allen Newell and Herbert A. Simon, which proved mathematical theorems by using symbolic reasoning. Expectations ran high.
Then progress slowed. In the 1970s and 1980s, ambitious projects such as expert systems and other forms of symbolic AI ran into trouble, because the programs could not cope well with uncertainty, context or the untidiness of the real world. Funding and enthusiasm faded in a period that is often called the “AI winter.”
Interest returned toward the end of the twentieth century. Better algorithms, such as backpropagation for training neural networks, together with more computing power and huge collections of data, made machine learning practical. In recent decades, advances in deep learning, reinforcement learning, natural language processing and computer vision have driven rapid growth. AI now supports virtual assistants, recommendation systems, self-driving vehicles, medical diagnosis and advanced robots, and it has changed industries such as healthcare, transport, finance and entertainment.
Researchers are now exploring quantum computing, cognitive architectures and approaches inspired by biology. As they do, the ethical and social questions grow more urgent. The goal must be AI that is built and used in line with human values, safety and accountability.
The story of AI is the story of a very old curiosity about the nature of intelligence, and it is far from finished. Table 1.1 summarizes the main stages.