Other meanings of Artificial intelligence
Computer science
Artificial Intelligence is a landmark textbook by Patrick Henry Winston, first published in 1977 by Addison-Wesley. The book systematically introduced the core concepts of symbolic AI, including search, knowledge representation, and learning, and became a standard reference for undergraduate AI courses for decades. Its clear, example-driven exposition influenced a generation of computer scientists and shaped the early curriculum of the field.
The book grew out of Winston's teaching at the MIT Artificial Intelligence Laboratory, where he was a protégé of Marvin Minsky. The first edition (1977) was among the first textbooks to codify the then-nascent field of artificial intelligence, drawing on the symbolic tradition exemplified by the Lisp programming language and early expert systems. It was revised in 1984 (second edition) and 1992 (third edition), each time incorporating new developments such as neural networks and machine learning, while retaining its focus on core symbolic methods. The third edition added a chapter on connectionist models and expanded coverage of logic programming. The book's longevity reflects its role in defining AI as a coherent discipline separate from computer science generally.
The text is organized around the central problem of building intelligent agents that reason, learn, and perceive. Early chapters cover search algorithms (depth-first, breadth-first, A*), knowledge representation (frames, semantic nets, production rules), and natural language understanding. Later chapters address vision, learning (induction, explanation-based learning, neural networks), and planning. Each chapter includes worked examples, exercises, and Lisp code fragments. A distinctive feature is the use of the "blocks world" domain—a simulated environment of colored blocks on a table—to illustrate planning and perception. The book also discusses the historical context of the field, including the Dartmouth Conference (1956) and the work of pioneers like John McCarthy, Allen Newell, and Herbert A. Simon.
Winston's textbook was widely adopted in universities across the United States and Europe, selling over 200,000 copies across its editions. It was praised for its pedagogical clarity and for bridging theory with practical implementation. However, critics noted that its emphasis on symbolic AI came at the expense of emerging connectionist and statistical approaches, which gained prominence in the 1990s. The third edition attempted to address this but was seen by some as insufficiently deep on neural networks. Despite this, the book remains a classic reference for the "Good Old-Fashioned AI" (GOFAI) paradigm. Its influence waned after the publication of Russell and Norvig's Artificial Intelligence: A Modern Approach (1995), which adopted a more agent-based, probabilistic framework.
Patrick Winston originally intended the book to be titled Artificial Intelligence: An MIT Perspective, but the publisher opted for the simpler title. The first edition's cover featured a photograph of the iconic MIT AI Lab "moon" robot, a six-legged walking machine. The book's example programs were written in MACLISP, a dialect of Lisp developed at MIT; later editions included Common Lisp code. Winston also produced a companion video series, Artificial Intelligence: The MIT Perspective, which was distributed on VHS tapes to universities. An obscure fact: the third edition contains a footnote (page 512) acknowledging that the author's own AI program, "Sundry," could not solve the famous "Sussman anomaly" in planning—a problem later solved by the NONLIN system. The book was never translated into Japanese, despite MIT's strong ties with Japanese AI research in the 1980s.
The third edition (1992) is the most widely cited and remains in print.
Help improve the encyclopedia. Reports go straight to the site manager.