John Mccarthy
John McCarthy (1927–2011) was an American mathematician and computer scientist who named the field of artificial intelligence, invented the Lisp programming language, championed time-sharing, founded the Stanford Artificial Intelligence Laboratory, and spent his career pursuing a logical route to machine common sense. Few people shaped both the practical tools and the foundational ideas of early AI as he did, and he received the ACM Turing Award in 1971.
His formation was mathematical, and the mathematics shaped the style of everything that followed. Born in Boston in 1927, McCarthy earned a bachelor's degree in mathematics at Caltech in 1948 and a doctorate at Princeton in 1951, then held posts at Princeton, Stanford, Dartmouth and MIT. In 1955, as a Dartmouth mathematics instructor, he wrote the Dartmouth proposal with Minsky, Rochester and Shannon and chose the phrase artificial intelligence for its title, and he organized the 1956 summer workshop that followed. From early on he held that intelligent behavior should be captured by formal, declarative descriptions that a program could reason with, a position that put him in friendly, enduring disagreement with colleagues who favored psychological or neural models.
Lisp is his most widely felt technical legacy. In 1958, at MIT, McCarthy developed the language for symbolic computation on the IBM 704 and published its theory in "Recursive Functions of Symbolic Expressions and Their Computation by Machine, Part I" (Communications of the ACM, 1960). The design rested on a few ideas that were new to practical programming: programs and data share one representation, the list; functions can be defined recursively and passed as values; conditional expressions are first-class; and memory is reclaimed automatically by garbage collection. Steve Russell, one of McCarthy's students, famously realized that the paper's evaluator could be hand-coded as a working interpreter, which made the notation an actual language and not just a mathematical device. Lisp became the main language of AI research for decades, and its descendants Scheme and Common Lisp, together with ideas it popularized, run through today's Python, JavaScript and functional languages.
His second great project was making computers shareable. Annoyed by batch processing, in which each user waited hours for one run, McCarthy proposed time-sharing in the late 1950s: a machine rapidly switching among many users, each of whom feels in sole command of it. His 1959 ideas fed MIT's Compatible Time-Sharing System, developed under Fernando Corbató, and the model of interactive, multi-user computing that followed. He also talked of computing as a public utility, an idea that reads like an early description of cloud services. On the AI side he and Minsky started the MIT AI project in 1959, and McCarthy moved to Stanford in the early 1960s, where he founded the Stanford AI Laboratory in 1963 and spent the rest of his career; SAIL became one of the field's great centers, training students in robotics, vision, speech and mathematical theorem proving.
His research agenda was common sense through logic. The Advice Taker, described in "Programs with Common Sense" (1959), imagined a program that accepts statements in a formal language and draws their consequences to decide what to do. Working with Patrick Hayes he introduced the situation calculus in 1969, a logical way of describing actions and how they change the world, which exposed the frame problem: how to say what does not change when something happens. In 1980 he proposed circumscription, a form of nonmonotonic reasoning that lets a reasoner jump to conclusions, such as assuming only the known objects exist, and retract them when more facts arrive. He also contributed to computer chess and the alpha-beta pruning idea, and to the mathematical theory of computation, and was awarded the Kyoto Prize in 1988 and the US National Medal of Science in 1990.
A few corrections are worth making. He did not write the first interpreter of Lisp, and he was not the person who first built a working AI program; his gifts were naming, conception and formal framing. His emphasis on logic is sometimes presented as a failed program, since statistical and neural methods now dominate, yet the questions he raised about knowledge representation, defaults, causal reasoning and what a system should do given partial information remain unsolved at the heart of modern AI, and are visible whenever language models are paired with explicit knowledge, planning or program-based reasoning. He was also stubborn about the label: he rejected the view that AI had to imitate human psychology, saying the goal was machines that solve problems intelligently by whatever method. He died at Stanford in October 2011.
Read John McCarthy as the person who gave AI its name and its logical ambitions: a coiner of terms, builder of tools such as Lisp and time-sharing, and a lifelong advocate of the idea that common sense can be written down in formal language and reasoned with.