The short answer: artificial intelligence as a field was officially founded in 1956 at the Dartmouth Summer Research Project, but the underlying ideas and mechanical automata go back centuries. The term “artificial intelligence” was coined by John McCarthy for that 1956 workshop, making that the moment AI became a named discipline rather than a scattered set of philosophical and mechanical experiments.
Background: Ancient Ideas and Mechanical Roots
The desire to create thinking machines is much older than the term “AI.” Ancient Greek myths featured mechanical servants like Talos, a bronze automaton. In the 13th century, the scholar Albertus Magnus was rumored to have built a talking mechanical head (the story is likely legend, but it shows the idea was alive). The 18th and 19th centuries brought real progress: Jacques de Vaucanson’s mechanical duck (1739) could flap, eat, and digest, and Charles Babbage designed the Analytical Engine (1837) with Ada Lovelace writing what is considered the first algorithm intended for a machine. These were not AI, but they established the twin pillars: mechanical computation and the notion that reasoning might be reducible to steps.
The Birth of the Field: 1950–1956
Three key developments led directly to the founding of AI as a formal discipline:
- Alan Turing’s 1950 paper “Computing Machinery and Intelligence” opened with the question “Can machines think?” and proposed the Turing Test as a practical way to judge machine intelligence.
- Early neural network work: In 1943, Warren McCulloch and Walter Pitts published a mathematical model of artificial neurons. In 1958, Frank Rosenblatt built the Perceptron, a physical machine that could learn to recognize patterns.
- The Dartmouth Workshop (1956): John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon organized a summer research project at Dartmouth College. McCarthy coined the term “artificial intelligence” for this event, and the attendees (including Allen Newell and Herbert Simon, who had already built the Logic Theorist program) left convinced that machine intelligence was achievable within a generation.
So while Turing’s work is often called the intellectual foundation, 1956 is the universally accepted birth date of AI as a named field.
The First AI Programs and the Optimistic Era (1956–1974)
The years right after Dartmouth saw explosive progress. Newell and Simon’s Logic Theorist proved mathematical theorems; later, their General Problem Solver aimed to solve any well-defined problem. In 1966, Joseph Weizenbaum created ELIZA, a natural language program that simulated a psychotherapist and fooled many users into thinking it understood them. Meanwhile, the first industrial robot, Unimate, began working on a General Motors assembly line in 1961. Funding poured in from DARPA and other agencies, and researchers confidently predicted machines would beat the world chess champion within a decade (it actually took until 1997, when IBM’s Deep Blue beat Garry Kasparov).
The AI Winters: When Progress Stalled
By the mid-1970s, the optimism collapsed. The problems were far harder than expected: computer memory was tiny, processing was slow, and early neural networks had fundamental limitations (Minsky and Papert’s 1969 book Perceptrons showed single-layer networks could not solve even simple problems like XOR). Governments and companies cut funding, leading to the first “AI winter” (1974–1980). A second winter hit in the late 1980s when the expert systems boom (rule-based programs like MYCIN) failed to scale and maintain themselves, and the Lisp machine market collapsed. These winters taught the field that hype without results leads to funding droughts.
The Modern Resurgence: Machine Learning and Deep Learning
The current era of AI began not with a single discovery but with a slow accumulation of data, algorithms, and computing power. Key milestones include:
- 1997: IBM’s Deep Blue defeated world chess champion Garry Kasparov.
- 2011: IBM’s Watson won the TV quiz show Jeopardy! against human champions.
- 2012: A deep neural network (AlexNet) won the ImageNet computer vision competition by a huge margin, sparking the deep learning revolution.
- 2016: Google DeepMind’s AlphaGo beat top Go player Lee Sedol, a feat previously thought decades away.
- 2020s: Large language models like GPT-3 and its successors showed that scaling up transformer networks (introduced in 2017) could produce fluent text, code, and images, bringing AI into everyday use.
These advances are built on the same mathematical foundations as the 1950s work, but with vastly more data and hardware.
Common Misconceptions About AI’s Origins
Three myths come up constantly:
- “Turing invented AI in 1950.” Turing proposed a test and asked the question, but he never used the term “AI” and did not organize the field. That credit goes to McCarthy and the Dartmouth group.
- “AI is a recent invention.” The field is nearly 70 years old; what changed is capability, not the concept.
- “The first AI was a robot.” The first AI programs were software running on room-sized computers, not physical robots. Unimate was a robotic arm, but it was not intelligent—it followed pre-programmed paths.
Frequently Asked Questions
Who actually coined the term “artificial intelligence”?
John McCarthy, a mathematics professor at Dartmouth College, coined the term in 1955 when writing the proposal for the 1956 Dartmouth Summer Research Project. He later founded the MIT AI Lab and Stanford AI Lab.
What was the first AI program ever written?
The Logic Theorist, created by Allen Newell, Herbert Simon, and Cliff Shaw in 1955–1956, is generally considered the first AI program. It could prove mathematical theorems from Principia Mathematica and even found a shorter proof for one theorem than the original.
Why did AI fail in the 1970s and 1980s?
The failures were due to combinatorial explosion (too many possibilities to search), limited computer memory and speed, and the discovery that early neural networks could not solve certain basic problems. Funding dried up when promised results did not materialize, causing the AI winters.
Is the Turing Test still used today?
It is rarely used as a formal benchmark, but it remains a cultural touchstone. Modern AI systems can often pass a casual version of the test, but researchers now focus on specific tasks like image recognition, translation, and reasoning benchmarks rather than a single all-purpose test.
What was the first commercial AI product?
Expert systems like XCON (1980) at Digital Equipment Corporation are often cited as the first commercially successful AI applications. XCON configured computer systems from customer orders, saving the company an estimated $40 million per year by the mid-1980s.
How does the history of AI affect today’s technology?
Every modern AI system—from voice assistants to self-driving car research—traces its lineage to the algorithms and concepts developed in the 1950s and 1960s: search trees, neural networks, logic, and probabilistic reasoning. Understanding the history helps explain why certain approaches (like deep learning) dominate today and why the field has cycled between hype and disappointment.
Related
Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.
More cards in this drawer
- Skip Ryobi and Milwaukee? Harbor Freight Tools to Buy
- Funny Subreddits That Make You Laugh
- Why Retro Gaming Is Still Worth Playing Today
- Neptune Facts That Will Blow Your Mind
- Survival Crafting Game Fatigue: Why & How to Fix It
- Sell Your EV Now Before Prices Plummet?