A short history of AI and computers
The history of AI and computers runs from calculating engines designed in the 1820s, to electronic computers in the 1940s, to AI named as a field in 1955 and 1956. Early AI followed rules that people wrote. Then came systems that learned from examples. Newer systems generate text from patterns they learned in huge collections of examples.
Last reviewed: 20 October 2026 7 min read
What was the first computer?
It depends on what you count as a computer. Each "first" below is a different kind of first.
| Kind of first | Machine | When |
|---|---|---|
| First design for a general-purpose, programmable machine | Charles Babbage's Analytical Engine (never finished) | conceived in 1834 |
| First electronic machine used to break secret messages | Colossus, at Bletchley Park in England | working in early February 1944 |
| First large electronic computer shown to the public | ENIAC, at the University of Pennsylvania | announced to the public on February 14, 1946 |
| First program run from an electronic computer's own memory | The Manchester "Baby" | June 21, 1948 |
So a fair short answer is: Babbage designed the first programmable computer, and the first electronic computers were built in the 1940s.
When was AI invented?
The name "artificial intelligence" was first used in a written proposal dated August 31, 1955, for a summer study in 1956. Dartmouth College, where that 1956 workshop took place, calls it the birth of the field.
The ideas are older. In 1950 Alan Turing asked whether machines can think, and suggested a test he called the imitation game.
A timeline of AI and computers in 25 moments
Each row has one source. Dates are as precise as the source allows.
| # | Date | What happened | Source |
|---|---|---|---|
| 1 | 1821 | Charles Babbage begins designing a calculating machine, his Difference Engine. In 1834 he conceives a bigger, programmable one, now known as the Analytical Engine. | Computer History Museum |
| 2 | 1843 | Ada Lovelace publishes notes with a step-by-step set of operations for Babbage's engine, often called the first published computer program. | Computer History Museum |
| 3 | 1890 | Herman Hollerith's punched-card machines count the US census, which finishes months ahead of schedule. | US Census Bureau |
| 4 | 1936 | Alan Turing describes an imaginary machine that can carry out any calculation written as steps. | Proceedings of the London Mathematical Society |
| 5 | December 1943 | Warren McCulloch and Walter Pitts publish a simple mathematical model of a brain cell, the starting idea behind artificial neural networks. | Bulletin of Mathematical Biophysics |
| 6 | February 1944 | Colossus, an electronic machine at Bletchley Park, is working, helping to break secret wartime messages. | The National Museum of Computing |
| 7 | February 1946 | ENIAC, a 27-ton electronic computer built at the University of Pennsylvania, is announced to the public. | Penn Today |
| 8 | December 1947 | John Bardeen and Walter Brattain build the first working transistor, the tiny switch inside every chip. | Computer History Museum; Nobel Prize 1956 |
| 9 | June 21, 1948 | The Manchester "Baby" runs the first program stored in an electronic computer's own memory: seventeen instructions long. | Computer History Museum |
| 10 | October 1950 | Alan Turing publishes "Computing Machinery and Intelligence" and proposes the imitation game. | Mind (journal) |
| 11 | 1955 to 1956 | A proposal dated August 31, 1955 uses the words "artificial intelligence" and plans a summer study at Dartmouth in 1956. | Stanford copy of the proposal; Dartmouth College |
| 12 | July 1958 | Frank Rosenblatt's perceptron is shown to the public: a machine that learns from examples rather than following only written rules. | Cornell University |
| 13 | September 12, 1958 | Jack Kilby demonstrates the first working integrated circuit: many parts on one piece of material. | Texas Historical Commission; Nobel Prize 2000 |
| 14 | July 1959 | Arthur Samuel publishes "Some Studies in Machine Learning Using the Game of Checkers", using the phrase "machine learning". | MIT-hosted copy of the paper |
| 15 | April 19, 1965 | Gordon Moore predicts that the number of parts on a chip will keep doubling, at first every 12 months. | Computer History Museum |
| 16 | January 1966 | Joseph Weizenbaum at MIT describes ELIZA, a program that answers typed English using rules triggered by key words. | Communications of the ACM |
| 17 | October 29, 1969 | The first message is sent over ARPANET, an early network that grew into the internet. Only "l" and "o" arrive before the system crashes. | UCLA |
| 18 | 1971 | The first advertisement for a microprocessor, a computer's processor on a single chip, appears. | Computer History Museum |
| 19 | October 9, 1986 | David Rumelhart, Geoffrey Hinton and Ronald Williams show how a neural network can learn by sending its errors backward to adjust itself. | Nature |
| 20 | March 1989 | Tim Berners-Lee writes the first proposal for the World Wide Web at CERN. | CERN |
| 21 | April 30, 1993 | CERN makes the web's software available to anyone, with no royalties to pay. | CERN |
| 22 | May 11, 1997 | A chess computer defeats world chess champion Garry Kasparov. | Computer History Museum |
| 23 | June 2009 | Researchers present ImageNet, a very large collection of labeled photos for training and testing vision systems. | ImageNet |
| 24 | 2012 | A neural network from the University of Toronto wins the ImageNet image challenge with an error rate of about 15%, against about 26% for the next team. | ImageNet |
| 25 | June 12, 2017 | Researchers publish "Attention Is All You Need", introducing the transformer, a design for working with sequences of words. They first test it on translation. | arXiv |
One more date, for the curious: on October 8, 2024, John Hopfield and Geoffrey Hinton received the Nobel Prize in Physics for discoveries that enable machine learning with artificial neural networks.
How did AI change from rules to examples to generation?
AI history follows one thread: rules, then examples, then generation.
Rules. The first AI programs followed rules that people wrote. ELIZA is a clear case: a person wrote rules that matched key words and built replies from them. Rules are easy to read and check. They break when the world does something the rule-writer did not expect.
Examples. From the late 1950s, some researchers built machines that find their own rules from examples. The perceptron learned from labeled examples. In 1986, a method for sending errors backward through a network made learning in bigger networks practical. Large labeled collections such as ImageNet gave these networks enough examples to learn from.
Generation. The newest systems also learn from examples, but instead of picking a label they produce new output, such as a translated sentence. A generated answer can sound right and still be wrong, so it still needs checking.
All three kinds are in use today. Which machines count as AI is still a good question to argue about. "What Counts as AI?" is also a question on the learning path the curriculum is built around.
How do the computers and the AI fit together?
The two histories need each other. AI ideas such as the artificial neuron (1943) and the perceptron (1958) came early. Learning from millions of examples needed fast chips and huge amounts of stored data, which came after. The transistor, the integrated circuit and the microprocessor are part of AI's history too.
What this timeline does and does not show
It shows:
- 25 dated moments, each checked against its source.
- The thread from rules, to examples, to generation.
It does not show:
- Every important person or machine. Many people worked on each idea; a timeline names only a few.
- The whole story outside the US and the UK. Computing and AI grew in many countries.
- Which events mattered most. Historians disagree, and so can you.
Talk about it at home
- For ages 8 to 10: "Babbage's engine was never finished. Can an idea matter even if the machine is not built?"
- For ages 11 to 13: "ELIZA followed written rules. The perceptron learned from examples. Which would be easier to check, and why?"
- For ages 14 to 18: "Pick one event. Who paid for it, who used it, and who was affected?"
Where does Mechna Academy fit?
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10. In each mission, a child guesses how a machine decides, then opens its Peek Hatch to check, changes one thing and tests it on a case it has never seen. Children learn how AI works this way, and they never chat with an AI.
Questions
The name "artificial intelligence" first appeared in a proposal dated August 31, 1955, for a summer study at Dartmouth College in 1956. Dartmouth calls that 1956 workshop the birth of the field. Earlier ideas, such as Alan Turing's 1950 imitation game, came before the name.
It depends on what you count. Charles Babbage designed the first programmable computer, the Analytical Engine, in the 1830s, but it was never finished. The first electronic computers, such as Colossus and ENIAC, were built in the 1940s.
No single person did. Charles Babbage designed a programmable machine, Ada Lovelace wrote steps for it, Alan Turing described the idea of a general computing machine, and teams in the 1940s built the first electronic ones.
One early example is ELIZA, described by Joseph Weizenbaum at MIT in 1966. It answered typed English using rules that a person wrote, triggered by key words. It did not learn from examples.
Frank Rosenblatt's perceptron, shown in 1958, learned from examples. In 1959 Arthur Samuel published a paper on a checkers program that used the phrase "machine learning".
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10.
Mechna Academy is a city of machines for curious children aged 8 to 18, opening first to ages 8 to 10.
For grown-ups: join the waitlistPlaces are limited. The waitlist is for parents and guardians 18 and older.