When was AI invented?
AI was invented as a field in two steps: a proposal dated 31 August 1955 and a workshop in the summer of 1956. Here are the dates every page agrees on, each with the document behind it.

| Year | What happened | Who | Source |
|---|---|---|---|
| 1950 | Asks "Can machines think?" and proposes the imitation game | Alan Turing | Mind, October 1950 |
| 31 Aug 1955 | Proposes a "2 month, 10 man study of artificial intelligence" | McCarthy, Minsky, Rochester, Shannon | Dartmouth proposal |
| Summer 1956 | The Dartmouth workshop is held | Same four organizers | Dartmouth proposal, Wikipedia |
| 1974 to 1980 | First AI winter: the US and UK governments stop funding undirected AI research in 1974 | Governments, after the Lighthill report | Wikipedia |
| 1980 to 1987 | Expert systems bring a boom, then the Lisp machine market collapses in 1987 | Industry and Japan's Fifth Generation project | Wikipedia |
| May 1997 | Deep Blue beats Kasparov 3.5 to 2.5 (Wikipedia gives 11 May) | IBM | IBM history page, Wikipedia |
| 2012 | AlexNet wins the ImageNet challenge, a deep learning breakthrough | Deep learning researchers | Wikipedia |
| March 2016 | AlphaGo beats Lee Sedol 4 to 1 | DeepMind | DeepMind |
| 12 Jun 2017 | The Transformer paper is posted | Vaswani and seven coauthors | arXiv |
| 30 Nov 2022 | ChatGPT is announced | OpenAI | OpenAI |
So the answer depends on what you count. The idea of a thinking machine is older than 1950. The name and the research field date from 1955 to 1956. Chat assistants most people use came in 2022.
Who invented AI?
No single person invented AI, but four people put the name and the plan on paper in 1955. The Dartmouth proposal lists J. McCarthy of Dartmouth College, M. L. Minsky of Harvard University, N. Rochester of I.B.M. Corporation and C.E. Shannon of Bell Telephone Laboratories.
McCarthy is the person usually credited with the term, and Wikipedia says he coined it at the workshop. The proposal itself, which is where the phrase appears in its title, is dated a year earlier. That is a small correction to a claim repeated across the web, including on the thats-ai.org history page, which says the term was first used by McCarthy at the 1956 Dartmouth project, and on Wikipedia, which says he introduced it at the 1956 workshop.
Turing is the other name. His 1950 paper, "Computing Machinery and Intelligence," opens with the sentence "I propose to consider the question, 'Can machines think?'" and calls his test the imitation game. The phrase artificial intelligence belongs to the 1955 proposal.
Who is the father of AI?
Sources split between two people. Some give the title to Turing for the 1950 paper, and others give it to McCarthy for the name and the 1956 workshop. Neither title is official, and both men can be defended.
A fair way to say it: Turing posed the question and McCarthy, Minsky, Rochester and Shannon founded the field. Minsky co-organized Dartmouth with McCarthy, and Shannon, per the thats-ai.org page, described the first chess playing program in 1950. Nobody awarded the title, so treat any page that sounds certain with suspicion.
What did the Dartmouth proposal actually say?
The proposal asked for a two month study and rested on one conjecture. It says: "We propose that a 2 month, 10 man study of artificial intelligence be carried out during the summer of 1956 at Dartmouth College in Hanover, New Hampshire."
The conjecture is the part worth reading twice. The study was to proceed on the basis of the conjecture that "every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it." That sentence is the founding bet of the field.
The proposal also asked how to make a machine "use language, form abstractions and concepts", and listed neuron nets and randomness and creativity among its topics. Wikipedia adds that the workshop was considered a failure at the time. Ten people for two months didn't solve intelligence.
AI timeline: from Turing in 1950 to ChatGPT in 2022
The chart shows the shape: a burst of ideas in the 1950s, a stall in the 1970s, and then a run of public milestones. Sources give only a year for the perceptron paper, the Lighthill report and the 1974 funding cut, and only a month for Deep Blue, the Dartmouth workshop and AlphaGo, so those marks sit at the start of the period and are not exact days. These are the dated events behind it, from Wikipedia unless noted.
- 1943: McCulloch and Pitts publish "A Logical Calculus of the Ideas Immanent in Nervous Activity," an early model of artificial neurons.
- 1956: Newell and Simon complete the Logic Theorist and present it at the Dartmouth workshop.
- 1957 and 1958: Rosenblatt introduces the perceptron in a 1957 lab report and describes it in a 1958 paper.
- 1966: The ALPAC report criticizes machine translation efforts.
- 1969: Minsky and Papert publish Perceptrons.
- 1973: The Lighthill report on UK AI research appears.
- 1974: The US and UK governments stop funding undirected AI research.
- 1980: Searle presents the Chinese Room argument.
- 1980 to 1987: Expert systems bring a boom of investment.
- Late 1980s and 1990s: A second AI winter follows.
- 2012: AlexNet wins the ImageNet challenge.
- 12 June 2017: Eight authors post the Transformer paper, "Attention Is All You Need," on arXiv.
- 2024: The Nobel Prize in Physics goes to John Hopfield and Geoffrey Hinton, and the Chemistry prize to David Baker, Demis Hassabis and John Jumper.
The 2017 paper matters because it proposed "a new simple network architecture, the Transformer, based solely on attention mechanisms." GPT's own name stands for generative pre-trained transformer. What is an LLM explains how that design became a chat model.
What were the AI winters and expert systems?
An AI winter is a stretch when funding and interest drop after hype fails to deliver. Wikipedia dates the first one from 1974 to 1980. Criticism from James Lighthill and pressure from the US Congress led the US and British governments to stop funding undirected AI research in 1974. Earlier, after the 1966 ALPAC report on machine translation, the US National Research Council ended its support after spending 20 million dollars.
Expert systems ended it. These programs encoded human expertise as rules, and Wikipedia says a system called R1 was completed at Carnegie Mellon in 1980 for Digital Equipment Corporation. In 1981 Japan's Ministry of International Trade and Industry set aside 850 million dollars for the Fifth Generation computer project, and by the late 1980s AI had become a billion dollar industry. The thats-ai.org page adds that the Fifth Generation effort largely failed.
The second winter followed. The market for specialized AI hardware, the Lisp machines, collapsed in 1987 when desktop computers from Apple and IBM outperformed them, and Wikipedia says that by the end of 1993 more than 300 AI companies had shut down, gone bankrupt or been acquired. The field then revived through machine learning and, from 2012, deep learning.
What do three real moments show about AI history?
Three cases show how the field works better than a date list does.
Case 1, the proposal that promised too much. In 1955 four researchers asked for a two month study and a conjecture about simulating intelligence. In 1974 the US and UK governments stopped funding undirected research. The stretch that followed is the first AI winter, covered above.
Case 2, a game as a test. IBM says Deep Blue beat Garry Kasparov 3.5 to 2.5 in May 1997 in New York and became the first computer system to defeat a reigning world champion under standard tournament time controls. In March 2016 AlphaGo beat Lee Sedol, an 18 time world champion, 4 to 1, and DeepMind says more than 200 million people watched. Both wins were public and easy to score, which is why they became the milestones people remember. A system built to win one game is a different thing from a general one, a split AGI vs AI covers.
Case 3, a chat box. OpenAI announced ChatGPT on 30 November 2022 with the line "We've trained a model called ChatGPT which interacts in a conversational way." The same post warns that "ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers." That limit is as old as the idea of a talking machine, and Turing's test never claimed to measure truth.
Was AI invented before 1950?
Yes, if you count the ideas. The thats-ai.org history page lists mythic and mechanical precedents, such as Talos in Greek myth and a musical robot built by Ismail al-Jazari in 1206. It says Charles Babbage designed the Analytical Engine in 1837, with Ada Lovelace publishing the first computer algorithm for it.
Those aren't AI research. They are earlier answers to the question of whether a thing can be made to act like a mind. Wikipedia's page covers the same precursors, then dates the birth of the field to 1941 through 1956, which includes the 1943 neuron network paper of Pitts and McCulloch. Artificial minds are an old idea, and AI research is a mid twentieth century one.
Which dates do pages get wrong?
Four mistakes recur, and each has a fix.
- Saying the term appeared in 1956. The proposal that used it is dated 31 August 1955. The fix: say proposed in 1955, held in 1956.
- Crediting Turing with the term. The name comes from the 1955 proposal, five years after his paper. The fix: credit him with the question and the imitation game.
- Calling ChatGPT the invention of AI. It was announced 30 November 2022, 67 years after the proposal. The fix: call it the moment chat AI reached the public.
- Trusting a model's date with no source. OpenAI's own announcement warns about plausible but incorrect answers. The fix: open the primary document, as in the steps above.
A fifth is subtler: treating the Turing test as proof a machine thinks. Turing called it the imitation game, and a game result is a different claim.
What questions should you ask an AI about the history of AI?
Paste these into Claude, ChatGPT, Gemini or Whizi. Each asks for a source so you can check the answer yourself.
- Dates: "When was the term artificial intelligence first written down, and in what document? Give the exact date and a link to the primary text, and say if the term is often credited to a different date."
- People: "List the four authors of the 1955 Dartmouth proposal with their institutions at the time, and say what each is best known for. Use only sources you can link."
- Turing: "Summarize Turing's 1950 paper in six sentences. Quote one line exactly, give the journal, volume and page range, and say what the imitation game is."
- Winters: "Explain why funding for AI fell in the 1970s and again in the late 1980s. For each, name the report or event and its date, and mark any claim you can't source."
- Disagreement: "Where do historians disagree about the birth of AI? Give two views, the evidence for each, and a source I can open for both."
- Compare: "Compare the Dartmouth proposal of 1955 with how AI is described today. List three claims that came true, three that did not, and a source for each."
One app for the models this history led to
Whizi puts Claude, GPT and Gemini, the chat models that descend from the 2017 Transformer design, in one app. Pro carries all three, and its first week is $1.99. The free plan gives 10 messages a day on Whizi AI, Auto and Claude Haiku, with no card.
Web search is on by default in a new chat, so you can ask for dated sources, and side by side Compare on Pro puts one prompt to two models so you can read where their dates differ. For a quick history question, Wikipedia or the Stanford proposal is faster than any chat app, and Whizi doesn't beat a primary document. See Whizi pricing.
Where does AI history go wrong?
Sources disagree on dates and on credit. The thats-ai.org page, for example, has no section on anything after AlphaGo in 2016, so it can't answer the question of when chat AI began. Wikipedia carries the 2017 to 2026 period in a section of its own, but its summary is a secondary source.
Even the founding story is contested. Wikipedia says Dartmouth was considered a failure at the time, yet it is cited as the birth of the field. Read the proposal, then read the dates in a few places, and trust the pattern, not a single page. For the future, AGI vs AI separates what is claimed from what exists, and what is AI defines the terms.
- Say proposed in 1955 and held in 1956, not just 1956
- Credit Turing with the 1950 question, not the term
- Open the primary document before quoting a date
- Ask for a link behind every date an AI gives you
- Ask a second model when two sources disagree
Frequently asked questions
Who invented AI?
No one person did, but John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon wrote the 1955 Dartmouth proposal that named the field. McCarthy is usually credited with the term. Alan Turing posed the founding question in 1950 and is the other name people give.
Who is the father of AI?
Sources name either Alan Turing or John McCarthy, and neither title is official. Turing asked whether machines can think in his 1950 paper. McCarthy led the Dartmouth proposal in 1955 and workshop in 1956. A fair summary credits Turing with the question and McCarthy with the name.
When was AI invented for the public?
Chat AI reached the public on 30 November 2022, when OpenAI announced ChatGPT. Research AI is older: Deep Blue beat Kasparov in May 1997, and AlphaGo beat Lee Sedol in March 2016. The field itself dates from the 1955 Dartmouth proposal.
Was the Turing test the start of AI?
No, it came earlier. Turing published it in Mind in October 1950, five years before the Dartmouth proposal. He called it the imitation game, and the proposal that named the field is dated 1955.