History of Artificial Intelligence Timeline, 1943 to 2023

Artificial intelligence timeline at a glance

The table runs through the main milestones in the history of AI, using the corrected dates. The Genuine Impact infographic on this page covers 1943 to 2023; the later entries are added from the record.

The birth of AI, 1943 to 1956

The idea came before the computers did. In 1943 McCulloch and Pitts showed that simple on and off units modelled on brain cells could, in principle, carry out logic, which is the root of every neural network since [1]. In 1950 Turing sidestepped the question “can machines think?” by proposing a test: if a person holding a typed conversation can’t tell the machine from a human, the machine passes [2]. A year later Minsky and Edmonds built SNARC, a randomly connected network of about 40 artificial synapses that learned to solve a maze [3].

The field got its name in 1955, when John McCarthy, then a young mathematics professor at Dartmouth College, wrote a funding proposal with Minsky, Nathaniel Rochester and Claude Shannon for a summer research project. The proposal is credited with introducing the term “artificial intelligence”, and the 1956 workshop that followed is treated as the founding event of AI as a field [4].

Early optimism and the first AI winter

The first two decades produced programs that seemed remarkable for their time. The Logic Theorist proved mathematical theorems, the General Problem Solver (1957) tried to solve any problem that could be written as a set of rules, and Rosenblatt’s perceptron learned to recognise simple patterns [5]. ELIZA, written by Joseph Weizenbaum at MIT between 1964 and 1967, imitated a therapist by matching patterns in what people typed. Weizenbaum was surprised that some people, including his secretary, treated it as if it had human feelings [6].

Researchers promised more than the computers of the day could deliver. Machine translation failed to live up to expectations, the 1969 book Perceptrons exposed the limits of single-layer networks, and in 1973 the Lighthill report to the UK government criticised AI for failing to reach its “grandiose objectives” [7]. Funding was cut in Britain and by DARPA in the US, starting the first AI winter, which lasted roughly from 1974 to 1980 [7].

Expert systems and the second AI winter

AI came back in the 1980s in a more practical form. Expert systems captured the rules a specialist uses and applied them automatically. The first commercial one, XCON, configured computer orders for Digital Equipment Corporation and was estimated to save the company $40 million over six years. By 1985 companies were spending over a billion dollars on AI [7]. At the same time, in 1986, Rumelhart, Hinton and Williams showed how backpropagation could train networks with several layers, which revived interest in neural networks [8].

The boom didn’t last. Expert systems were expensive to maintain and broke when faced with situations outside their rules, and in 1987 the market for specialist AI computers collapsed. The second AI winter ran from 1987 into the 1990s [7].

Machine learning and deep learning, 1997 to 2017

From the late 1990s, faster computers and much more data shifted AI from hand-written rules to systems that learn from examples. The public milestones were mostly games. Deep Blue beat Kasparov in their 1997 rematch in New York, the first time a reigning world chess champion lost a match to a computer under tournament conditions; Kasparov had won their first match in 1996 [9]. IBM Watson won Jeopardy! in 2011.

2012 is usually seen as the start of the deep learning era. AlexNet, a neural network trained on graphics chips, won the ImageNet image recognition contest by a wide margin, and over the next few years most other approaches gave way to deep learning [10]. In the same year a Google Brain network trained on 10 million YouTube images learned to recognise cats without being told what a cat was [11]. The infographic credits that cat network to DeepMind, but it was Google Brain’s work.

DeepMind, bought by Google in 2014, made its name with AlphaGo, which beat Lee Sedol, one of the world’s best Go players, 4 to 1 in March 2016 [13]. The infographic dates this to 2015, but the match was in 2016. In 2017 AlphaZero taught itself chess, shogi and Go from the rules alone and beat the strongest existing programs [14]. The most important development of 2017 was less visible: Google researchers published “Attention Is All You Need”, the paper that introduced the transformer, the design that large language models are built on [10].

Generative AI, 2020 onwards

OpenAI’s GPT-3 (2020) showed that a very large transformer trained on text could write fluent prose. The same year DeepMind’s AlphaFold 2 predicted protein structures at the CASP14 competition in November 2020 with far higher accuracy than any other entry [15]. The infographic puts this in 2021, the year the method was published in Nature and the database of predictions released [15].

The big change for the public was ChatGPT, released on 30 November 2022. It reached 100 million monthly users in two months, the fastest growth of any consumer app at the time, and had 900 million weekly users by February 2026 [16]. The infographic, made in 2023, leaves it out. GPT-4 followed in March 2023, and the same period brought the first big legal fights, including the January 2023 class action by artists Sarah Andersen, Kelly McKernan and Karla Ortiz against Stability AI, Midjourney and DeviantArt [10].

In 2024 the science behind modern AI was recognised with two Nobel prizes: physics for Hinton and Hopfield’s neural network research, and chemistry for Hassabis and Jumper’s work on AlphaFold [8][10]. The EU AI Act, the first broad law regulating AI, came into force on 1 August 2024, with its rules phased in over the following years [17].

Questions about the history of AI

When was AI first invented?

AI became a field of study at the Dartmouth workshop in the summer of 1956, and the term “artificial intelligence” dates from the 1955 proposal for it [4]. The ideas go back further, to McCulloch and Pitts’s neuron model in 1943 and Turing’s 1950 paper [1][2].

Who is the father of artificial intelligence?

John McCarthy is most often given the title, because he named the field and organised the Dartmouth workshop. Marvin Minsky, Nathaniel Rochester and Claude Shannon, his co-organisers, and Alan Turing are also called founders [4].

What was the first AI program?

The Logic Theorist, written by Allen Newell, Cliff Shaw and Herbert Simon and first demonstrated in 1956, has been described as the first AI program. It proved 38 of the first 52 theorems in chapter two of Principia Mathematica [18].

What is an AI winter?

A period when funding and interest in AI fall sharply after promises are not met. There were two major ones, from about 1974 to 1980 and from 1987 into the 1990s [7].

When did AI become popular?

AI research boomed after 2012 with deep learning, but public use took off with ChatGPT in late 2022. Google searches for “AI” began to climb steeply in 2022 [10][16].

Sources

1. Wikipedia, Warren Sturgis McCulloch.
2. Wikipedia, Computing Machinery and Intelligence.
3. Wikipedia, Stochastic Neural Analog Reinforcement Calculator.
4. Wikipedia, Dartmouth workshop.
5. Wikipedia, General Problem Solver.
6. Wikipedia, ELIZA.
7. Wikipedia, AI winter.
8. Wikipedia, History of artificial intelligence.
9. Wikipedia, Deep Blue versus Garry Kasparov.
10. Wikipedia, Timeline of artificial intelligence.
11. Wikipedia, Google Brain.
12. Wikipedia, Google DeepMind.
13. Wikipedia, AlphaGo versus Lee Sedol.
14. Wikipedia, AlphaZero.
15. Wikipedia, AlphaFold.
16. Wikipedia, ChatGPT.
17. Wikipedia, Artificial Intelligence Act.
18. Wikipedia, Logic Theorist.
19. Genuine Impact, genuineimpact.substack.com, the creator of the infographic on this page, which lists ChatGPT and Futurism as its sources.

Last reviewed 6 November 2026