How Long Has AI Been Around?

AI has been a field of research for about 70 years. The term "artificial intelligence" first appeared in a 1955 proposal for a summer workshop at Dartmouth College, held in 1956 and widely considered the founding event of the field. The underlying ideas are older: Alan Turing asked whether machines could think in 1950. AI became part of most people's daily lives much more recently, especially after ChatGPT launched in November 2022.

When did AI start?

AI started as a formal field in the summer of 1956, at the Dartmouth Summer Research Project on Artificial Intelligence in Hanover, New Hampshire.

The workshop grew out of a proposal dated August 31, 1955, written by John McCarthy of Dartmouth College, Marvin Minsky of Harvard University, Nathaniel Rochester of IBM and Claude Shannon of Bell Telephone Laboratories. It opened: "We propose that a 2 month, 10 man study of artificial intelligence be carried out during the summer of 1956 at Dartmouth College." That document is the first known use of the phrase "artificial intelligence." Dartmouth describes the 1956 gathering as "the birth of this field of research."

Some ideas came before the name:

• 1943: Warren McCulloch and Walter Pitts published a mathematical model of how neurons might work, an early foundation for neural networks.
• 1950: Alan Turing published "Computing Machinery and Intelligence," which opens with the question "Can machines think?" He proposed a test he called the "imitation game," now known as the Turing test.

Why was AI created?

AI was created to find out whether machines could do things that seemed to require human intelligence, and to build machines that could.

The Dartmouth proposal was based on what its authors called a 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." Its authors hoped to make machines "use language, form abstractions and concepts, solve kinds of problems now reserved for humans, and improve themselves."

That goal combined scientific curiosity about how thinking works with a practical ambition: automating complex mental work such as solving problems, proving theorems and understanding language.

A timeline of AI

The early years: big ambitions in the 1950s and 1960s

Most early work focused on symbolic AI: programming computers with logic and rules. The Logic Theorist proved mathematical theorems. ELIZA imitated a psychotherapist and convinced some users they were talking to a person, even though it simply matched patterns in their words. Work on Shakey, the first mobile robot to reason about its own actions, produced the A* search algorithm that is still used for tasks like finding driving directions.

A second approach also began in this period. In 1958, Frank Rosenblatt's perceptron, a machine that learned from examples, was unveiled with considerable press attention. This learning-based approach is the ancestor of today's neural networks.

The AI winters: when progress stalled

Early promises outran what computers of the time could deliver, and enthusiasm collapsed twice.

The first AI winter. In 1973, a report by Sir James Lighthill, commissioned by the UK's Science Research Council, was sharply critical of AI research. According to an archive of the report's history, it "provoked a massive loss of confidence in AI" in UK academia that "persisted for almost a decade." Neural network research had also lost momentum after Marvin Minsky and Seymour Papert's 1969 book Perceptrons highlighted the limits of simple networks.

The second AI winter. Interest and investment in AI revived in the 1980s. At a 1984 AI conference, researchers Roger Schank and Marvin Minsky warned that inflated expectations would lead to another collapse. According to IBM's history of AI, the prediction came true within about three years, and the period of reduced funding and slower progress became known as the second AI winter.

The shift to machine learning

While headline AI struggled, a quieter change was underway: moving from hand-written rules to systems that learn from data.

In 1986, David Rumelhart, Geoffrey Hinton and Ronald Williams published a paper in Nature describing backpropagation, a method for training multi-layer neural networks. By the late 1990s, neural networks were reading several million bank checks per day. Separately, in 1997, IBM's Deep Blue, which relied on brute-force computing power to evaluate 200 million chess positions per second, beat world chess champion Garry Kasparov 3.5 to 2.5, the first time a computer defeated a reigning world champion in a match under standard tournament controls.

For how this learning-based approach works, see What Is Machine Learning?.

The deep learning breakthrough

In the 2010s, three things came together: huge datasets, much faster computer chips and better training methods. Neural networks with many layers, known as deep learning, suddenly outperformed older approaches.

The turning point came in 2012. A University of Toronto team's deep neural network won the ImageNet image recognition challenge with a top-5 error rate of 15.3%, far ahead of the next-best team's 26.2%.

In March 2016, DeepMind's AlphaGo beat Go champion Lee Sedol 4 to 1 in Seoul. In 2017, Google researchers published "Attention Is All You Need," introducing the Transformer, the architecture behind today's large language models. Three pioneers of deep learning, Yoshua Bengio, Geoffrey Hinton and Yann LeCun, received the 2018 ACM Turing Award.

The generative AI era

The 2020s brought AI that creates text, images and code. OpenAI's GPT-3, described in 2020, had 175 billion parameters and could produce convincing text from simple prompts.

The moment AI went mainstream came on November 30, 2022, when OpenAI released ChatGPT, a free chatbot anyone could use. OpenAI reported more than 900 million weekly ChatGPT users by February 2026. Scientific recognition followed: in 2024, John Hopfield and Geoffrey Hinton shared the Nobel Prize in Physics for foundational work enabling machine learning with artificial neural networks, and Demis Hassabis and John Jumper shared half of the Nobel Prize in Chemistry for protein structure prediction.

By 2025, major AI companies were releasing AI agents, systems that can carry out multi-step tasks on a computer rather than only answer questions. Anthropic released an early "computer use" capability in October 2024, and OpenAI introduced ChatGPT agent in July 2025. For how the models behind this era work, see What Is a Large Language Model?.

When did AI become popular?

AI became widely popular with the general public after ChatGPT's release in November 2022. Before that, AI was already in everyday products, including spam filters, recommendations and voice assistants, but most people did not think of those as "using AI."

Adoption since then has been fast. A Pew Research Center survey conducted in February 2026 found that about half of U.S. adults use AI chatbots. Stanford's 2026 AI Index reported that 88% of organizations in the surveys it cites were using AI in 2025.

Who created AI?

No single person created AI. It was built by many researchers over decades.

• John McCarthy led the 1955 proposal that named the field and organized the 1956 Dartmouth workshop.
• Alan Turing framed the central question of machine intelligence in 1950.
• Marvin Minsky, Nathaniel Rochester and Claude Shannon co-authored the Dartmouth proposal.
• Allen Newell, Herbert Simon and J. Clifford Shaw created the Logic Theorist.
• Frank Rosenblatt built the perceptron, an early learning machine.
• Geoffrey Hinton, Yann LeCun and Yoshua Bengio led the development of deep learning.

Has AI been invented yet?

Yes, in the sense that matters for everyday use. AI systems that recognize speech, translate languages, recommend content and generate text and images are real and widely used. Each of these systems is designed for specific kinds of tasks.

Artificial general intelligence, or AGI, is a different matter: AI that equals or surpasses human performance on all or almost all cognitive tasks. The International AI Safety Report 2025, prepared by an international panel of experts, describes AGI as a "potential future AI" and notes that the term "lacks a universal definition." Its 2026 edition describes current AI as strong at some difficult tasks while still failing at some simpler ones.