The history of AI moves from hand-written rules to systems that learn statistical patterns from data. Major turning points include Turing’s 1950 imitation game, the naming of AI at Dartmouth in 1956, expert systems, specialized AI victories such as Deep Blue, GPU-powered deep learning, the 2017 Transformer, and today’s generative, multimodal, and agentic systems.
Key milestones in AI history
Turing reframes the question
Alan Turing publishes “Computing Machinery and Intelligence” and proposes the imitation game as an operational way to discuss machine intelligence.
Artificial intelligence gets its name
The Dartmouth Summer Research Project brings researchers together around the proposal that aspects of learning and intelligence might be described precisely enough for a machine to simulate.
The perceptron learns from examples
Frank Rosenblatt’s perceptron demonstrates an early trainable model: input features are combined with adjustable weights to produce a decision.
Rules and expert systems expand
Researchers encode specialist knowledge as rules. These systems can be useful in narrow settings, but are expensive to maintain and struggle outside their prepared domain.
Deep Blue defeats Kasparov
IBM’s chess computer defeats the reigning world champion in a match, showing the strength of specialized algorithms and large-scale parallel search — not general human-like intelligence.
GPUs accelerate deep learning
AlexNet uses a deep convolutional neural network and GPUs to achieve a major ImageNet result. Data, parallel hardware, and improved training methods combine to move computer vision forward.
The Transformer changes sequence modeling
“Attention Is All You Need” introduces a Transformer architecture built around attention, allowing relationships across a sequence to be modeled with high parallelism during training.
Generative AI reaches a mass audience
Conversational interfaces make large generative models easy to use. Models increasingly connect text, images, audio, video, tools, and actions.
Why were there “AI winters”?
AI progress has never been a straight line. At several points, expectations outran computing power, available data, or the reliability of existing methods. Funding and enthusiasm fell. These slowdowns are called AI winters. They are a useful warning: impressive demonstrations do not automatically become robust, general-purpose systems.
What changed in modern AI?
Modern advances came from a combination, not a single invention: larger datasets, faster specialized processors, improved algorithms, large-scale training, and interfaces that made the systems widely accessible. Capabilities expanded, but core limitations remain. A model can produce a convincing answer without a reliable way to know whether that answer is true.
Common questions
Who invented artificial intelligence?
No single person invented AI. Alan Turing helped frame machine intelligence; John McCarthy coined the term “artificial intelligence” for the 1956 Dartmouth project; and many researchers developed the field’s theories, hardware, datasets, and applications.
When was artificial intelligence invented?
There is no single invention date. 1956 is often treated as the birth of AI as a named research field, while important foundations appeared earlier.
Was Deep Blue machine learning?
Deep Blue was primarily a specialized chess system built around massive search, evaluation functions, databases, and expert input. Its achievement should not be confused with today’s general-purpose generative models.
Primary and institutional sources
Research papers explain mechanisms; institutional resources provide guidance. Manufacturer pages describe their own products and are not independent evaluations.
- Computing Machinery and Intelligence (1950) — Alan M. Turing · Mind
- Frank Rosenblatt and the perceptron — Cornell University
- Deep Blue: search, evaluation and expert input — IBM
- ImageNet Classification with Deep Convolutional Neural Networks (2012) — Krizhevsky, Sutskever & Hinton · NeurIPS
- Attention Is All You Need (2017) — Vaswani et al.
- Introducing ChatGPT (30 November 2022) — OpenAI
- Le projet de Dartmouth de 1956 — Dartmouth College
Continue with the companion in ChatGPT
Explore museum themes, ask questions and test your understanding with quizzes in ChatGPT. The companion and the game keep separate progress.
Discover the ChatGPT companionContent reviewed: 6 September 2026. Navigation and resources updated: 1 October 2026. Project method and limits.