A Short History of Artificial Intelligence
A teen timeline of AI history: from Turing and the 1956 Dartmouth workshop through expert systems, AI winters, deep learning and today's large language models.
Key takeaways
- AI as a field was named at the 1956 Dartmouth workshop, but its roots go back to Turing in the 1950s
- Progress came in waves, with bursts of hype followed by 'AI winters' of lost funding and interest
- The modern boom is driven by big data, powerful GPUs and the deep learning breakthroughs of the 2010s
- Transformers (2017) led to the large language models that power today's chatbots
Older than you think
People imagine AI as a 21st-century invention. In truth, the dream of thinking machines is ancient, and the scientific field is over seventy years old. Knowing its history helps you see today's boom clearly: as the latest chapter, not a sudden miracle. For a quick refresher on what AI even means, see Meet Artificial Intelligence.
The foundations (1940s-1950s)
The story really starts with mathematician Alan Turing. In his 1950 paper, he asked, "Can machines think?" and proposed the Turing test: if a machine's typed answers are indistinguishable from a human's, perhaps it counts as intelligent. The idea still sparks debate today.
In 1956, a summer workshop at Dartmouth College gathered researchers including John McCarthy, Marvin Minsky and Claude Shannon. McCarthy coined the term "artificial intelligence", and the field was officially born. The mood was wildly optimistic; some predicted human-level machines within a generation.
Early excitement and the first winter (1960s-1970s)
Early programs were genuinely impressive for their time. They proved maths theorems, played checkers and held simple text conversations. ELIZA (1966) mimicked a therapist well enough to fool some users.
But the optimism ran ahead of reality. Computers were slow, data was scarce, and the hard problems, like understanding language or seeing the world, proved far tougher than expected. When results disappointed, funding dried up. This was the first AI winter.
Expert systems and a second winter (1980s)
AI bounced back with expert systems: programs that captured a specialist's knowledge as thousands of hand-written "if-then" rules. Companies used them for tasks like diagnosing equipment faults. For a while, AI was big business again.
Yet these systems were brittle. They could not learn, they broke on anything outside their rules, and they were expensive to maintain. By the late 1980s the hype collapsed once more, triggering a second AI winter.
The quiet build-up (1990s-2000s)
Away from the spotlight, machine learning matured. Instead of hand-coding rules, researchers let systems learn from data, the approach explained in What Is Machine Learning?. A milestone arrived in 1997, when IBM's Deep Blue beat world chess champion Garry Kasparov. The internet, meanwhile, was quietly creating something AI desperately needed: oceans of data.
The deep learning revolution (2010s)
Around 2012, several forces combined and AI exploded:
- Big data from the web gave models huge amounts to learn from.
- GPUs, chips built for video games, turned out to be perfect for training neural networks fast.
- Better algorithms made deep, multi-layer networks finally trainable.
A neural network called AlexNet crushed an image-recognition contest in 2012, and deep learning took off. Suddenly machines could recognise images, transcribe speech and translate languages at remarkable levels. To see how these networks work, read Neural Networks Explained.
The age of large models (2017-now)
In 2017, researchers introduced the transformer, an architecture brilliant at handling sequences like language. It made it possible to train enormous large language models on vast text collections.
This led directly to the generative AI you use today: chatbots, image generators and coding assistants. Progress since has been astonishingly fast, raising fresh questions about jobs, truth and safety.
What history teaches
The pattern is clear: bursts of hype, then harder reality. AI has been "about to change everything" before, and twice it stalled. Today's tools are genuinely powerful, but the past is a reminder to stay curious and sceptical, and to ask what these systems truly can and cannot do.
Quick quiz
Test yourself and earn XP
What year and event gave AI its name as a field?
At the 1956 Dartmouth Summer Research Project, John McCarthy and others coined the term 'artificial intelligence' and launched it as a field.
What did Alan Turing propose in 1950?
In his 1950 paper, Turing proposed the 'imitation game', now called the Turing test, to ask whether a machine's responses could be indistinguishable from a human's.
What is an 'AI winter'?
An AI winter is a stretch when promised breakthroughs failed to arrive, so funding and enthusiasm collapsed. There were two major ones.
What mainly powered the deep learning boom of the 2010s?
Huge labelled datasets, powerful graphics processors (GPUs), and better algorithms combined to make deep neural networks finally work well.
What 2017 breakthrough led to modern chatbots?
The transformer architecture, introduced in 2017, made it practical to train very large language models, leading to systems like GPT and today's chatbots.
FAQ
There is no single father, but key figures include Alan Turing, whose ideas laid the foundation, and John McCarthy, who coined the term 'artificial intelligence' in 1956. Many others shaped the field.
Yes, repeatedly. Bold predictions in the 1960s and 1980s outran reality and triggered 'AI winters'. That history is a good reason to stay both excited and sceptical today.
Keep exploring
More in AI