21 Facts About AI We Bet You Didn’t Know

Artificial intelligence is the field of computer science that builds systems able to do tasks we normally link with human thinking: recognising images and speech, understanding language, making predictions and learning from data. It feels new because of chatbots and image generators, but the field is more than 70 years old.

Here are 21 facts about AI, from its early history to recent milestones, that many people don't know.

A short history of AI

In 1950 Alan Turing published "Computing Machinery and Intelligence", which asked whether machines can think and proposed the imitation game, now known as the Turing test. The term "artificial intelligence" followed a few years later, and the 1956 Dartmouth workshop is usually treated as the birth of the field.

Early optimism was followed by two "AI winters", in the 1970s and again in the late 1980s, when funding dried up because results fell short of promises. Progress returned in the 1990s and 2000s with machine learning, more data and faster hardware. Since 2012, deep learning has driven most of the big advances, from speech recognition to the large language models behind today's AI assistants.

If you want to understand how the main terms relate, read our explainer on artificial intelligence vs machine learning.

21 facts about AI

1. The idea is ancient

In Homer's Iliad, the god Hephaestus is served by golden mechanical maidens who can speak and think. Greek myth also tells of Talos, a bronze giant built to guard Crete. People imagined thinking machines long before anyone could build one.

2. Artificial neurons date back to 1943

Warren McCulloch and Walter Pitts published a mathematical model of a neuron in 1943. It is the distant ancestor of the neural networks used today.

3. "Artificial intelligence" was coined in 1955

John McCarthy, then at Dartmouth College, used the term in a 1955 proposal, written with Marvin Minsky, Nathaniel Rochester and Claude Shannon, for a summer research workshop held at Dartmouth in 1956.

4. One of the first AI programs played draughts

Christopher Strachey, a British schoolteacher and early programmer, wrote a draughts (checkers) program that ran on the Ferranti Mark 1 computer in 1952.

5. "Machine learning" is from 1959

Arthur Samuel at IBM popularised the term while building a checkers program that improved by playing games against itself.

6. The perceptron arrived in 1958

Frank Rosenblatt's perceptron was an early machine that learned to classify simple images. It was overhyped at the time, but its core idea lives on in modern neural networks.

7. People trusted a chatbot in the 1960s

Joseph Weizenbaum's ELIZA, built at MIT in the mid-1960s, mimicked a therapist using simple pattern matching. Some users still believed it understood them. The tendency to read understanding into a machine's replies is now called the ELIZA effect.

8. Hard things are easy and easy things are hard

Moravec's paradox, described by roboticist Hans Moravec in the 1980s, observes that computers find logic, maths and chess relatively easy but struggle with things a toddler does without effort, such as walking, grasping objects and recognising faces.

9. Expert systems hit a knowledge bottleneck

In the 1970s and 1980s, expert systems tried to encode human expertise as rules. Getting that knowledge out of experts' heads and into the system was slow and expensive. This "knowledge acquisition bottleneck" is one reason machine learning, which learns from data instead, took over.

10. Deep Blue lost its first match

IBM's Deep Blue won a game against world champion Garry Kasparov on 10 February 1996, the first time a computer had beaten a reigning champion under tournament conditions. Kasparov still won that match. Deep Blue won the rematch in 1997.

11. Watson won Jeopardy! in 2011

IBM's Watson beat two of the show's best champions, which showed that computers could answer questions asked in everyday language.

12. A 2012 image contest started the deep learning boom

A neural network called AlexNet won the ImageNet image recognition challenge by a wide margin. Its use of GPUs to train a deep network kicked off the deep learning era.

13. AlphaGo beat Lee Sedol in 2016

DeepMind's AlphaGo beat one of the world's best Go players 4–1. Go has so many possible positions that experts had expected this milestone to be at least a decade away.

14. The "T" in GPT comes from a 2017 paper

Google researchers introduced the transformer architecture in the paper "Attention Is All You Need". Transformers now power most large language models, including the GPT family, Claude and Gemini.

15. ChatGPT launched in November 2022

OpenAI released ChatGPT on 30 November 2022. It brought generative AI to a mass audience almost overnight and pushed businesses everywhere to work out what AI could do for them.

16. AI work won two Nobel Prizes in 2024

John Hopfield and Geoffrey Hinton won the Nobel Prize in Physics for foundational work on neural networks. Demis Hassabis and John Jumper of Google DeepMind shared the Nobel Prize in Chemistry, with David Baker, for AlphaFold's protein structure prediction and for computational protein design.

17. Stephen Hawking was unsure which way it would go

In 2016 Hawking said powerful AI would be either the best or the worst thing ever to happen to humanity, and that we did not yet know which. That debate about benefits and risks still shapes AI policy.

18. The EU passed the first broad AI law

The EU AI Act entered into force on 1 August 2024. It sorts AI systems by risk and phases in obligations over several years, with the strictest rules for high-risk uses.

19. AI can invent new sounds

Google's Magenta team released NSynth in 2017. It uses a neural network to blend the sounds of different instruments into new ones that musicians can play.

20. AI helps count wildlife

Wildbook uses computer vision to identify individual animals from their stripes and spots. In Kenya it has been used in the Great Grevy's Rally, where volunteers photograph endangered Grevy's zebras and the software identifies each animal to estimate the population.

21. Language models can sound confident and still be wrong

Large language models predict likely text. They don't look facts up unless they are connected to a search tool or a company's own data. That is why they sometimes "hallucinate" plausible but false answers, and why business AI systems are usually grounded in trusted data and checked by people.

What this means for businesses

AI has moved from research labs into everyday tools: customer support assistants, document processing, sales forecasting, fraud detection and coding help. Our post on applications of artificial intelligence covers common uses, and how to build AI apps walks through the process.

The practical lesson from 70 years of AI history is to start with a specific problem, measure the results and keep people in the loop where mistakes are costly.

Conclusion

AI has a long history of big promises, setbacks and breakthroughs. The current wave is different because the tools are good enough, cheap enough and easy enough to use that most businesses can put them to work.

If you want to use AI in your business, Sterling Technolabs can help you find the right use case through AI consulting and then build it with our AI software development and generative AI development teams.