📜 The History of AI
5th Grade 🤖 Artificial Intelligence
The idea of a thinking machine is very old, but the science of artificial intelligence began in the middle of the 20th century. Since then it has gone through times of huge excitement, times of disappointment, and breakthroughs that surprised everyone.
Before computers
Ancient myths already told of statues and artificial figures that came to life — the Greeks, for example, told of Talos, a giant bronze man who guarded an island. In the 1800s the English mathematician Ada Lovelace worked on programs for a calculating machine designed by Charles Babbage, and many people consider her one of the first programmers. She wrote that a machine like that could work with things other than numbers, like musical sounds — but that it would only do what it was told to do.
Turing and Dartmouth: the birth of a field
In 1950 the British mathematician Alan Turing wrote a paper that asked: “Can machines think?” Instead of arguing about the definition of thinking, he suggested a game: a person exchanges written messages with two hidden partners — one human and one computer — and tries to guess which one is the computer. If it’s hard to tell, maybe we can say the computer is behaving intelligently. Today that idea is called the “Turing test.”
In 1956 a group of scientists met at a summer workshop at Dartmouth College in New Hampshire. In the proposal for that workshop, written back in 1955, the term “artificial intelligence” appeared for the first time, coined by the scientist John McCarthy. The researchers were very optimistic and thought that within a few years they would build machines as smart as people. It turned out to be much harder than that.
Rules, excitement and winters
In the first decades, most researchers tried to teach computers using rules they wrote by hand. In the 1960s a simple chat program called ELIZA was built that repeated the user’s words back as a question. It didn’t understand a thing, but many people felt it was “listening” to them — an important lesson that is still true today: it’s easy for us to imagine that a talking machine understands.
Later it became clear that writing rules for every situation in the world is nearly impossible. Funding dropped and the excitement faded. Periods like these are called “AI winters,” and there were several of them. Meanwhile, other researchers kept developing a different idea: machines that learn from examples instead of rules.
Chess, Go and deep learning
In 1997 IBM’s computer Deep Blue beat world chess champion Garry Kasparov in a match. Deep Blue checked an enormous number of possible moves every second — that was computing power, not learning like we know it today.
Later, as computers got faster and huge amounts of pictures and text piled up online, big neural networks started succeeding at tasks that had been very hard, like recognizing pictures and speech. This approach is called “deep learning.” In 2016 the program AlphaGo beat Lee Sedol, one of the best Go players in the world. Go is a board game with so many possibilities that it’s impossible to check them all, and the program also learned from huge numbers of games.
The age of chatbots
In late 2022 an AI chatbot that can write text and answer questions on all kinds of topics was released to the public. In a very short time, huge numbers of people around the world started using tools like this, and schools and families started talking about how to use them wisely and fairly.
The field keeps changing very fast, so it’s important to learn the basic ideas that don’t change: AI learns patterns from data, it can make mistakes, and people are responsible for how it’s used.
When you look at the whole story from above, you see an interesting pattern: again and again people thought the smart machine was just around the corner, and again and again it turned out there was a lot more to learn. On the other hand, there were also happy surprises — tasks that seemed impossible, like winning at Go, were reached earlier than expected. So it’s smart to be careful about both big promises and big fears, and to check the facts.
Key facts
- In 1950 Alan Turing proposed the game now called the “Turing test.”
- The term “artificial intelligence” first appeared in connection with the 1956 Dartmouth workshop.
- In 1997 Deep Blue beat world chess champion Garry Kasparov.
- In 2016 AlphaGo beat Lee Sedol at the game of Go.
- In late 2022 an AI chatbot was released to the public, and chatbots quickly spread around the world.
Quick quiz
Who proposed a test in 1950 to check whether a machine behaves intelligently?
At which event did the term “artificial intelligence” first appear?
What happened in 1997?
What were the “AI winters”?
What did people learn from the chat program ELIZA?
Which program beat Lee Sedol at Go in 2016?
How did Deep Blue play chess?
When were AI chatbots released to the general public and quickly spread around the world?
💭 Think and talk about it
- Why do you think the researchers in 1956 believed they’d build a smart machine within a few years? What did they get wrong?
- If you were a judge in a Turing test, what question would you ask to find out which one is the computer?
- What can we learn from the fact that people felt ELIZA was “listening” to them?
Questions kids ask
Who invented artificial intelligence?
There’s no single inventor. Alan Turing laid some of the foundations, John McCarthy coined the term in the proposal for the 1956 Dartmouth workshop, and since then thousands of researchers all over the world have added to the field.
Has a computer passed the Turing test yet?
People argue about it, because the test isn’t fully defined and can be run in different ways. What matters more is remembering that a program that sounds like a person doesn’t necessarily understand like a person.
Why has AI moved forward so fast in recent years?
Mostly thanks to a combination of more powerful computers, enormous amounts of digital data and deep learning methods that use big neural networks.