Introduction to AI
Senior 1 · ICT
Introduction to AI
Lesson 2 of 5
22 min
A little over seventy years ago, "a machine that thinks" was a question for mathematicians and science-fiction writers. Today AI runs inside cheap phones. Knowing the kinds of AI helps you separate what is real from what is only in films, and knowing its history shows that progress came in bursts, with long pauses in between.
| Type | What it can do | Does it exist? |
|---|---|---|
| Narrow AI (weak AI) | One task, or a small group of tasks, often very well | Yes: every AI in use today, from face unlock to chatbots |
| General AI (strong AI, AGI) | Learn and do any intellectual task a human can, and move skills from one area to another | Not yet. Researchers disagree about when, or whether, it will arrive |
| Super AI (superintelligence) | Be more intelligent than the best humans in almost every area | No. It is an idea discussed by scientists and writers |
Worked example
A chess program beats every human player but cannot recognise a cat or hold a conversation. However brilliant it is at chess, it is narrow AI, because its ability does not carry over to other tasks. Even a modern chatbot that writes essays, answers questions and translates is still counted as narrow AI by most experts: it cannot, for example, learn to drive a car or cook a meal the way a person can.
| Year | Event |
|---|---|
| 1950 | Alan Turing, a British mathematician, asks "Can machines think?" and proposes the imitation game, now called the Turing test: if a person chatting in writing cannot tell whether they are talking to a machine or a human, the machine passes. |
| 1956 | A summer workshop at Dartmouth College in the USA launches AI as a field of research. The name "artificial intelligence" comes from John McCarthy's proposal for it. |
| 1966 | ELIZA, one of the first chatbots, is built by Joseph Weizenbaum at MIT. It matched keywords to reply, but many users felt it understood them. |
| 1970s and late 1980s | "AI winters": progress is slower than promised, so money and interest dry up. |
| 1997 | IBM's Deep Blue beats world chess champion Garry Kasparov in a match. |
| 2011 | IBM Watson wins the TV quiz show Jeopardy! against two former champions. |
| 2012 | "Deep learning" with large neural networks makes a big jump in recognising images. |
| 2016 | Google DeepMind's AlphaGo beats Lee Sedol, one of the world's best players of the board game Go. In Rwanda, Zipline drones start flying blood to hospitals. |
| 2022 | ChatGPT is released, and AI chatbots quickly reach hundreds of millions of people. |
Tip
Memory hook: Narrow = now (every AI you can use today); General = human-level at everything (not yet); Super = beyond humans (not yet). For history, remember the anchors 1950 Turing, 1956 Dartmouth, 1997 Deep Blue, 2022 ChatGPT.
Classify as narrow, general or super AI: (a) a program that translates Kinyarwanda into English; (b) an imagined robot that can learn any job a human can do; (c) an imagined system far cleverer than all scientists together.
Solution: (a) Narrow AI, which exists today; (b) general AI, which does not exist yet; (c) super AI, which does not exist.
Describe the Turing test in one or two sentences.
Solution: A person has a written conversation with a hidden partner, which may be a machine or a human. If the person cannot reliably tell that the partner is a machine, the machine passes the test.
What was an "AI winter", and why did it happen?
Solution: A period (in the 1970s and again in the late 1980s) when funding and interest in AI fell sharply, because the results were much weaker than researchers had promised.
Put in order: AlphaGo beats Lee Sedol; the Dartmouth workshop; Deep Blue beats Kasparov; ELIZA is built.
Solution: Dartmouth workshop (1956) → ELIZA (1966) → Deep Blue (1997) → AlphaGo (2016).
Summary
Sources and credits
The worked example, the timeline selection and the practice items were written by Ganzaa.
This lesson adapts CC BY-SA material and is shared under CC BY-SA 4.0.
