What is artificial intelligence? A plain-English explanation
What AI actually is, how it learns, what it can and can't do, and why it suddenly seems to be everywhere. No jargon, no maths.
First published · AI Digested
Artificial intelligence, or AI, is the name for computer systems that can do things we normally think need human intelligence. Understanding speech, recognising a face in a photo, translating between languages, writing an email, spotting a tumour on a scan, beating a grandmaster at chess. When a computer does any of these, we call it AI.
That's the whole definition. Everything else is detail about how.
The old way: telling computers exactly what to do
For most of computing history, programs were lists of rules written by people. A tax calculator follows rules: if income is over this amount, apply this rate. It never surprises you, and it never gets better on its own.
Early AI researchers, from the 1950s onwards, tried to build intelligence the same way. They wrote thousands of rules about chess, medicine or grammar. This worked for narrow, tidy problems. It failed badly for messy, real-world ones.
Try writing rules for recognising a cat. Pointy ears? Some cats have folded ears. Whiskers? So do dogs. Four legs? Not if it's sitting down or half out of shot. Humans recognise cats instantly but cannot explain how. That turned out to be the core problem.
The new way: learning from examples
Modern AI takes a completely different approach, called machine learning. Instead of writing rules, you show the computer a huge number of examples and let it work out the patterns itself.
Show a system a million photos labelled "cat" or "not cat", and it gradually adjusts itself until it can label photos it has never seen. Nobody tells it about ears or whiskers. It finds its own clues, some of which a human would never think of.
The most powerful kind of machine learning uses neural networks, systems of millions or billions of simple connected calculations, loosely inspired by brain cells. When those networks have many layers, it is called deep learning. Almost every impressive AI you have heard of since 2012 is deep learning.
The key idea: traditional software follows rules people wrote. Modern AI learns its own rules from data. That's why it can do things nobody knew how to program, and also why it sometimes behaves in ways nobody expected.
What about ChatGPT?
Chatbots like ChatGPT, Claude and Gemini are built on large language models. These are neural networks trained on an enormous amount of text: a large part of the public internet, books, articles and computer code.
Their training task sounds almost silly: read some text and predict the next word. But doing that well, across billions of examples, forces the model to pick up grammar, facts, writing styles and patterns of reasoning. The result can answer questions, write essays, explain ideas and write software.
This is why chatbots feel so different from earlier AI. Older AI did one job. A language model can attempt almost anything you can describe in words.
AI you already use every day
You were using AI long before chatbots arrived:
- Your phone camera uses AI to sharpen photos, blur backgrounds and recognise faces.
- Your email uses AI to filter out spam.
- Maps apps use AI to predict traffic and estimate arrival times.
- Netflix, Spotify, YouTube and TikTok use AI to decide what to show you next.
- Your bank uses AI to flag unusual payments that might be fraud.
- Voice assistants like Siri and Alexa use AI to turn your speech into text.
- Online translation tools use AI to translate between languages.
What AI is good at, and what it isn't
AI is very good at:
- spotting patterns in huge amounts of data,
- doing repetitive tasks quickly and consistently,
- producing a decent first draft of text, images or code,
- working in many languages,
- staying available at 3am.
AI is still bad at, or unreliable for:
- being sure it's right. Chatbots can state false things with total confidence. These are called hallucinations.
- common sense about the physical world, in ways that can be surprising.
- knowing recent events, unless it is connected to search.
- taking responsibility. It can't be held accountable for a decision, so people still need to be.
Narrow AI and "general" AI
Almost all AI so far is narrow AI: good at a specific job. A chess engine cannot drive a car.
Language models are broader than anything before them, which is why there's so much talk about AGI, artificial general intelligence: AI that could do almost any intellectual task a person can. The biggest AI companies say AGI is their goal. Experts disagree, strongly, about when or whether it will arrive and what it would mean. See our glossary for more.
Where to go next
Common questions
Is AI the same as a robot?
No. A robot is a physical machine. AI is software. Some robots use AI to see and move, but most AI, like ChatGPT or your spam filter, has no body at all.
Does AI actually think?
Not in the way people do. AI systems process patterns in data very effectively, and the results can look like thinking, but they have no feelings, wants or experiences. Whether future systems could genuinely understand is an open and hotly debated question.
Who invented AI?
No single person. Alan Turing asked whether machines could think in 1950, and John McCarthy coined the term 'artificial intelligence' in 1955 for a workshop held at Dartmouth College in 1956. Thousands of researchers have built on their work since.
Is AI dangerous?
AI brings real, present-day risks, such as convincing scams, false information and unfair automated decisions, and some experts worry about larger risks from much more powerful future systems. It also brings real benefits. Knowing how it works is the best protection.
Sources and further reading
Spotted a mistake or something out of date? Tell us and we will fix it. Read how we research and write our articles.
Read next
AI vs machine learning vs deep learning: what's the difference?
AI, machine learning, deep learning, generative AI: people use these words as if they mean the same thing. Here's how they fit together, in one simple picture.
Updated 22 November 2026
Machine learning explained: how computers learn from examples
What machine learning is, the three main ways machines learn, real examples you use every day, and why data matters more than anything else.
Updated 22 November 2026
How ChatGPT works: large language models explained simply
How do chatbots like ChatGPT, Claude and Gemini actually work? A plain-English explanation of large language models: tokens, training, prediction and why they sometimes get things wrong.
Updated 22 November 2026