What is AI? Explained in simple terms, with no maths

AI, or artificial intelligence, is software that learned how to do a task by studying a huge number of examples, instead of following rules a programmer wrote by hand. The AI most people use today, chat assistants like Claude, ChatGPT and Gemini, learned from an enormous amount of text. They work by predicting what words should come next. That simple idea, done at a giant scale, is enough to write, explain, translate, summarise and write code.

Normal software vs. AI

A calculator follows rules. Someone wrote down exactly what to do for every button, and it does the same thing every time.

AI is built differently. Nobody writes the rules. Instead:

1. You collect a vast number of examples.
2. A program adjusts billions of internal numbers until it gets good at predicting those examples.
3. What’s left is a model: a file full of numbers that has picked up the patterns.

It’s closer to how a child learns a language than to how a calculator works. Nobody teaches a child grammar rules first. They hear millions of sentences and absorb the patterns.

How a chat AI really works

When you type a question, the AI doesn’t look up an answer in a database. It does something stranger:

• It reads your message.
• It predicts the most fitting next piece of text, a word or part of a word.
• It adds that piece, then predicts the next one, and the next.

Sentence by sentence, an answer appears. Because it learned from so much human writing, the patterns it picked up include facts, reasoning steps, writing styles and how to structure code.

This explains two things people find confusing.

Why it sounds so human. It learned from human writing, so it writes like us.

Why it’s sometimes confidently wrong. It’s producing what sounds right, and usually that is right. When it isn’t, there’s no alarm inside the AI that goes off. People call this a hallucination.

What AI is good at

• Explaining things at whatever level you ask for
• Drafting and editing emails, posts, reports and messages
• Summarising long documents
• Translating between languages, and between formal and casual tone
• Brainstorming options you hadn’t thought of
• Writing code, from a spreadsheet formula to a whole app
• Answering “how do I…” questions with steps

I use it for the last two every day. The games in my arcade and most of the tools in my portfolio were built by describing what I wanted to an AI assistant and checking the result, which is called vibe coding.

What AI is bad at

• Knowing what it doesn’t know. It may invent a source, a quote or a number.
• Recent events. A model learned from text up to a certain date. Unless it can search the web, it won’t know what happened after.
• Exact arithmetic and counting, unless it uses a calculator tool behind the scenes.
• Your private context. It doesn’t know your situation unless you tell it.
• Remembering. Most assistants start each conversation fresh.

A good working rule: treat AI like a brilliant, fast assistant who is new to your life and occasionally bluffs. Give it context, and check anything that matters.

Words you’ll hear

• Model. The trained “brain”, like Claude or GPT.
• Prompt. What you type to the AI.
• Token. A small piece of text, roughly three-quarters of a word. Models read and write in tokens.
• Context window. How much text the AI can keep in view at once.
• Agent. An AI that doesn’t just answer, but takes steps: searching, editing files, running programs.

There’s a longer list in my glossary of AI terms for non-technical people.

Is AI actually intelligent?

It depends what you mean. It doesn’t have feelings, goals or experiences the way people do. But “it only predicts the next word” undersells it. To predict text well, a model has to pick up a lot about how the world works, and it can combine ideas in ways that weren’t in its examples.

For everyday purposes, the philosophy matters less than the practical question: is it useful for this task, and how will I check the result?

How to get better answers

Three habits make most of the difference.

1. Give context. Say who you are, what you’re trying to do and who it’s for. “Explain mortgages” gets a textbook. “Explain mortgages to a first-time buyer in India who is comparing fixed and floating rates” gets something useful.
2. Ask for a format. A table, five bullet points, a step-by-step list, a one-paragraph summary.
3. Go back and forth. The first answer is a draft. Say what’s wrong with it. “Shorter.” “Less formal.” “You missed the cost.”

And one safety habit: for health, money, legal questions or anything you’ll repeat to others, ask for sources and check them.

Do you need to be technical to use AI?

No. If you can write a clear message, you can use AI. The skills that matter are ordinary ones: explaining what you want, noticing when an answer is off, and asking a follow-up.

If you want to go one step further, two small things pay off quickly: learning what Markdown is, because it’s the format AI writes in, and learning the basics of GitHub, because it’s where AI-made work gets saved.