Glossary: AI words, familiar and unfamiliar
At a glance
Some words about AI keep coming up again and again, but if their meaning isn't clear, the whole conversation feels hard. This page explains those words in plain English, with examples. The words are arranged so that reading from start to finish makes each one easier to follow — the basic words first, then specific names and risk words.
The terms
AI (artificial intelligence)
When a computer or a program seems to think like a person — it answers questions, understands writing, or helps with decisions — that's called AI, short for Artificial Intelligence. Think of it as a very well-read assistant who has learned from lots of books, but doesn't feel anything itself.
Algorithm
A set of step-by-step instructions for doing a task. A cooking recipe lists what to do, step by step — a computer algorithm is much the same, except that instead of cooking it gives instructions for calculations or decisions.
Data
A heap of information — numbers, words, pictures, anything that can be stored. AI learns from this data, just like a student learns by reading books.
Machine Learning
A way of teaching a computer by showing it lots of examples (data) so it spots patterns on its own, instead of giving it rules directly. For example, if you show a child cats and dogs again and again, they learn to tell the two apart by themselves — nobody writes down the rules.
Model
After machine learning from lots of data, the finished "knowledge" that comes out is called a model. Apps like ChatGPT and Gemini are really built on top of models.
Training
Showing a model hundreds of thousands of examples so it can learn is called training. It's a bit like years of study at school — the difference is that it's a computer doing the learning.
Chatbot
A program that can hold a conversation with people through writing or speech. AI assistants like ChatGPT and Gemini are chatbots, though today's chatbots are built on much more advanced AI than the older ones.
ChatGPT
An AI chat app made by a company called OpenAI. You ask in writing and it answers in writing, and you can ask questions by showing it pictures too.
Gemini
An AI assistant made by Google. It's easy to start with if you already have a Google account (Gmail).
Copilot
An AI assistant made by Microsoft. The name says it all — it's like a "co-pilot", built to sit beside you and help with your work.
Claude
An AI assistant made by a company called Anthropic. Many people use it for reading and understanding text, for writing, and for conversation.
Prompt
The question or instruction you write to AI is called a prompt. For example, "Write me a short note asking for a day off" is a prompt. The clearer your prompt, the more useful the answer.
Token
When AI reads or writes text, it breaks the text into small pieces and works with them — those pieces are called tokens. A short word can be a single token, while a long word may be split into two or three pieces. You don't need to know this for everyday use, but it's good to know what it means when you hear the word.
Multimodal
The ability to understand more than one kind of information at once — not just text, but also pictures or a voice — is called multimodal. For example, showing a picture and asking about it in writing is multimodal use.
Voice assistant
A helper program that you can talk to, and that answers in speech or writing. Phones have had familiar examples for a long time, and many AI apps today are adding the ability to ask questions by voice too.
Image generation
When you describe a picture in words and AI draws a new picture to match, that feature is called image generation. For example, if you write "draw a village scene beside a river", AI makes a picture for you on its own.
Deepfake
A picture or video made with AI that makes it look as if someone is saying or doing something they never really said or did. It can look so real that even people who know the person can be fooled — so it's best not to trust a strange or suspicious video too easily.
Hallucination (AI making things up)
Sometimes AI says something that isn't actually true, yet sounds completely confident. That's called a hallucination. Imagine asking someone a question they can't answer, but instead of saying "I don't know" they smoothly make up a guess — AI sometimes does exactly this. So if you get important information from AI, it's good to check it once more.
Bias
AI learned from lots of data written by people, so the unfairness or one-sided views that people have can sometimes show up in AI's answers too. That's called bias. For example, if someone grows up reading only one kind of news, their opinions may lean in one direction — something similar can happen with AI.
Robot vs AI
A robot is a physical machine that can move or do things with its hands. AI is the "brain" part that thinks or decides. A robot can have AI inside it, and AI can also live only in a phone app, with no body at all — ChatGPT, for example, has no body; it only talks in text.
AGI
AGI means an AI that could think and learn about anything, in any situation, as flexibly as a human. Today's AI tools are good at particular kinds of tasks, but this level of fully human-like intelligence doesn't really exist yet — it's still a topic for research and debate.
App vs Website
An app is a program installed on your phone that takes up space in the phone's memory. A website is a page opened in a browser (such as Chrome), with nothing to install. Many AI tools can be used both ways — as an app, or just by opening the website in a browser.
Account
A personal space you create on an app or website with your identity, where your information and your conversation history are saved. You usually sign up with an email address or a phone number.
Cloud
Storing information or doing calculations on big computers out on the internet, instead of on your own phone or computer, is called the cloud. For example, if your phone's photos are saved in a photo-storage app, they're really stored in the cloud, not on your phone. Most AI apps also work by using big computers in the cloud, so they usually need the internet to run.
OTP
The one-time code that arrives on your phone when you log in to an account or make a payment is called an OTP (One-Time Password). It's never okay to tell it to anyone — your bank or mobile wallet will never call you and ask for your OTP.
Scam / phishing
Trying to trick someone out of money or information by lying or using a fake identity is called a scam. Phishing is one kind, where fake messages or links that look like they come from a real bank or organisation are sent to steal information. AI can now write such fake messages to look even more convincing, so you need to be more careful than before.
Privacy
Controlling who can see your personal information and who can't is called privacy. When you ask AI a question, it's safest not to include sensitive information such as your national ID number, passwords or bank account details.
Open source
A program or model whose inner workings are shared openly, so anyone can look at it, use it or change it for their own needs, is called open source. The opposite is a tool whose inner workings are kept only by the company that made it.
Free tier
The part of an app you can use for free is called the free tier. It usually comes with some limits (for example, how many questions you can ask per day), and those limits can change over time — so the app itself has the most up-to-date information.
Subscription
Paying every month to use an app's extra features is called a subscription — the opposite of the free tier. This page doesn't recommend that you take out any paid subscription anywhere — it only tells you what the word means.
Automation
When you set up a rule once, and a task then happens by itself instead of you doing it by hand again and again, that's called automation. For example, an alarm you set on your phone rings by itself every day — that's a kind of automation too, though it doesn't need AI.
Digital literacy
The skill of using the internet, apps and now AI safely and sensibly is called digital literacy. Reading this whole glossary is itself part of growing your digital literacy.