What Is Artificial Intelligence in Simple Words?

By Dr Sam Donegan · MLAI · Updated

The short answer

Artificial intelligence, or AI, is technology that uses information to work out an answer, make a prediction, or create something. It helps phones turn speech into text, email services spot spam, and chatbots draft replies. Its answers can still be wrong.

Test your understanding →

You probably use AI already

You don't need a robot or a complicated app to come across AI. Think about:

  • Your inbox: a spam filter spots patterns in messages and decides which look suspicious.
  • Your phone: speech recognition turns what you say into written words.
  • Your music app: a recommendation system suggests songs based on listening patterns.
  • A chatbot: a writing tool drafts a reply from your instructions.

Each does a particular job. Being good at suggesting a song doesn't mean a system can give reliable advice about your health or finances.

What makes it AI?

A simple timer follows a rule: turn the light on at 7 pm. A spam filter has a different task. It has to work out whether a new message looks like spam, even if it hasn't seen that exact message before.

That step of working out an answer is called inference. The OECD's explanation of AI uses this idea to describe systems that make predictions, recommendations, decisions or new content from the information they receive.

Many AI tools learn patterns from examples. Others use knowledge and logic put together by people. An app can combine AI with ordinary software, such as a rule that asks you to approve a reply before it's sent.

How AI works, with one example

Imagine you run a small online shop and want help sorting customer emails. Here's a made-up message and one way an AI tool might handle it. This is an illustration, not a result from a tested product.

  1. 1. A message comes in

    “My order arrived, but one of the two mugs is missing. Can you send the other one?”

    Input

  2. 2. AI suggests a category

    The tool suggests “delivery problem” so the message can go to the right person. The model is the part of the tool that works out this suggestion.

    Model and output

  3. 3. A person checks

    You read the message, check the order and decide what to do. The AI suggestion hasn't checked your stock or sent a replacement.

    Human check

What could go wrong?

The tool might label the message “order not delivered”. That sounds close, but the parcel did arrive. Only one mug is missing. If you trust the label without reading the email, you could give the customer an unhelpful answer about tracking their parcel.

You correct the label to “missing item” and check what was packed. Keeping a record of mistakes helps you decide whether the tool is useful. It doesn't automatically teach the model; someone has to decide how those corrections are used.

AI, machine learning and generative AI: what's the difference?

You'll hear these terms together, but they mean different things.

Artificial intelligence

The broad field. It includes systems that recognise speech, suggest songs, make predictions or create content.

Example: your phone turning speech into text.

Machine learning

One way to build AI: use examples to find patterns, instead of writing an instruction for every possible case.

Example: a spam filter trained on messages people have already labelled.

Generative AI

AI that creates text, images, audio, video or code. Today's generative AI tools use machine learning.

Example: a chatbot drafting a product description from your notes.

Try it yourself

AI or ordinary software?

Choose AI or Not AI, then see why. You can also open all the answers below. Some apps mix AI with ordinary rules; focus on what each example describes.

0/8 answered
1. A calculator adds the exact numbers you type.
2. An email service spots spam by learning from earlier messages.
3. A light switches on at 7 pm because a timer was set for 7 pm.
4. A streaming service suggests a show based on what you've watched.
5. A contact form sends the same confirmation email after every submission.
6. A phone turns speech into text using examples of recordings and their written words.
7. Payroll adds overtime pay whenever hours worked go above a set limit.
8. A writing tool creates a new product description from your instructions.
Read all answers and explanations
  1. 1. A calculator adds the exact numbers you type. Not AI. The calculator follows set maths rules. It doesn't need to learn patterns or work out what you mean.
  2. 2. An email service spots spam by learning from earlier messages. AI. It uses patterns from earlier messages to judge a new one. An ordinary rule can then move suspected spam to a separate folder.
  3. 3. A light switches on at 7 pm because a timer was set for 7 pm. Not AI. You set the time, and the light follows that instruction. That's automation; it doesn't need AI.
  4. 4. A streaming service suggests a show based on what you've watched. AI. In this example, a model uses viewing patterns to predict what you might enjoy. Different services may make recommendations in different ways.
  5. 5. A contact form sends the same confirmation email after every submission. Not AI. Someone wrote the email in advance. The form sends that same message each time, so there's no AI involved in this step.
  6. 6. A phone turns speech into text using examples of recordings and their written words. AI. The model works out which words match the sounds. Background noise, accents and unfamiliar names can still trip it up.
  7. 7. Payroll adds overtime pay whenever hours worked go above a set limit. Not AI. This step follows a set rule about hours and pay. Other parts of the software might use AI, but this calculation doesn't need it.
  8. 8. A writing tool creates a new product description from your instructions. AI. The tool uses generative AI to write a description. Check the details: it might add a feature the product doesn't have.

These made-up examples are for practice. Your score isn't a qualification or a test of any particular app.

Want to learn with other people?

Explore MLAI talks, workshops and community catch-ups. New to AI? Check the event description for who it's for, what you need to know, the cost and whether it's online or in person.

Find an AI event

What is AI good at?

AI can help with jobs such as sorting lots of messages, transcribing a recording, spotting patterns and getting a first draft on the page. It's most useful when you can check whether it did the job well.

It can also miss details, repeat unfair patterns in its training data, or make up an answer. A confident tone doesn't tell you whether it's right. The NIST guide to generative AI risks explains why plausible but false answers need attention.

Four questions to ask before using an AI answer

  1. Is this information okay to share? Use made-up details while you learn. Keep personal information and confidential work out of public chatbots. The OAIC's privacy guidance explains what organisations should check.
  2. Can I check the answer? Look up names, numbers, dates and quotes in the original source. Open any links: a chatbot can invent a reference.
  3. What happens if it's wrong? A bad song suggestion is easy to skip. A decision about someone's health, money or job needs much more care and qualified help.
  4. Who makes the final decision? Be clear about who checks the work, fixes mistakes and can stop the tool from taking an action.

If you're using AI at work, the Australian Government's AI adoption guide goes into more detail on testing tools and deciding who is responsible.

Try a small task

Write three made-up customer messages: one asking about opening hours, one reporting a missing item, and one asking to change an order. Ask a chatbot to sort them by topic and explain its choices. Then read each message yourself. Did it miss anything? Would you use its suggestions?

This gives you a first look at how a tool behaves. Three examples aren't enough to decide whether it's ready for real customers.

Keep exploring

For more familiar examples, read how AI is used in everyday work and life. If you've heard about tools that take actions for you, our guide to AI agents explains what changes when software can do more than suggest an answer.

A few common questions

What is artificial intelligence in one sentence?

AI is technology that uses information to work out an answer, make a prediction, or create something, such as a reply or an image.

Is all automation AI?

No. A timer that turns a light on at 7 pm follows a fixed instruction. It doesn't need AI. An app can also use both: AI might suggest a reply, while an ordinary rule stops it being sent until you approve it.

Are AI and machine learning the same thing?

No. AI is the wider field. Machine learning is one way to build it, by finding patterns in examples. A spam filter, for instance, can learn from messages already marked as spam.

What makes generative AI different?

Generative AI creates text, images, audio, video or code. A chatbot drafting an email is one example. A convincing answer can still contain mistakes or made-up details.

Does AI think like a person?

A chatbot can sound like a person, but that doesn't show that it understands or has feelings. You can judge whether its answer is useful without assuming there's a human-like mind behind it.

What should I avoid putting into a public AI chatbot?

Keep other people's personal information and confidential work out of public chatbots. The OAIC advises organisations not to enter personal information, especially sensitive information, into publicly available generative AI tools. Use made-up examples when you're learning.
About this guide

The shop example and quiz are made up to help explain the ideas. They aren't product tests or a formal assessment. AI tools helped with research, writing and code. The sources below explain the definitions and advice in more detail.

Published . Updated with simpler explanations, a worked example and clearer next steps.

Sources and further reading

For a closer look at the ideas in this guide.

Disclaimer: This article provides general information and is not legal or technical advice. For official guidelines on the safe and responsible use of AI, please refer to the Australian Government’s Guidance for AI Adoption →

Next up

What Is Artificial Intelligence Used For in Everyday Work and Life

What is artificial intelligence used for? Learn the main everyday and business uses of AI, from pattern recognition and recommendations to automation and decision support.

Close-up of coworkers using an AI assistant on a laptop for everyday work tasks and decision support

Want to learn with other people?

Explore MLAI talks, workshops and community catch-ups. New to AI? Check the event description for who it's for, what you need to know, the cost and whether it's online or in person.

Find an AI event