What is AI?
Your phone guesses your next word, and nobody taught it that word. That is AI in a nutshell. It learns patterns from examples, and it predicts.
First, what do you think?
Two quick questions. There are no wrong guesses yet.
What is AI?
The whole story in three sentences. AI is software that learned from examples instead of rules. From those examples it learned patterns. With those patterns, it predicts what most likely comes next.
You already know it. You type "See you" on your phone, and it suggests tomorrow. Your email puts an advert for cheap watches in the spam folder. A streaming app shows you a film you actually like.
Nobody wrote a rule for any of these. Nobody told your phone: "After see you, suggest tomorrow." The phone saw millions of messages. In many of them, see you was followed by tomorrow. So that is its best guess.
That is the difference that matters.
- Normal software follows rules. A person writes them down. "Over 50 euros, delivery is free." The computer does exactly that, every time.
- AI learns from examples. You show it lots of emails marked spam and not spam. It finds the patterns by itself. Then it predicts for a new email it has never seen.
The part that learned the patterns is called a model. You can think of it as a big set of learned patterns, packed into software.
AI stands for artificial intelligence. The name is older than the technology you use today. It sounds like a thinking machine. But the AI in your phone, your email and your chatbot doesn't think. It predicts.
AI learns patterns from examples, and then predicts what most likely comes next.
A first small step: today, watch your phone's word suggestions. Every suggestion is a prediction.
What the predictor does
In this course we call AI the predictor. It is the one idea that explains almost everything.
The rule: the predictor gives the answer that fitted best in its examples. Not the answer that is true. Not the answer you want. The one that fits the pattern.
An example. You type "I'll be there in five". Your phone suggests minutes. Most of the time that's right. But maybe you meant five o'clock. The phone doesn't know what you mean. It only knows what usually comes next.
That is why AI is often right, and sometimes strangely wrong. When your situation looks like the examples, the prediction is good. When it's unusual, the prediction can miss.
The same idea works for every kind of AI:
- A spam filter predicts: is this email spam?
- A photo app predicts: is there a cat in this picture?
- A chatbot predicts: which word comes next in the answer?
Try it yourself
Some programs follow rules a person wrote. Others learned from examples. Can you tell them apart?
Everything runs in your browser. Nothing you type is sent anywhere.
In practice
Three people meet AI without calling it that. The predictor explains what happens.
Case 1The strange suggestion
Sem (11) types a message to his gran: "Thanks for the lovely". His phone suggests day. He wanted to write jumper.
Think the phone is stupid, or that it knows him.
The phone is neither. It predicted the word that most often followed in its examples. Thanks for the lovely day is common. Jumper is not.
See the suggestion for what it is: a guess. Take it when it fits. Type your own word when it doesn't.
You stay in charge. The predictor helps you type faster, but it doesn't decide what you say.
Case 2The email in the spam folder
Fatima (34) works in an office. An important email from a new customer ends up in her spam folder.
Assume someone set a rule to block that customer.
Nobody did. The spam filter predicted spam. The email looked like the spam it learned from. A new address, a link, the word offer.
Mark it as not spam, and add the customer to your contacts. Check the spam folder now and then.
The predictor only knows its examples. A new customer doesn't look like them yet, so you keep an eye out.
Case 3"It knows what I like"
Henk (68) is surprised. His streaming app suggests a film about sailing, his favourite hobby.
Think the app listens in, or knows him personally.
It doesn't need to. Henk watched three sailing documentaries. People who watched those often liked this film too. The app predicted he would as well.
Know that what you watch shapes what you see next. Want something different? Search for it yourself.
The suggestions stop feeling like spying. They're predictions from your own viewing.
By situation
Spell checkers, translation apps and chatbots are all predictors. They're often right. But they predict what usually fits, not what your teacher asked. Read what they suggest before you use it.
Many work programs now have AI in them: email, spreadsheets, customer service. When one of them sorts, suggests or writes something, it's a prediction. Ask yourself: what examples did it learn from? Do they look like my situation?
Your phone, your TV, your smart speaker and your bank app use AI. Seeing it as a predictor takes away the mystery. It isn't watching you personally. It learned patterns from millions of people, and you're one of the examples.
Myths and mistakes
| Claim | True? | Why |
|---|---|---|
| AI is a robot | No | Most AI is software, in your phone or on a server. You never see it. |
| AI is brand new | No | Spam filters and word suggestions have used it for years. Chatbots are the newest kind. |
| Someone programmed every answer | No | Nobody wrote the answers. The model learned patterns from examples and predicts from them. |
| AI knows what you mean | No | It predicts what usually comes next. That is often what you mean, but not always. |
Related topics
Next
- How AI learnswhere the patterns come from, and what a model is.
- Language modelsthe predictor behind every chatbot.
Related
- AI around youwhere you meet AI every day.
Sources
- explainerGoogle for Developers. What is Machine Learning?.Machine learning trains software, called a model, to make useful predictions from data instead of hand-written rules. Four kinds: supervised, unsupervised, reinforcement learning and generative AI. Examples: spam, cats in photos, weather.
- governmentEuropean Parliament. What is artificial intelligence and how is it used?.AI in daily life: recommendations in online shops, web search, digital assistants, machine translation, smart thermostats, safety features in cars, cybersecurity.
Test
You get 5 questions about the whole page. Get 4 right, and you've mastered the topic.
About the tools in this course. Every tool is described in the same way, in alphabetical order, without a recommendation. Names, prices and features change often: the maker's own site is always right. This site was built with the help of Claude, one of the tools described here.