B2 · How Artificial Intelligence Works: Data and Learning

What Is Data?

Data is a collection of facts, such as numbers, words, measurements, pictures or sounds, gathered to answer a question. Information is what data becomes once it is organised and understood, so that it can guide a decision.

8 min Beginner Jump to the activity

What you need: Nothing. This page starts from the beginning, with no programming and no formulas.

Definition

How would you decide where to hold a picnic that everyone enjoys?

Data is a collection of facts, such as numbers, words, measurements, pictures or sounds, gathered to answer a question. Information is what data becomes once it is organised and understood, so that it can guide a decision.

In simple words

You ask ten friends where they want the picnic: six say the park, three say the lake, one says the zoo. Those ten answers are data. "Most people want the park" is information. Deciding to go to the park is the decision. AI works the same way: it turns huge amounts of data into useful information.

Why it matters

AI learns from data, so data is the raw material of every AI system. If you understand what data is and how it becomes information, you understand where AI's answers come from, and why bad data leads to bad answers.

Key terms

Data
Recorded facts, such as numbers, words, measurements, pictures or sounds.
Information
Data that has been organised so that it means something and answers a question.
Knowledge
Understanding built from information and experience, which lets you decide what to do.
Dataset
A collection of data gathered for one purpose, such as all the answers to a survey.
Raw data
Data exactly as it was collected, before any sorting, cleaning or counting.
Noise
Data that is irrelevant or wrong and hides the pattern you are looking for.

Explained in detail

Data is older than computers

The word "data" comes from Latin and means "things that are given". People have recorded data for thousands of years. Clay tablets from ancient Mesopotamia list sacks of grain and heads of cattle. Rulers have counted their people in censuses since ancient times, and India's census is today one of the largest counting exercises in the world. A shopkeeper's account book, a doctor's register of patients and a farmer's record of rainfall are all data. Computers did not invent data. They made it possible to store and study far more of it.

From data to wisdom

Data by itself says very little. The number 39 means nothing until you know it is a body temperature in degrees Celsius. People who work with data often describe four levels. Data is the raw fact: 39. Information is data with meaning: "the child's temperature is 39 degrees". Knowledge is understanding what follows: "39 degrees is a high fever". Wisdom is knowing what to do: "give fluids, and see a doctor if it does not come down". AI systems work mostly at the first two levels. They are very good at turning large amounts of data into information. Knowledge and wisdom still depend heavily on people.

LevelWhat it isExample
DataRaw facts31, 33, 35, 36, 38
InformationFacts with meaningThe afternoon temperature rose every day this week
KnowledgeUnderstanding the patternA heat wave is building
WisdomKnowing what to doMove the sports day to the morning

Every dataset starts with a question

Good data work begins with a clear question, because the question decides which data is worth collecting. If you want to know which snacks to sell at a school fair, the ages of the customers matter and their shoe sizes do not. Collecting everything "just in case" produces noise: data that fills pages, costs effort to store and hides the pattern you need. The same is true for AI. A model built to predict rainfall does not improve when you feed it cricket scores.

Why AI needs so much of it

A person can learn what a mango is from one or two mangoes. A machine learning model finds patterns only by comparing very many examples, so it needs data in large quantities: thousands of labelled photos, millions of sentences, years of weather records. The digital world now produces such quantities every day, from phones, sensors, cameras, payments and websites. This flood of data, together with faster computers, is the main reason AI has improved so quickly in recent years.

How it works

StagePicnic example
QuestionWhere should we hold the picnic?
DataTen answers: park, lake, park, zoo…
InformationSix of ten want the park
DecisionBook the park for Sunday

Try it yourself

From raw facts to a decision

A group is planning a picnic. Watch answers turn into a decision.

Collect data to plan an event

Try this with family or friends.

  1. Write one question, such as "Which snack should we buy for the party?"
  2. Ask at least eight people and write down every answer.
  3. Count the answers for each choice.
  4. Write one sentence of information, such as "Most people want samosas".
  5. Make your decision and say which data it came from.

Think about it

  • Would the answer change if you asked only your best friends?

Data or information?

Is each item raw data, or information that someone has worked out?

  • This week is getting hotter every day

  • Blue is the most popular colour

  • A list of 200 names and their favourite colours

  • Photos from a traffic camera

  • Temperatures: 31, 33, 35, 34, 36 °C

Common mistakes

  • Thinking data means only numbers. Words, photos and sounds are data too.
  • Thinking more data is always better. Data that doesn't fit the question is noise.
  • Mixing up data and information. Data needs organising before it can guide a decision.

Check yourself

Question 1

1. Which is information rather than data?

Question 2

2. Why does AI need data?

Question 3

3. Is a song recording data?

Summary

  • Data is collected facts: numbers, words, pictures and sounds.
  • Information is organised, understood data that guides decisions.
  • AI turns large amounts of data into information.

Finished reading?

Sources: Elements of AI (University of Helsinki)