Skip to content
EN

Lesson 1 of 8 · 5 min · +100 XP

What is AI, really?

Artificial intelligence (AI) is software that learns patterns from examples instead of only following rules a person wrote. Shown many labelled photos, texts or sounds, it builds a model that can recognise, predict or create new ones. That makes AI strong at patterns but not always right, so treat its output as a clever guess and check it.

▶ Try it

Why it matters

A normal program follows rules a person wrote. AI learns its own rules from examples, which is why it can recognise faces and write text, and also why it can be confidently wrong.

In 5 minutes you'll be able to

  • Normal programs follow written rules. AI learns patterns from examples.
  • What AI learns is a model that turns new inputs into predictions.
  • Treat AI output as a clever guess you check, not as the truth.

What you'll practise

  1. Rules vs patterns

    Two ways to make a computer do something clever.

    📜 Written rules
    A normal program does exactly what a person wrote: “if the total is over 50, add free delivery”. Predictable, and it never surprises you.
    🖼️ Learned from examples
    AI is shown thousands of examples, like photos labelled “cat” or “not cat”, and finds the pattern itself. Nobody writes the rule.
    🧩 The model
    What AI learns is stored as a model: a huge set of numbers that turns a new input into a guess.
    🎯 A guess, not a fact
    AI's output is always a prediction: the most likely answer based on its examples. Usually useful, sometimes wrong.

  2. Rules or learned?

    Did a person write the rule, or did the machine learn it from examples?

    • A calculator works out 12 × 7 → Written rules
    • A phone unlocks when it recognises your face → Learned (AI)
    • An alarm clock rings at 7:00 → Written rules
    • A spam filter that has seen millions of junk emails → Learned (AI)
    • A website rejects passwords shorter than 8 characters → Written rules
    • Lumi writes a birthday poem for your gran → Learned (AI)
    • A map app predicts traffic from past journeys → Learned (AI)
    • A heating system switches on below 19 °C → Written rules

  3. Lab calls

    Three moments at Lumen Lab. What's really going on?

    1. Noor wants a program that spots ripe bananas in photos. Writing a rule for every shade of yellow is impossible. What's the AI way?

      • Show it lots of photos labelled “ripe” and “not ripe” (Answer)
      • Write an even longer list of colour rules

    2. Lumi tells you your essay is “really strong”. What is that, really?

      • A prediction of what fits, based on patterns in text (Answer)
      • Lumi's honest personal opinion
      • A fact that has been checked

    3. A friend says: “AI is basically a robot brain that understands everything.” What's closer to the truth?

      • It's software that finds patterns in data (Answer)
      • It thinks and understands like a person

Remember

AI learns patterns from examples instead of following written rules, so its answers are smart guesses, not facts.

AI doesn't follow your rules. It learns patterns from examples, and patterns can be wrong.

Key facts

  • The term “artificial intelligence” was introduced in a 1955 proposal for a summer research project at Dartmouth College, which took place in 1956. · Source: McCarthy, Minsky, Rochester & Shannon, A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence, 1955
  • The OECD defines an AI system as a machine-based system that infers from the input it receives how to generate outputs such as predictions, content, recommendations or decisions. · Source: OECD, Recommendation of the Council on Artificial Intelligence (definition updated 2023)
  • Since 2 February 2025, the EU AI Act has asked organisations that provide or use AI to work on their staff's AI literacy. Since a 2026 amendment, the duty is to take measures that support it, without guaranteeing a set level. · Source: EU AI Act (Regulation (EU) 2024/1689), Article 4, as amended by Regulation (EU) 2026/1744
  • Most AI you meet every day is narrow: it is trained for one kind of task, such as recognising faces or predicting text, and doesn't transfer that skill to unrelated jobs.

Updated:

Questions people ask

What is the difference between AI and machine learning?

AI is the broad goal of making machines do tasks that seem to need intelligence. Machine learning is the main way modern AI gets there: the system learns patterns from examples instead of being given every rule.

What are examples of AI in everyday life?

Face unlock on phones, spam filters, voice assistants, translation apps, recommendations on streaming and shopping sites, traffic predictions in map apps and chatbots that write text.

Is AI the same as a robot?

No. A robot is a machine that moves in the physical world, and many robots follow simple written rules. AI is software; it can run inside a robot, but most AI runs on phones, computers and servers.

Does AI think like a human?

Not really. AI finds statistical patterns in data and uses them to make predictions. It can sound human, but it has no feelings, goals or understanding of the world the way people do.

Related terms