Learnai-foundationsWhat AI Actually Is

Lesson · 8 min

What AI Actually Is

Cut through the hype and define AI like an engineer.

Artificial Intelligence is a family of computer programs that perform tasks we used to think needed human intelligence: recognizing faces, understanding language, recommending videos, driving cars. It is math, data and a lot of engineering rather than magic or consciousness.

The three ingredients

  • Data — examples the system learns from (photos, sentences, clicks).
  • A model — a giant adjustable formula that turns inputs into outputs.
  • Training — the process of tuning the model until its answers improve.
AI doesn’t "know" facts the way you do. It finds patterns. A chatbot that writes a great poem is not reasoning like a poet; it is predicting which words usually come next.

That single idea explains both why AI feels brilliant and why it fails in weird ways. If you remember one thing from this module: AI is pattern-matching at enormous scale.

Try it yourself

Try it — order

Arrange the machine learning loop in the order it actually runs.

1Measure how wrong the guesses are
2Collect examples (data)
3Nudge the knobs to be less wrong
4Repeat until predictions improve
5Build a model with adjustable knobs

Knowledge check

1

Which of these is the best one-line definition of modern AI?

2

What happens during "training"?

3

A chatbot writes a perfect essay about a city. What does that prove?