02
Lesson 2 · 2 min
How machines learn
A machine learns by guessing, checking how wrong it was, and adjusting itself a tiny bit, billions of times.
Imagine teaching a computer to guess house prices in Nairobi. You give it thousands of past sales: size, area, number of rooms, and the real price.
At first its guesses are random. After each guess it sees the real answer and nudges its internal settings so the next guess is a little closer. These settings are called parameters. Big models have billions of them.
Repeat this enough times and the guesses become good. The finished result is called a model. Using the model later to answer new questions is called inference.
One important catch: a model only knows what was in its training data. If the examples were unfair or out of date, the model will be too. This is where bias comes from.
Do it now
Find the training data
- 1.Think of an app you use daily that makes suggestions: music, shopping, maps or video.
- 2.Ask yourself: what examples did it learn from? Probably your past clicks and other people's.
- 3.Ask a chatbot the prompt below to check your thinking.
I use [APP NAME]. What kind of data did its recommendation system probably learn from, and what kind of bias could that create? Answer in plain language, under 120 words.
Words from this lesson
- Machine learning
- Teaching computers by showing them many examples so they find the patterns themselves, instead of being given every rule.
- Training data
- The examples, such as text, images or recordings, that a model learns its patterns from.
- Model
- The trained file of learned patterns that does the actual work inside an AI product. One model can power many apps.
- Parameters
- The adjustable numbers inside a model that are set during training. Big models have billions of them.
- Inference
- The moment a trained model is actually used to produce an answer, as opposed to the earlier learning phase.
- Bias
- When an AI treats some people or ideas unfairly because of skewed patterns in the data it learned from.