MEPX
Chapter 2 of 12All chapters
  1. 1What AI means
  2. 2How machine learning works
  3. 3Training data
  4. 4Kinds of learning
  5. 5Neural networks
  6. 6Large language models
  7. 7Limits and hallucination
  8. 8Bias and fairness
  9. 9Using AI well
  10. 10Privacy and safety
  11. 11Beyond text
  12. 12AI and work

Chapter 2 of 12

How machine learning works

Guess, measure the error, adjust.

The training loop

A model starts with random numbers, makes a prediction, compares it against the right answer, and nudges the numbers to reduce the error. Repeat that billions of times and the numbers encode a useful pattern.

  • The measure of wrongness is the loss function, and training minimises it.
  • Nobody writes the rule. It emerges in the numbers, which is why it cannot simply be read back.

Training and inference

Training is the expensive one-off process. Inference is running the finished model on a new input, which is cheap by comparison and is what happens when you use one.