BInvestigatorClassical machine learning· about 20 min· fundamentals that rarely change· verified 2026-09-20· EN
Rule-based systems and machine learning
Be able to explain the difference between a program where a human has written the rules and a system that has found the rules itself from examples, and determine which one suits a given problem.
Prerequisites
- BAlgorithmic thinkingrequired
- BData in everyday liferequired
Everyday explanation
Calculating VAT is easy to write as a rule: price × 0.25. Deciding whether an image depicts a cat is almost impossible to write as rules — how do you describe 'cat-ness' in steps?
Rule-based: the human writes the steps. Machine learning: the human shows thousands of examples (image + 'cat'/'not cat') and the computer finds a pattern itself.
Intuition
| Rules | Learning | |
|---|---|---|
| Who finds the pattern? | The human | The computer, from data |
| Suits when… | the rules can be written down | examples are numerous and rules are unclear |
| Errors occur when… | a rule is missing | examples are few, wrong, or biased |
| Can the answer be explained? | yes, step by step | often only partially |
Both are algorithms. The difference is where the rules come from.
Mastery means
- Give examples of problems that suit rules versus learning
- Explain why 'recognising a cat in an image' is difficult to write rules for
Sign in to do the exercises and build your mastery up.
Sources
Leads to
Part of the goals (37)
- AI for beginners
- Build an AI service that survives production
- Builder — collect data, train a model and test AI critically
- AI, ethics and society
- Language models in practice
- Data: collect, clean, document
- An AI service in operation
- Classical machine learning in practice
- Computers that see and hear
- Interpreting a language model
- AI safety in practice
- Responsible AI in practice
- Build a RAG system you can trust
- Classical ML for real
- Train your first neural network
- Discoverer — build a game and train a machine
- Build an agent you can trust
- Build a memory system for an agent
- Multimodal systems
- AI in production
- Fine-tune and run your own models
- Build a voice interface
- Train an agent with reward
- Training neural networks for real
- Build an NLP system end to end
- Frontier Lab — an independent research project
- Evals in practice
- Deep reinforcement learning
- Fine-tune a model with LoRA
- Image classification with convolutional networks
- Run models more cheaply: quantisation
- Understand how generative AI works
- Reproduce a paper
- Build a transformer from scratch
- Generative models in depth
- Seeing and hearing with AI
- Statistics for experiments