Artificial intelligence is the broad field of making computers perform tasks that seem intelligent. Machine learning is a subset of AI where systems learn patterns from data instead of following hand-written rules. Deep learning is a subset of machine learning that uses large neural networks. Generative AI is an application, mostly of deep learning, that creates new content.
Nested circles
The easiest way to picture the relationship is as circles inside each other:
- Artificial intelligence (AI) is the largest circle: any technique that lets computers do things we associate with intelligence.
- Machine learning (ML) sits inside AI: systems that learn from data.
- Deep learning (DL) sits inside ML: learning with large, multi-layered neural networks.
- Generative AI overlaps mostly with deep learning: models that create text, images, audio or video.
Artificial intelligence
AI includes older approaches that do not learn at all, such as rule-based "expert systems" written by hand, and search algorithms that play chess by exploring possible moves. If a system behaves intelligently, it falls under AI, regardless of method.
Example: a chess program that searches millions of possible moves using hand-crafted evaluation rules.
Machine learning
Instead of writing rules, you give the system examples and let it learn the pattern. Classic ML methods include linear regression, decision trees, random forests and support vector machines. They work very well on structured data such as spreadsheets.
Example: predicting house prices from size, location and age, using past sales data.
Deep learning
Deep learning uses neural networks with many layers, loosely inspired by the brain. Given enough data and computing power, deep networks can learn very complex patterns, which is why they dominate image recognition, speech and language.
Example: recognising faces in photos, or transcribing speech.
Generative AI
Generative models produce new content rather than just classifying or predicting. Large language models and image generators are the best-known examples.
Example: writing an essay draft, generating an illustration from a description.
Quick comparison
| Learns from data? | Typical data | Example | |
|---|---|---|---|
| Rule-based AI | No | Hand-written rules | Simple expert system |
| Classic ML | Yes | Tables, structured data | Loan default prediction |
| Deep learning | Yes | Images, audio, text | Image recognition |
| Generative AI | Yes | Huge text and media collections | Chat assistants, image generators |
What should you learn?
- Want to use AI tools well? Start with AI literacy and prompting.
- Want to build AI apps? Learn Python and model APIs; see generative AI development.
- Want to analyse data and build predictive models? Learn classic ML first; see machine learning and data science.
- Want to research new models? Plan for strong maths and usually a university degree.
Frequently asked questions
Is ChatGPT machine learning or deep learning?
Both. It is a deep learning model, which is a type of machine learning, which is part of AI.
Should I learn classic machine learning before deep learning?
Usually yes. Classic ML teaches core ideas such as training, evaluation and overfitting in a simpler setting.
Is data science the same as machine learning?
Data science is broader. It includes collecting, cleaning, analysing and communicating data, and may or may not involve ML.
