Short answer
Artificial intelligence (AI) is the broad goal of making machines do tasks that normally need human intelligence. Machine learning (ML) is a part of AI where systems learn from data instead of following hand-written rules. Deep learning (DL) is a part of machine learning that uses neural networks with many layers. Think of three nested circles: all deep learning is machine learning, and all machine learning is AI, but not the other way round. Today's chatbots and image generators are generative AI built with deep learning.
The three terms in one picture
- AI is the outer circle: any technique that makes a computer act intelligently, from a chess program with hand-written rules to a self-driving car.
- Machine learning sits inside AI: the computer learns patterns from examples.
- Deep learning sits inside machine learning: the learning is done by deep neural networks.
- Generative AI and large language models are applications of deep learning that create new text, images, audio or code.
Comparison table
| Artificial intelligence | Machine learning | Deep learning | |
|---|---|---|---|
| What it is | The whole field of intelligent machines | Systems that learn from data | ML with many-layered neural networks |
| Coined | 1956 (Dartmouth workshop) | 1959 (Arthur Samuel) | Popular from the 2006 to 2012 period |
| How it works | Rules, search, logic or learning | Statistical models trained on data | Neural networks trained with backpropagation |
| Data needed | Varies, can be none | Moderate, often structured | Large amounts, often unstructured |
| Hardware | Any computer | CPUs are often enough | GPUs and AI accelerators |
| Human feature design | Rules written by experts | Often needs hand-picked features | Learns features automatically |
| Examples | Chess engines, route planners, expert systems | Spam filters, credit scoring, recommendations | Image recognition, speech, ChatGPT |
What is artificial intelligence?
The term "artificial intelligence" was coined for a 1956 workshop at Dartmouth College organised by John McCarthy and others. Early AI focused on rules and logic: experts wrote down how to make decisions, and the computer followed them. This is sometimes called symbolic AI or "good old-fashioned AI". It still powers many systems, such as route planning and game AI.
What is machine learning?
Arthur Samuel, an IBM researcher who built a checkers program that improved by playing itself, popularised the term "machine learning" in 1959. Instead of writing rules, you give the system examples and let it find the patterns. Common types:
- Supervised learning: learn from labelled examples (emails marked spam or not spam).
- Unsupervised learning: find structure in unlabelled data (grouping customers by behaviour).
- Reinforcement learning: learn by trial and error from rewards (game-playing agents, robot control).
Classic ML methods such as decision trees, random forests and linear regression are still widely used, especially for spreadsheets and business data.
What is deep learning?
Deep learning uses neural networks with many layers that learn their own features directly from raw data such as pixels, sound waves or text. It took off around 2012, when a deep network trained on GPUs (AlexNet) dramatically beat older methods at image recognition. Since then deep learning has driven most AI breakthroughs: speech recognition, machine translation, AlphaGo, protein structure prediction and large language models.
Deep learning needs a lot of data and a lot of computing power, which is why it runs on GPUs with tensor cores. See GPU vs CPU for AI.
Where do generative AI and LLMs fit?
Generative AI is deep learning used to create new content. Large language models generate text and code using the transformer architecture, and diffusion models generate images and video. When these models use tools and take actions, they become AI agents. The long-term goal of matching human ability across all tasks is called AGI.
Which should you learn first?
- To use AI tools: start with how LLMs work and good prompting.
- To build ML models: learn Python, basic statistics and classic ML with scikit-learn, then deep learning with PyTorch. Our Python volume and learning paths are good starting points.
- To understand AI hardware: learn how GPUs run neural networks in our AI and hardware fundamentals course.
Frequently asked questions
Is machine learning the same as AI?
No. Machine learning is one way of building AI. Some AI systems use hand-written rules instead of learning from data.
Is ChatGPT machine learning or deep learning?
Both. ChatGPT is built on large language models, which are deep neural networks, and deep learning is a type of machine learning.
Does deep learning always need a GPU?
Training large deep learning models effectively requires GPUs or other accelerators. Small models can run on a CPU, especially for inference.
Sources and further reading
- McCarthy et al., A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence (1955)
- Samuel, Some Studies in Machine Learning Using the Game of Checkers, IBM Journal (1959)
- LeCun, Bengio and Hinton, Deep learning (Nature, 2015)




Leave a Reply