AI is the whole field of computers doing thinking-type tasks. Machine learning is the main way AI is built today: the computer learns from examples instead of following rules a person wrote. Deep learning is one kind of machine learning that uses neural networks with many layers. They fit inside each other like nesting dolls: all deep learning is machine learning, and all machine learning is AI, but not the other way round.
- AI is the umbrella: any machine doing tasks we think of as needing a mind, whether it learned them or was given rules.
- Machine learning is AI that learns patterns from examples, like a spam filter trained on emails people marked as junk.
- Deep learning is machine learning with large, many-layered neural networks. It needs far more data and computing power.
- ChatGPT, Claude and Gemini sit in the smallest doll: they are deep learning models, so they are machine learning and AI too.
| AI | Machine learning | Deep learning | |
|---|---|---|---|
| What it is | The whole field: machines doing thinking-type tasks. | One way of building AI: learning from examples. | One kind of machine learning: many-layered neural networks. |
| Where the skill comes from | Hand-written rules, search, or learning. | Patterns found in data, not rules typed in. | Patterns found in very large amounts of data. |
| Everyday example | A 1980s expert system, or a chess computer. | A spam filter. | Face unlock, voice assistants, chatbots. |
| Fits inside | Nothing: it is the biggest doll. | AI. | Machine learning, and so AI. |
Each row is sourced in the sections below.
Think of “vehicle”, “car” and “electric car”. Every electric car is a car, and every car is a vehicle, but a bicycle is a vehicle that is not a car. AI, machine learning and deep learning work the same way: each word is a narrower group inside the one before.
How do AI, machine learning and deep learning fit together?
The plain point: they are nested, not separate. Each one sits inside the one above it.
IBM, the computer company, describes AI as “a series of nested or derivative concepts”, and sets out the order: “Directly underneath AI, we have machine learning”, and “Deep learning is a subset of machine learning that uses multilayered neural networks”.1 The same page puts generative AI, the kind that writes and draws, one level further in: it “refers to deep learning models that can create complex original content”.1
So the words are not rivals. Someone saying “AI” today usually means the smallest doll.
What counts as AI, if it is not machine learning?
The plain point: AI is defined by what a system does, not how it learned to do it. Some AI never learned anything from examples.
NIST, the US government’s standards agency, defines an AI system by its outputs: one that can “generate outputs such as predictions, recommendations, or decisions influencing real or virtual environments.”2 Nothing in that definition says the system has to learn.
In the 1980s, AI mostly meant expert systems: programs that stored a specialist’s knowledge as facts and rules written by people. A European Commission report describes them as having “the knowledge base – a collection of facts, rules, and relationships on a specific domain” and an engine for applying them.3 IBM’s own chess computer, Deep Blue, is another example: IBM files it under “reactive machines” with “no memory”, built “to perform a very specific task”.4
That is the part of AI that machine learning does not cover: rules or search, not examples.
What makes machine learning different?
The plain point: nobody types in the rules. The computer is shown examples and works out the patterns itself.
Google’s course for developers gives the definition in one line: “Machine learning (ML) is a way to train software, called a model, to make predictions or generate content using data.”5 IBM stresses the break with normal programming: the models “make decisions or predictions without explicit, hard-coded instructions.”6
The classic example is a spam filter. People mark emails as junk; the model studies them and learns what junk tends to look like. Google uses spam as its own example of learning from labelled examples.7 A rule-based filter would need a person to list every trick spammers use. A learned one picks up new patterns from new examples.
What makes deep learning “deep”?
The plain point: “deep” means many layers in the network, not deep thinking.
A neural network passes information through layers of small calculating units. Google’s glossary calls a model deep once it is “A neural network containing more than one hidden layer.”8 IBM puts the line at “at least 4 layers” and adds that modern networks are often much deeper.9
The extra layers come at a cost. IBM says deep learning “requires an exceedingly large amount of data and computational resources for training”, and that it took off partly thanks to powerful graphics chips (GPUs).10 In return it handles messy things such as speech and photos: Apple says its “Hey Siri” detector uses a deep neural network.11 IBM’s AI page goes further: “Some form of deep learning powers most of the artificial intelligence (AI) applications in our lives today.”1
Where do ChatGPT and generative AI fit?
The plain point: in the smallest doll. Chatbots run on deep learning models, so they are all three at once.
ChatGPT, Claude and Gemini run on large language models. IBM calls those “a category of deep learning models trained on immense amounts of data”.12 Image makers and voice tools are deep learning models too: together they are generative AI, which IBM places inside deep learning (quoted above).
So “Is ChatGPT AI or machine learning?” has a simple answer: both, and deep learning as well. The narrower word just tells you more about how it was made.
Why does the difference matter to you?
The plain point: the narrower the word, the more it tells you about how a product behaves.
- “Uses AI” tells you little. It could be a fixed set of rules or a chatbot. In this site’s reading, the useful question is what the product does: sort, recommend, recognise or create.
- Learned systems inherit their examples. A machine-learning model guesses from the data it saw, so gaps and mistakes in that data carry through. That is why answers from these tools are worth checking; see hallucination.
- Deep learning explains the scale. Its hunger for data and computing power is behind the giant data centres and the short list of companies at the leading edge; see what AI runs on.
AI is the field, machine learning is learning from examples, and deep learning is machine learning with many-layered neural networks. Every chatbot is all three; plenty of AI is only the first.
Questions people ask
Is all AI machine learning?
No. Expert systems ran on rules people wrote, and IBM classes its Deep Blue chess computer as a reactive machine with no memory (sources above). Most AI built today is machine learning, but the word AI is older and wider.
Is deep learning better than machine learning?
Not always; it is one kind of machine learning with different costs. IBM notes it needs far more data and computing than classic methods (source above). In this site’s reading, a small, well-defined job with tidy data may not need it.
Is ChatGPT deep learning?
Yes. It runs on a large language model, which IBM calls a kind of deep learning model (source above). That makes it machine learning and AI as well.
What is the difference between a neural network and deep learning?
A neural network is the structure; deep learning is training big ones with many layers. Google calls a network deep once it has more than one hidden layer (source above); see neural network meaning.
Where to go next
- “Give me each word on its own.” Read artificial intelligence, machine learning and deep learning.
- “How is one actually built?” See how AI is made, step by step.
- “What are the other ways of sorting AI?” See types of AI, explained and narrow AI.
- “How did we get here?” Read the history of AI.
- “I am completely new.” Start with AI explained for complete beginners, then the AI glossary.
QUICK CHECK · 3 QUESTIONS
Pick an answer to see if you got it. Nothing is recorded.
1. How do AI, machine learning and deep learning relate to each other?
- They are three separate fields that do not overlap
- They fit inside each other: deep learning inside machine learning, inside AI
- Deep learning is the biggest group and AI sits inside it
Show the answer
Answer: They fit inside each other: deep learning inside machine learning, inside AI. The page describes them as nesting dolls: all deep learning is machine learning, and all machine learning is AI.
2. Which of these is AI but not machine learning, according to the page?
- A 1980s expert system built from rules people wrote
- A spam filter trained on emails marked as junk
- A chatbot running on a large language model
Show the answer
Answer: A 1980s expert system built from rules people wrote. Expert systems stored a specialist’s knowledge as hand-written facts and rules, so they did not learn from examples.
3. What does the word deep mean in deep learning?
- The model understands ideas more deeply than people
- The model was trained for a very long time
- The neural network has many layers
Show the answer
Answer: The neural network has many layers. The page explains that deep refers to the number of layers in the network, not to deep thinking.
- 1IBM, What Is Artificial Intelligence (AI)?, updated 2 Oct 2026, read at source 19 Oct 2026“a series of nested or derivative concepts” … “Directly underneath AI, we have machine learning” … “Deep learning is a subset of machine learning that uses multilayered neural networks”. ↑agenerative AI “refers to deep learning models that can create complex original content”. ↑b“Some form of deep learning powers most of the artificial intelligence (AI) applications in our lives today.” ↑c
- 2NIST, Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1, January 2023, read at source 1 Oct 2026 (as cited on this site’s artificial intelligence meaning page): “an engineered or machine-based system that can, for a given set of objectives, generate outputs such as predictions, recommendations, or decisions influencing real or virtual environments.”↑
- 3European Commission Joint Research Centre (Delipetrev, Tsinaraki & Kostić), AI Watch: Historical Evolution of Artificial Intelligence, EUR 30221 EN, 2020, read at source 5 Oct 2026 (as cited on this site’s history of AI page): “Expert systems had two components: 1. the knowledge base – a collection of facts, rules, and relationships on a specific domain; and 2. the inference engine that described how to manipulate and combine these symbols.”↑
- 4IBM, Types of Artificial Intelligence, read at source 5 Oct 2026 (as cited on this site’s types of AI page): reactive machines have “no memory and are designed to perform a very specific task”; Deep Blue “beat chess grandmaster Garry Kasparov in the late 1990s”.↑
- 5Google, What is Machine Learning?, updated 27 Jan 2026, read at source 1 Oct 2026 (as cited on this site’s machine learning meaning page): “Machine learning (ML) is a way to train software, called a model, to make predictions or generate content using data.”↑
- 6IBM, Machine Learning, updated 2 Apr 2026, read at source 1 Oct 2026 (as cited on this site’s machine learning meaning page): “This pattern recognition ability enables machine learning models to make decisions or predictions without explicit, hard-coded instructions.”↑
- 7Google, Supervised Learning, updated 25 Aug 2025, read at source 1 Oct 2026 (as cited on this site’s machine learning meaning page): “Supervised learning uses labeled data to train models that predict outcomes for new, unseen data.” … “like identifying spam or predicting precipitation.”↑
- 8Google, Machine Learning Glossary: ML Fundamentals, read at source 1 Oct 2026 (as cited on this site’s deep learning meaning page): deep model, “A neural network containing more than one hidden layer.”↑
- 9IBM, What Is Deep Learning?, 15 Sep 2025, read at source 1 Oct 2026 (as cited on this site’s deep learning meaning page): “‘deep learning’ refers to the training of models with at least 4 layers (though modern neural network architectures are often much ‘deeper’ than that).”↑
- 10IBM, What Is Deep Learning?, 15 Sep 2025, read at source 1 Oct 2026 (as cited on this site’s deep learning meaning page): “Relative to classic machine learning techniques, deep learning requires an exceedingly large amount of data and computational resources for training.” … “enabled in part by advancements in high-performance graphic processing units (GPUs).”↑
- 11Apple Machine Learning Research, Hey Siri: An On-device DNN-powered Voice Trigger for Apple’s Personal Assistant, October 2017, read at source 5 Oct 2026 (as cited on this site’s machine learning meaning page): “The ‘Hey Siri’ detector uses a Deep Neural Network (DNN) to convert the acoustic pattern of your voice at each instant into a probability distribution over speech sounds.”↑
- 12IBM, What Are Large Language Models (LLMs)?, 6 Oct 2021, read at source 5 Oct 2026 (as cited on this site’s LLM meaning page): “Large language models (LLMs) are a category of deep learning models trained on immense amounts of data, making them capable of understanding and generating natural language.”↑
Read at source 19 Oct 2026: IBM, What Is Artificial Intelligence (AI)? (updated 2 Oct 2026).
Reused from this site’s meaning pages and articles, with the read dates recorded there: NIST AI 100-1 (artificial intelligence meaning, 1 Oct 2026); Google’s machine learning course, supervised learning page and glossary, IBM’s machine learning, deep learning and LLM pages, and Apple’s “Hey Siri” paper (machine learning, deep learning and LLM meaning pages, 1–5 Oct 2026); the European Commission JRC report (history of AI, 5 Oct 2026); IBM’s types of AI page (types of AI, 5 Oct 2026).
The comparisons on this page (nesting dolls; vehicle, car and electric car) and the advice marked “this site’s reading” are this site’s own ways of explaining, not claims from these sources.
Free printable AI cheat sheets, and AI news every week.
Sixteen one-page AI guides and checklists in one free PDF, no sign-up. And if you want to keep up, the newsletter covers what changed in AI each week, in plain English, with the sources.
SUBSCRIBE FREE ↗ GET THE CHEAT SHEET PACK