Short answer: pick the book that matches the question you are asking. Many AI reading lists put plain-English books next to programming textbooks and do not say which is which. This list is sorted by what you want to know. The order is our editorial choice, it is not a ranking, and each description only repeats what the publisher’s own page says about the book (checked 14 October 2026).
What a book can teach you about AI, and what it cannot
A good book teaches the ideas that change slowly: what the technology is trying to do, why it can sound confident and still be wrong, and who gets hurt when it is used badly. A book is a good place to learn those.
A book cannot give you two things:
- The current screen. A book is fixed when it is printed. Tool menus, buttons and plans change faster than a new edition can catch up, so a written walkthrough of a particular tool may not match what you see.
- Practice. You learn to prompt by typing a prompt, reading a weak answer and trying again. Reading about prompting helps, but only doing it builds the skill.
So use a book for understanding and a hands-on lesson for practice.
If you want to understand what AI is and where it falls short
Start here if your real question is “what actually is this thing?”
Artificial Intelligence: A Guide for Thinking Humans by Melanie Mitchell. Macmillan sells the current paperback under its Picador imprint (copyright 2019, with a preface added in 2025). Macmillan’s page says the book asks how intelligent the best AI programs really are, how they work, what they can do and when they fail. It also says the book looks at the gap between the hype and what AI has actually achieved.
Artificial Intelligence Basics: A Non-Technical Introduction by Tom Taulli, published by Apress in 2019. The publisher’s page calls it a non-technical introduction to ideas such as machine learning, deep learning and natural language processing, with real-world case studies. Two things to know before you choose it. First, the page aims it mostly at people bringing AI into an organisation. Second, it was published before the chat tools most beginners now use.
If you would rather see a chat tool than read about one, the free lesson on what AI chat tools are covers the basics in one short sitting.
If you want to use AI in your own work
This is the “how do I actually use it?” question. It is also where books help least, because the tools keep changing.
Co-Intelligence: Living and Working with AI by Ethan Mollick, published by Portfolio (Penguin Random House) in April 2024. Penguin Random House says the book asks you to treat AI as a co-worker, co-teacher and coach, and uses real examples from business and education. Of the books on this list, it is the most about everyday use.
Even so, any book about using AI describes the tools as they were when it went to print. Read it for the why and the when. For the how, go somewhere that shows today’s screens.
A hands-on course is that place. The free AI for Beginners: Start Here course is short and gives you the exact prompts we used, so you can copy them and try them yourself. When you want better answers, the lesson on the four parts of a good prompt is a good next step.
If you want to judge the hype and the risks
You may not want to use AI yet at all. You may just want to know whether to believe what you are being told.
AI Snake Oil by Arvind Narayanan and Sayash Kapoor, published by Princeton University Press in September 2024. Princeton’s page says the book explains how AI works and why it often does not, where it might be useful or harmful, and how to tell when companies are using hype to sell AI products that do not work. It also explains the differences between types of AI.
The Alignment Problem by Brian Christian, published by W. W. Norton. Norton describes it as a look at what goes wrong when we build AI systems, and at the people trying to fix it. The page gives examples such as hidden bias in systems that sort job applications. It is about risk and is not a how-to guide.
Whichever you read, practise checking what an AI tool tells you. It costs nothing. The lesson on how to check what it tells you turns the scepticism in these books into a habit. The post on what “AI responses may include mistakes” means explains the warning you will keep seeing under chat boxes.
Books to skip for now if you do not code
One example shows the pattern.
Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Aurélien Géron, published by O’Reilly. O’Reilly’s page says the book uses Python frameworks, with code examples and exercises throughout. It also says programming experience is all you need to get started. That suits someone who already programs. If you do not, it is the wrong first book.
A simple test: if the title names a code library, or the publisher says you need programming experience, leave it for now. Come back to it if you decide to learn to code.
How to pair a book with a short course
A book gives you the ideas that last. A course gives you the current screens and a chance to type something yourself. Each one covers what the other misses.
- Choose one book from the section that matches your question.
- Open AI for Beginners: Start Here alongside it. Its lessons have dated screenshots, so you can see when each screen was captured.
- After each chapter, try one thing in a chat tool. Keep it small, such as asking the tool to explain an idea from the chapter in plain English.
- When the book says something about what AI can or cannot do, test a small version yourself and check the answer.
Checkpoint: by the end of your first chapter, you should have read something and tried something. If you have only read, do step 3 before you turn the page.
For the bigger picture beyond books, see the post on the best ways to learn AI if you do not code.
Which one to start with
Choose by your question, not by what is popular.
- To understand what AI is: start with Mitchell.
- To use it in your work: start with Mollick, with the free course open alongside.
- To judge the claims: start with AI Snake Oil.
One book and one course are enough to start. Finishing one book is worth more than starting three.
Sources
- Penguin Random House: Co-Intelligence by Ethan Mollick (Portfolio)
- Macmillan: Artificial Intelligence: A Guide for Thinking Humans by Melanie Mitchell (Picador)
- Princeton University Press: AI Snake Oil by Arvind Narayanan and Sayash Kapoor
- W. W. Norton: The Alignment Problem by Brian Christian
- Springer Nature Link: Artificial Intelligence Basics: A Non-Technical Introduction by Tom Taulli (Apress)
- O'Reilly: Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow, 3rd Edition