16 Must-Read Books for Understanding Artificial Intelligence
A practical playbook for partnering with AI at work, school, and in daily life—clear, current, and immediately applicable.
This book is suitable for listening to while commuting to work (which is what I did.) I am a big fan of Ethan Mollick. His works bridge the gap between the technologist and the user.
I found the book to be focused on how a normal person can use large Language Models like ChatGPT and Claude (Click here for an ever changing list) it has helped me to be more effective in my work life and private life.
The sharpest business lens on AI: treat AI as "cheap prediction" to redesign processes and strategy.
11/16/2025
I am currently reading this book. As someone with a background in Economics and 3 years trying to figure out how to use AI/ML in a business setting, Power and Prediction really hits home. Using electricity’s adoption pattern as a model to predict Machine Learning’s adoption pattern has been very useful for me. The projects I have been working on have been, as Agrawal et al state, point solutions. Using the technology to do what we already do within the existing systems. The real benefits will occur once we know how to use ML effectively and build systems around ML. As I work to uncover opportunities and vet AI as a Service providers, I am starting to look for vendors who are thinking in systems instead of functions.
A superb tour of bias, safety, and how to align systems with human goals; grounded in real failures and fixes.
11/16/2025
I am listening to this book as I drive to work. I have been surprised by this book. I thought it would be full of tropes about fairness and ethics in an ivory tower sort of navel gaze or conversations about how to get AI to care about humanity as pets, so it works in our behalf instead of its own. It is not that at all. It really delves into issues that need to be addressed to make these tools useful. The historical understanding of how people have tried to manage themselves or others through algorithms really helps broaden the conversation and allows us to gather lessons about how to handle when our stories about how the world works runs into how the world really works. When I was at the MIT Media Lab Conference three years ago I was struck by the problem of trying to build models that reflect how we want the world to be as opposed to how the world really is. This book lays out that problem – it will be interesting to see if the book has answers.Big-picture risks, governance, and why "containment" of powerful tech matters for institutions and society. (New afterword 2024)
This book is well worth reading, if only because of its author, Mustafa Suleyman. Suleyman is one of the entrepreneurs behind DeepMind, the company responsible for AlphaGo and AlphaFold. He comes from a social justice background, has become a highly successful capitalist, and now leads Microsoft’s efforts to develop personal agents—technology with the potential to most profoundly shape the lives of every man, woman, and child in the developed world, for better or for worse. The book argues that we are heading toward both the best of times and the worst of times. While unsettling to read, this prediction rings true. It calls the reader to action and has given me the time to position myself to ride the wave rather than be engulfed by it. This book is very much about the author’s beliefs. In particular he has great confidence in government’s ability to manage AI/ML to good endings. That feels naive to me. These tools will take time to figure out by the people using them and government regulators are always slower than the industries that are trying to create value out of the new thing – be it a chemical for mind alteration or a tool that lets you write 25 emails in the time it used to take you to do 2 emails. While reading about his beliefs is interesting I am reminded by my father’s injunction to me: “Nobody cares what you think, they only care what you can prove.”
A clear executive blueprint for when, where, and how AI creates value.
To Be Read.
The human story of modern AI—Hinton, LeCun, Hassabis, and others—useful context for why today's systems look the way they do.
To Be Read.
The case for "provably beneficial" machines and a future-safe AI design philosophy.
To Be Read.
How AI entwines with labor, the environment, data extraction, and power—great for policy and social-impact lenses.
This is the first book I read about AI when I first woke up to the revolution that was going on around me. I picked it up in a politically focused bookstore in Washington D.C. on a trip with my son. This book focuses on the political ramifications of Machine Learning. From the way models are trained to ownership and the possibility of a consolidation of power that will be dangerous to self-governance. I find it useful to understand the power dynamics I am operating in and how to plan and prepare for complaints and criticisms around the things I create with Machine Learning.
Classic on how opaque models warp education, credit, employment, and democracy; the accountability agenda starts here.
To Be Read.
How to structure ML projects, diagnose errors, and prioritize effort—perfect "manager's guide" to making ML work.
To Be Read.
A human-centered memoir from a pioneer of modern computer vision; pairs history with what responsible AI should be.
To Be Read.
What AI means in healthcare—for clinicians, patients, workflow, and ethics.
To Be Read.
A bracing critique of current methods; helpful to see limits and alternative paths.
To Be Read.
Insider history of how deep learning moved from labs to the economy; bridges tech and impact.
To Be Read.
The flagship textbook; skim select chapters for foundations that clarify what today's systems can/can't do.
To Be Read.
Essential on search, representation, and structural bias—how seemingly neutral systems shape society.
To Be Read.