The Best AI Books Worth Reading… and the Ones to Skip
I work in AI and read 150 books a year. Here’s every AI book I’ve reviewed so far, in the order I’d recommend reading them.
- Category: AI, Technology, Business, Policy
- Updated: June 2026
I’ve spent the last year reading through the wave of AI books flooding the market. Some are excellent. Several are quite good. A few you can skip entirely. If you’re a busy professional trying to make sense of the AI moment without wasting time on the wrong books, this is the guide I wish existed when I started.
I work in AI, read 150 books a year exclusively on audio, and set a high bar for nonfiction that’s insightful, concise, and trustworthy.
1. Co-Intelligence by Ethan Mollick
★★★★☆ | 4.5 hours
Wondering how to use AI? Start here.
Think you already know how to use AI? Still consider starting here.
This is a thought-provoking look at how to incorporate AI into your day-to-day work. Mollick’s frameworks (built from research on AI adoption, not just speculation) aren’t just a list of to-do’s that AI can help with; he shares how to think critically about using AI yourself as well as how to help your team use it.
Read it if: you want practical, jargon-free guidance for incorporating AI into your work.
2. The Optimist by Keach Hagey
★★★★☆ | 12 hours
Neutral and well-documented, this AI industry intro is the one I keep recommending. The first 40% covers deep background on Sam Altman (well done, but not of interest to everyone). The rest is the most concise, approachable overview of the industry I’ve read. It’s focused on OpenAI, but it also covers topics like the philosophical factions inside AI (safety-focused effective altruists vs. acceleration-focused technologists vs. profit-driven builders). Once you have that map, AI headlines start making a lot more sense.
Read it if: you want a balanced industry overview before diving into more opinionated accounts.
3. The Coming Wave by Mustafa Suleyman and Michael Bhaskar
★★★★☆ | 12 hours
The clearest policy-level framing of AI risk available, and the most level-headed voice in a space full of either cheerleaders or doom-sayers. Suleyman’s core argument is that AI is arriving at a moment of historically low institutional trust, with no reliable governance mechanism. It’s a simple, useful frame for understanding why AI policy is so hard. Some ideas have become common knowledge since its 2023 publication, but it’s still a great entry point for AI governance thinking.
Read it if: you want to understand AI risk and policy.
4. The Infinity Machine by Sebastian Mallaby
★★★★☆ | 15 hours
The most intellectually curious AI book I’ve read, and the one most willing to go deeper than the headlines. Centers on Demis Hassabis and DeepMind rather than Altman and OpenAI — a perspective that’s underrepresented in the AI book canon. This traces the competing academic approaches that shaped the industry, which helps contextualize what’s happening today. It takes a philosophical lens to the risks and unknowns around AI evolution and policy. The first third is a dense biography that you can choose to skim or even skip.
Read it if: you want to understand how AI developed as a technology and you’re ready to go a level deeper.
Full review: The Infinity Machine
5. Empire of AI by Karen Hao
★★★★☆ | 18 hours
An excellent choice if you want a thoughtful critique of how the AI industry is currently structured, including topics like the invisible labor of data annotation workers, the environmental cost of AI infrastructure, and the risks of power concentration in a handful of companies.
Unfortunately, those topics get buried inside a very long account of OpenAI’s internal politics, adding up to a whopping 18 hours on audiobook.
Read it if: you’re interested in AI’s immediate, measurable harms (labor, environment, power). Although this runs long, the topics it covers are important.
6. If Anyone Builds It, Everyone Dies by Eliezer Yudkowsky and Nate Soares
★★★★☆ | 6 hours
The title is not a metaphor. Hear directly from leaders of the “AI Doomers” faction of the industry why they truly believe this. Their core argument is that engineers don’t fully understand how AI models interpret goals, so at some point, some AI somewhere will set a goal that is destructive to humanity. Understanding this perspective is increasingly necessary for anyone working in or around AI, and the authors walk through it in a clear and engaging way.
Read it if: you want the AI doom argument made rigorously, by someone who helped build the field.
Full review: If Anyone Builds It, Everyone Dies
Excellent and AI-Adjacent: The Thinking Machine by Stephen Witt
★★★★★ | 10 hours
This biography of Jensen Huang and Nvidia’s rise from gaming chips to AI infrastructure dominance is the best strategic leadership book I’ve read in a while. If you’re here for OpenAI, ChatGPT, and the AI product landscape, this isn’t that. What it is: thought-provoking insight into how transformative technology bets actually get made, told through a biography of an important and complex character.
Read it if: you want leadership and strategy frameworks alongside AI.
Full review: The Thinking Machine
Outdated – Skip It: Supremacy by Parmy Olson
★★☆☆☆ | 9 hours
The best parallel coverage of OpenAI and DeepMind’s simultaneous evolution I’ve found — that part is genuinely useful. But the book presents strong opinions as documented fact, and having read more rigorous and recent accounts of the same events in the books listed above, I caught multiple oversimplifications. When it was published in 2024 it may have been the best accessible AI overview available, but today it’s not your best choice.
Skip it and read The Optimist for the OpenAI story, The Infinity Machine for DeepMind, or Empire of AI for another strong perspective about the dangers of AI.
New AI books added as reviewed — check back for additions.
