The Year We Finally Started Reading About AI Seriously
There was a moment in late 2024 when I noticed something shift in the bookshop. Customers stopped asking vague questions about “that AI thing” and started asking specific ones. They wanted to understand what was happening, not just react to headlines. By 2025, this curiosity had turned into something measurable: according to a recent Gallup survey, 61 percent of American adults expressed a genuine desire to understand artificial intelligence better, yet had no clear sense of where to begin. That gap between curiosity and access is exactly where the best books of 2025 have planted themselves.
The numbers tell their own story. Oxford University Press reported a 67 percent increase in AI-themed nonfiction submissions between 2023 and 2025. Publishers watched public anxiety about the technology shift from abstract to urgent, and they responded by opening their gates to serious thinking on the subject. When the Frankfurt Book Fair compiled its 2025 nonfiction trends report, AI and society emerged as the single fastest-growing category, with over 400 titles from major publishers on display. This wasn’t a passing trend manufactured by marketing departments. It was genuine demand meeting genuine supply.
Mustafa Suleyman’s Resurgence and the Question of Prediction
It’s curious how certain books gain a second life through circumstance. Mustafa Suleyman’s “The Coming Wave,” published in 2023, reclaimed its place at the center of conversations about artificial intelligence when Suleyman became CEO of Microsoft AI in 2025. Suddenly, his predictions about the trajectory of AI development weren’t merely the musings of a technology entrepreneur. They became the declared thinking of a man now directly shaping that very trajectory. Publishers watched orders spike as readers returned to its pages with fresh urgency, searching for clues about what might be coming next.
What makes “The Coming Wave” worth revisiting now is not its prophetic tone, though it certainly has one. It’s Suleyman’s refusal to offer false certainty. He writes about artificial intelligence and synthetic biology as interlocking technological forces that will fundamentally reshape how we work, govern, and organize society. Reading it in 2025, with concrete examples of AI implementation already reshaping industries, creates an uncanny dialogue between prediction and reality. You find yourself marking passages not because they were correct, but because they’re now obviously incorrect in ways that matter deeply.
Kate Crawford and the Architecture of Power
Kate Crawford’s work has long operated in the space where technical detail meets political analysis. Her 2021 book “Atlas of AI” established a framework for understanding artificial intelligence not as a neutral tool but as a system embedded in vast infrastructure, shaped by economics and power. In 2025, MIT Press published Crawford’s updated essays expanding this framework with new data and sharper analysis. The centerpiece finding is both elegant and troubling: five companies now control over 80 percent of the compute capacity required to train large language models globally. That concentration of computational power isn’t just a technical fact. It’s a structural truth about who gets to shape this technology.
Crawford’s 2025 work is invaluable precisely because it resists both techno-utopianism and resigned pessimism. She takes you through the physical reality of data centers, the geopolitics of rare earth materials, the labor practices hidden behind consumer-facing AI products. You begin to understand that when we talk about artificial intelligence, we’re actually talking about a complex system involving energy infrastructure, global supply chains, and massive concentrations of capital. The MIT Press AI and Society catalog 2025 reflects this expanded interest in understanding AI through its material and political dimensions.
Where Publishers See the Conversation Heading
The Frankfurt Book Fair’s 2025 report offers a window into what serious publishers believe matters most about artificial intelligence right now. With over 400 titles in the AI and society category, the fair showcased a remarkable range of approaches: technical deep dives, policy analysis, philosophical inquiry, and urgent cultural criticism. The Frankfurt Book Fair 2025 trends report highlighted that the conversation has matured. Publishers are no longer satisfied with “AI will change everything” as a thesis. They’re demanding specificity: which sectors, in what ways, affecting whom, and with what alternatives available.
That maturation shows up in the kinds of books that succeeded commercially in 2025. Memoirs from AI researchers offered insider perspectives on lab culture. Policy-focused works from economists and governance experts explored regulatory frameworks. Sociological studies examined how AI implementation was reshaping workplace dynamics. What’s striking is the absence of a single dominant narrative. There’s no consensus book, no one tome that everyone points to as definitive. Instead, there’s a loose constellation of serious works, each taking a different angle on the same underlying questions.
Finding Your Entry Point in a Growing Library
The difficulty facing someone who genuinely wants to understand artificial intelligence and its social implications is no longer a lack of resources. If anything, the opposite problem now exists: where does one begin? A 2026 Gallup poll confirmed what bookshop conversations had already suggested. More than 61 percent of American adults wanted to understand AI better but couldn’t identify a starting point. For readers without a technical background, Crawford’s updated work provides perhaps the most accessible entry into the deeper structural questions. For those comfortable with technical concepts who want to understand policy implications, Suleyman’s book remains useful despite its age. For readers who want multiple perspectives, the breadth of 2025 publications offers genuine variety.
What strikes me most about the books published in 2025 is their collective refusal to pretend that artificial intelligence is something we can simply understand once and then move on from. These are books written in the present moment about a technology in motion, addressed to readers who need to think carefully because it’s already affecting their lives. They’re probably not the kinds of books that age well. In five years, they’ll seem dated by new developments and new data. But that’s exactly what makes them valuable now. They’re asking the questions we need to ask today, in language we can actually engage with. That seems like the most important thing a book about AI could do in 2025.