The Coming Wave — Technology, Power, and the Twenty-First Century's Greatest Dilemma

An Insider’s Warning

Mustafa Suleyman co-founded DeepMind, one of the world’s most influential AI research laboratories, in 2010. He was present at the creation of some of the most significant AI systems of the past decade. He then wrote a book arguing that what he and his colleagues have built represents one of the most serious civilisational risks in human history.

This combination of positions requires accounting for, and Suleyman does provide one — though some readers will find it insufficient. The argument is essentially that the development of these systems was inevitable given the economic incentives and scientific opportunities; that the alternative to building them responsibly was having them built irresponsibly by someone else; and that having helped build them, he now has both a platform and an obligation to argue for containment.

Whether you find this reasoning adequate is a reasonable question. What is not reasonable is to dismiss the substantive case that The Coming Wave makes on the basis of who is making it.

The Convergence Thesis

The book’s central argument is structural: AI and synthetic biology — separately powerful enough to reshape society — are converging. The tools of AI are being applied to biological research; the techniques developed in biology are informing AI; the same underlying logic of self-improving, self-replicating systems runs through both. Together, they constitute a wave of technological change unlike anything the modern world has encountered.

Suleyman calls this convergence the “coming wave” and argues that it will arrive faster and with more force than most public discourse anticipates. The reason is what he terms the “containment dilemma”: the benefits of these technologies — in medicine, productivity, scientific discovery — are so large that enormous economic and political pressure exists to develop and deploy them quickly. But rapid development and deployment is precisely what makes safety measures hardest to implement.

The Nation-State Problem

One of the book’s most important arguments concerns governance. The nation-state system was designed to manage threats that respected national borders — military invasions, trade conflicts, conventional regulatory challenges. AI and synthetic biology do not respect those borders. A model trained in one country can be downloaded worldwide. A synthetic biology technique published in an academic paper is available to anyone who reads it.

This creates a governance gap that Suleyman argues cannot be filled by the existing international architecture. No current institution has the mandate, the technical expertise, or the enforcement mechanisms to regulate technologies of this kind. Creating such institutions would require the kind of international cooperation that has proven extremely difficult to achieve even on much simpler problems.

Suleyman is careful not to imply that this problem is insoluble. His “containment” framework — a set of principles rather than a blueprint — points toward what adequate governance might look like: technical safety measures embedded in the technology itself, international agreements with actual enforcement mechanisms, and the creation of new institutions with the specific authority and capacity required. But he is honest that the path from here to there is not clear.

The Economic Incentive Problem

The most uncomfortable section of The Coming Wave concerns economic incentives. The companies developing advanced AI are doing so because it is enormously commercially valuable; the governments supporting AI development are doing so because it is economically and strategically valuable; the researchers working on it are doing so because it is intellectually rewarding and professionally lucrative. These incentives are largely orthogonal to safety.

This is not a new observation — critics of the AI industry have been making it for years. But Suleyman makes it with specific authority: he describes, from the inside, the way that commercial pressures and safety requirements exist in tension within the organisations building these systems, and the way that tension is typically resolved.

What Distinguishes This Book

The Coming Wave is not the most technically detailed book on AI risks — The Alignment Problem is more thorough on the technical specifics. It is not the most philosophically sophisticated — Nick Bostrom’s work goes deeper into the formal structure of the risks. What distinguishes it is the combination of technical credibility, inside knowledge, and moral seriousness deployed in a book written for a general audience.

Suleyman writes clearly about complex systems without over-simplifying them. His alarm is calibrated rather than performative — he describes specific mechanisms of risk rather than gesturing at vague catastrophe. And his acknowledgement that he is himself implicated in what he is warning about gives the book a quality of moral complexity that makes it more, not less, trustworthy.

The Urgency

The most striking quality of The Coming Wave is its temporal urgency. Suleyman argues repeatedly that the window for establishing meaningful governance before the technologies become too powerful and too distributed to regulate is closing — that decisions made in the next few years will determine the trajectory of the next century. Whether or not this specific timeline is correct, the underlying argument — that there is a window, and that windows close — is sound.

Our rating: 4.3/5 — Essential reading on AI governance from someone with the authority and credibility to be taken seriously.