Minds We Create: AI and a Future Still Being Written offers a rigorous, accessible account of the most consequential technological development of the early twenty-first century - written for the informed general reader who wants clarity rather than hype.
Drawing on history, organisational theory, and governance literature as much as on technical AI research, the book argues that the central challenges of artificial intelligence are not primarily engineering problems. They are institutional problems: questions about how humans build and maintain complex systems under competitive pressure, how organisations respond to safety concerns when those concerns are inconvenient, and how governance frameworks develop - or fail to develop - when technology moves faster than the institutions meant to oversee it.
The book opens with two historical episodes - the near-launch of a nuclear torpedo aboard a Soviet submarine in 1962, and the FDA pharmacologist whose principled refusal to approve thalidomide kept a generation of American children safe - and uses these not as decoration but as analytical tools. Both stories illuminate the same structural question: what does it take for safety-conscious judgment to have weight, under pressure, when the incentives run the other way?
From that foundation, Minds We Create works through the technical substance of AI risk - alignment, interpretability, adversarial vulnerabilities, the specific failure modes of large language models - before turning to the governance landscape: what functioning collective oversight looks like, where current frameworks fall short, and what the most important unsolved problems are.
The book was completed in February 2026, as a real-time standoff between the US Department of Defense and an AI company over autonomous weapons and mass surveillance was unfolding - an episode the final chapters address directly.
Twenty-one chapters. A full glossary, bibliography, and index. Written for intellectually engaged adults with no prior technical background.
Audience: General nonfiction readers; policy professionals; technology executives; students of governance, ethics, and emerging technology.
Comparable titles: The Alignment Problem by Brian Christian; Human Compatible by Stuart Russell; The Age of Surveillance Capitalism by Shoshana Zuboff.
