Minds We Create

**Minds We Create: AI and a Future Still Being Written** is a nonfiction work on artificial intelligence safety, alignment, and governance for general readers with no prior technical background. The book opens with a specific image: a Soviet submarine officer in 1962, cut off from all communication in the middle of the Cuban Missile Crisis, who prevented a nuclear launch because the decision structure around him gave his judgment the weight it needed. That story is not a digression — it is the book's central argument made concrete. The safety of powerful systems depends less on the systems themselves than on the institutional structures, organizational incentives, and governance frameworks that determine whether careful judgment prevails under pressure. Drawing on historical cases including the Challenger disaster, early nuclear governance, and the management of shared commons, the book builds a framework for understanding AI risk as primarily a human and institutional problem rather than a purely technical one. The alignment challenge, interpretability, adversarial vulnerabilities, and the specific failure modes of large language models are all covered and explained without sacrificing accuracy, but always in service of a broader structural argument. The book's final chapter examines events from early 2026 — including a direct confrontation between a major AI company and the US Department of Defense over safety restrictions — as a live test case of the dynamics analyzed throughout. Author: Erik Bernath Publisher: Bernáth Erik Nándor Year: 2026 ISBN (paperback): 978-615-82993-0-5 ISBN (ebook/EPUB): 978-615-82993-2-9 ISBN (audiobook): 978-615-82993-4-3