Available Spring 2026
Experience is all you need
For seventy years, computer scientists trying to build thinking machines kept trying different approaches: logic, then search, then probability, then pattern recognition. Each worked within limits and failed beyond them. Experience Is All You Need proposes that those failures pointed at the same gap: what was missing wasn't a better design but experience itself.
Intelligence, this book argues, is not located in a brain, a type of code, or an inherited set of traits. It accumulates through contact with a world that pushes back: through error and its consequences, through time, through the particular knowledge that comes from being wrong when it costs something, and the argument is that the process works the same way whether it is unfolding in a person or a machine.
To show how that process unfolds, the book draws on two novelists who traced it with unusual care. Jane Austen's fiction is built around the gap between what a person believes about themselves and what experience will eventually force them to understand. Joseph Conrad asks a darker version of the same question: what happens to a person when the world doesn't give them time to learn, and the information it does give is hostile, distorted, or comes too late?
Reading both writers alongside contemporary machine learning research, the book builds an intuition for how intelligence actually develops, not through technical explanation but through the recognizable human experience of getting things wrong, being corrected, and slowly, imperfectly, getting better. Why can a system trained on vast quantities of text still fail in ways that a child would not? Why does learning from a description of an experience produce something fundamentally different from learning through having one?
Rejecting both human exceptionalism and technological mysticism, this is a book about what intelligence actually is, where it comes from, and what it would take to build it, in a machine or in a life.