Sixty-five years ago a small group of maverick thinkers gathered in a converted Buddhist museum on the outskirts of Cambridge, convinced that language held the key to unlocking artificial intelligence. Their leader was the extraordinary and eccentric Margaret Masterman – linguist, logician, philosopher, religious devotee, and forgotten prophet of AI.

With the discovery of an archive that had remained unopened for nearly forty years, Peter de Bolla has pieced together the remarkable story of Masterman's creation and direction of the Cambridge Language Research Unit (CLRU). Working on the fringes of the university, dismissed as an amateur and constrained by structural sexism, Margaret Masterman and the CLRU worked at an extraordinary pace to crack the problem of machine translation. Taking Turing's theoretical proposal of an 'electronic brain' and running furiously with it, they built the first machine that could think – a machine capable of engaging with human minds, and even, in Masterman's prophesised future, of doing philosophy alongside them.

Her audacious claim that a machine could generate meaning, not merely manipulate symbols made – and still makes – her both a prophetic voice and a heretic. Blending biography, intellectual history, the drama of scientific revolution, and the exhilarating discovery of AI, this book reveals how Masterman and her collaborators – from philosophers, linguists and botanists to quantum theorists – came to formulate a theory of language and mind that we are only now beginning to see in action in the large language models shaping our world today.

Peter de Bolla is professor of cultural history and aesthetics at the University of Cambridge.

This title views best in tablet-style eReaders.
This is a set of uncorrected page proofs. It is not a finished book and is not expected to look like one. Errors in spelling, page length, format and so forth will all be corrected by the time the book is published several months from now. Photos and diagrams, which may be included in the finished book, may not be included in this format. Uncorrected proofs are primarily useful so that you, the reader, might know months before actual publication what the author and publisher are offering. If you plan to quote the text in your review, you must check it with the publicist or against the final version. Please contact USPR@bloomsbury.com with any questions. Thank you!