His publication list is now commendable. This new book, How to Think about AI: A Guide for the Perplexed, explores the profound impact of artificial intelligence (AI) on humanity, addressing both its promises and risks.

The book has printed in the initial pages a dozen positive reviews, all stressing its importance, from a former prime minister and others, including several professors of computing science.

It is reassuring, and perhaps a mark of things to come, that Professor Susskind asserts from the outset that none of his book was written by AI.

The aim of the work is to provide readers with various ways of thinking about AI, a term that is defined as “how it works”, or alternatively, “what it can do”. The very short history of AI is highly instructive in that regard for those who know little or nothing of the subject matter.

Beginning with his personal history and decades of practical experience in AI, Professor Susskind acknowledges the rapid advancements in the field, exemplified by newer systems like ChatGPT, and the challenges they pose to public debate.

His work is in five parts: (1) understanding AI; (2) encouraging different ways of thinking; (3) practical applications; (4) addressing risks; and, (5) contemplating future possibilities, including the prospect of “conscious machines”.

The argument overall stresses the importance of balancing optimism with caution, urging that AI be discussed not only by technologists but by all sectors of society. In the book, AI is centred as the defining challenge of this age, with suggestions of reshaping industries, professions, and even human identity itself. There is plenty to reflect on. Not least of the specific novel issues is that of “not-us thinking”. The commonplace idea often heard that new machines may assist other professions, but not those of any given profession invited to consider it.

In that respect, Professor Susskind is alive to the approach by lawyers, particularly in relation to court work: “no robot could stand up and plead in court”. Yet, online courts, forced into place by recent necessity, require a different set of tasks.

There are undoubted risks in AI, with “tomes already written” on them. A classification of risks is proposed by Professor Susskind and that aspect should inform the looming decisions. Unlike an Act of Parliament, there is no appointed day for the universal introduction of AI: the technical advances supporting its development are exponential, universal, and subtle.

It is likely that there is ample uncontroversial material labelled ‘AG generated’ discoverable on the internet now, but such a label might not always be there. On that point, the BBC recently showed a short item on a news programme. A young woman spoke about the problems of climate change in impeccable English, and the viewer was advised at the end that the production was by AI, and the presenter fictitious.

This discursive book is an excellent summary of what seems to be a vast literature and presents the major issues in a helpful and instructive manner that is certainly not patronising. The modest size is illusory, first, as it encompasses explanations of major and critical issues that are pressing now, and not theoretical matters.

How to Think About AI: A Guide for the Perplexed by Richard Susskind. Published by Oxford University Press, pp. 224. £10.99.