The first two weeks of August were a holiday reading fest, although mainly fiction. Now that I’ve climbed the email mountain waiting for me when we got home, I have time to reflect on the small number of books relevant to this blog.
Just as we left for sunny Wales, I finished Karen Hao’s Empire of AI. It’s a biography of OpenAI up to late 2024 by a journalist whose beat at the MIT Technology Review was AI. The subtitle, ‘Inside the reckless race for total domination’, sets the tone. The book is carefully reported – and none of the dramatis personae come out of it well. I did thoroughly enjoy reading it but didn’t find it at all surprising.
The other AI book was Leif Weatherby’s Language Machines. I’d spotted it via the wonderful Henry Farrell who has written two detailed blog posts – here and here – about why the book is marvellous. He also explained what it says, which was useful as I didn’t understand it. Courtesy of this explanatory post, the key point in the book is that: “LLMs work as they do thanks to a remarkably usable mapping between two systems: the system of human language as it has been used and developed, and the system of statistical summarizations spat out by a transformer architecture, leading to a “merging of different structural orders.”” This figures. For example, Weatherby states that Shannon’s information theory is an implicit theory of language, and I can see this is so as language is how humans convey information. I can also see a parallel between LLMs and Wittgensteinian views of language. However, the book requires much more understanding of linguistics and structuralism than I have, so there were literally sentences I could not make sense of at all. Some points jumped out clearly. “The political risk in AI … is located not in the systems themselves but in the lazy methodological individualism we too easily revert to in thinking about them.” I’m on board for agreeing about the metholdological individualism but actually am not sure there is no political risk in the systems themselves.
Away from technology, I read The CIA Book Club by Charlie English, a thoroughly good read about exactly what the title says – how the CIA supported financially people smuggling books into Poland and reprinting and distributing them there, through the 1980s and the Solidarnosc era. One lesson to bear in mind: it takes real courage to read and share ideas that do not conform to the dictates of an authoritarian regime. The book made me wonder – again – why the remaining western democratic powers, and their security services, so easily forgot this lesson of the Cold War, that the battle of ideas and for minds is real battle.
Finally, when we got back I finished a book I’d started before leaving, David McWilliams Money: A Story of Humanity. As you’d expect it is superbly well written. It is a straightforward history and so ideal for people who haven’t read much on this before. The bit that was really new to me was a chapter on Roger Casement and the Congo – I learned that Casement was a veteran surveyor with much experience in Africa who told Joseph Conrad about the colonial horrors he had seen, so informing Heart of Darkness. The experiences made Casement an ardent anti-colonialist, a journey which ultimately saw him executed for treason by the British government which had previously knighted him. The book ends with a quick canter through MMT, crypto and MPesa, and chapters on the control of money in modern economies and the psychology of money. The book covers a huge span in its 350 or so pages, and is an enjoyable read.
Of the fiction in the pile below, I highly recommend Attica Locke’s trilogy set in Texas – this is the third. I’m a big fan of Natalia Ginzburg and – in a completely different register – enjoy the Elly Griffiths series. But there were no absolute turkeys in this year’s reading.
