Using genAI to read books

“Students … are stressed and overwhelmed … a reading list can start to feel like a mission they’re failing” (Tom Chatfield).  

I know the feeling. When I was a well-meaning but innocent young student, my response to reading lists was to start with the first book on the list, and once I had read the book in its entirety, I would then start on the second book … bad method, if only because I never even started the books at the bottom of the list (that must be some kind of bias, let’s say “top of the list bias”). Today, in the age of AI, we can do much better than that. Today, reading lists handed out to students are, I hope, better presented, with annotation, but with or without guidance, the challenge of getting to grips with new content remains.

Here is a tested way of dealing with a reading list that puts you in the driving seat without having read any of the titles … it sounds like cheating, but it will make you a more effective reader in the long run. Here is a digital-age solution to the reading list problem. It is designed for books, but would work in a similar way for other content, such as journal articles (although there is today a wide range of tools for speeding up the discovery process for articles). I call this method “the context method”, or, “what’s this book about?”

Let’s be realistic – you won’t read more than one or two of the books on a reading list in full. You might not even open some of them. But it doesn’t matter! The books have been included for some reason, not because they all offer the same experience, but (I would imagine) because they offer complementary angles to the subject you are studying.

First, a complaint about reading lists. The reading lists I was given at university were simply a list of authors and titles, with no description whatever. But let’s start with that, and see where it takes us. The actual example was The Italian Renaissance. I was studying English literature, and we were given a one-page reading list for some background reading. It comprised around ten or fifteen titles – quite typical for an academic list. It was, I think, in alphabetical order, and one of the first titles was Jacob Burckhardt, The Civilization of the Renaissance in Italy (1860).

1. Ask genAI: “Please give me a one-sentence summary of this book”. This summary will be your starting point. I used Copilot, out of convenience, but any of the large-scale LLMs will do fine. It told me Burckhardt’s Civilization of the Renaissance is about “the birth of modern individualism and secular culture”. At this point, you don’t want to agree or disagree with the books, simply to know what they claim. In the case of a work of literature, it’s not the summary you want, but some explanation of why it is felt to be so significant.
2. Try the prompt “Is [title] still read today? Why?” [Burckhard helped establish cultural history as a discipline … it forms a landmark in historiography]
3. Make a note of when the book was published. A book published in the 19th-century will be included because it has already established its position in the canon. A book published last year will still be the focus of much discussion.
4. Find a couple of contrary views. If the book (like Burckhardt) was published in the 19th century, there will be plenty of critiques of it. Ask genAI “Why is [book title] considered to be out of date today?
5. You now have your reading list with two bullet points for each book.

For each title, you should now have something like:

Title one

• why it is considered significant
• why it is considered superseded

Optionally, for a deeper engagement with each book, you could search for any reviews of that book – not reviews on Amazon or Goodreads, both about as useless as Trustpilot, but reviews with some authority. But this step is very much an “if you have time” option.

Using the context method, you have framed your reading list before you even begin. If your reading list was well curated, you could hope that you now have a view of the critical context for your subject: who are the major figures in the critical debate, how it has changed over the last 50 years, and so on. This echoes the approach of Pierre Fayard, whose How to Talk about Books you haven’t read makes it clear that even professional academics like Fayard read about books more than they read them. He describes the summary above as “the inner book”, the preordained ideas, the general cultural preconceptions, that surround any book. I’m not quite sure if he saw this as an advantage or disadvantage, but for me, the context of a book (what others think of it) is as important as the book itself. After all, some books, such as Newton’s Principia, will never be read, but it doesn’t mean you should know about what they say.  

The advantages of this method:

• This methodology provides an “onion-skin” approach. By all means read all the books in full, if you wish; nothing in this methodology prevents that. But if you stop at any point, you have a clear grasp of what the book is about. A later, more detailed reading may modify that first view, but the context view will remain valuable.
• It works for a reading list, or even without a reading list, if you are just discovering a new area of study by yourself. Ask genAI for “a non-specialist introduction to [my subject]”.
• When you start reading the books themselves, you are more likely to be engaging in a critical debate with the author.
• It uses genAI to do what genAI does well: to summarise an existing body of knowledge. At this stage, you aren’t interested in any new interpretations of the subject – that can come later. All you want to know is where this book fits in the big scheme of things. Is it a description of the current consensus? Is it superseded but pathbreaking because it raises questions that are still valid today?
• You have now gained a new title in the “collective library” (to use Pierre Fayard’s term). The collective library is “the set of books in a particular culture, which enables you to place that book in a cultural context, for example, Madame Bovary resembles many other 19th-century novels describing the heroine’s adultery.

Can this method be used universally?

• Textbooks are different: they have a semi-mythic status. The approved text is sanctioned by the teacher as the definitive guide. This makes it difficult to engage in critical reading. Textbooks, which are secondary, rather than primary texts, tend not to be critiqued in the same way as monographs. Awarding text-book status to one title has the downside of removing critical reading. Good textbooks should provide suggestions for further reading, and these suggestions can be used for the appraisal method described above,
• If you are studying literature (as opposed to just reading for pleasure) you can use genAI in a similar way. However, genAI tends to summarise the plot rather than to explain the significance. You might need to rephrase your prompt slightly to get this kind of insight: Hamlet, for example, uses soliloquies to explore mortality and uncertainty.
• For articles, there are several ways to assess importance

I have suggestions for how subsequently to read the book itself, but that will be for another post.