There is a lot of guilt around using Artificial Intelligence for reading and writing. Whether it is justified completely depends on the situation. A certainty is that more and more books disappoint me. It is apparent that they have been written with the help of AI (more like with the help of the author) – which is perfectly fine – but when the whole thing sounds like AI every three paragraphs, that is not enjoyable. It communicates indifference and warns me of possible hallucinations. Recent disappointments have been caused by two big names Morgan Housel and Ryan Holiday. I was looking forward to reading their new books and lost motivation once I realised how much AI was used. And I get it – intellectual delegation might be the only way they get a holiday.

I apologize if the editors are the ones to blame. Even though the press talks about how editors are struggling to humanize AI-generated content by authors, the exact opposite could also be taking place. Perhaps the editors are editing the books with AI precisely to make it more universally readable. People use AI so much they may feel more comfortable reading AI generated text. Due to this, they tend to be paraphrasing with AI, asking it to simplify text they would normally understand. Everything looks clearer in bullet point logic after all. It may be a secret bet in publishing that AI-sounding books will be preferred by that wider audience. Non-casual readers will hate this, but what is their alternative if all the big names use this system? Does it also mean that this audience will search more thoroughly for relatively unknown writers who are self-publishing? This is, assuming, again, that editors actually encourage AI use contrary to public common sense.

If a book has been written with AI, why not read it with AI? 

A book that has been written with AI is a piece of mud with some golden nuggets inside. An enjoyable reading experience is traded for an exhausting search for something meaningful. The richness of the language is essentially substituted with bullet point logic. Should the reader then spend their time managing the muddy slop? Just as AI emails deserve being replied with AI emails, similarly AI generated books deserve to be written by AI.   

How do we know it’s mud around gold? 

To understand the composition of a book, reading a few pages is enough as AI slop is immediately apparent; some experience using AI in daily work or research is enough. If the writing style is unique, on the other hand, you will also be able to pick this up. Instead of investing in a pointless introduction, go straight to the middle of the book and assess. Then consider the following: is it worth reading 350 pages for 3 valuable pages? Wouldn’t you use Ctrl + F in a PDF anyways if you were doing research before AI? If so, it is ethical, practical and valuable to use AI to find those pages immediately. Remember that garbage was written way before AI was even thought of.

Reading time as opportunity cost 

This frees up your time to invest in books worth reading. You should, then, aim to read ‘the Αncients’. In Slay the Spire 2, a new early access game I enjoy, Darv is an Ancient entity that brings you relics, items that power you up, from the original Slay the Spire. Similarly, when I say the Ancients in real life, I mean all those influential scientists, philosophers, personalities and thinkers who stand as the giant shoulders of today’s achievements and realisations. Luckily, also, there is enough bibliography written before 2022, the year ChatGPT emerged to cover a lifetime of reading needs and preferences. As Nassim Nicholas Taleb suggests in ‘Antifragile’, the older a book or a text is, the more valuable it becomes. At the same time, irrespective of AI, a large amount of books simply explore the same classic ideas over and over again. Just open a bestselling economics book and you will realize that the book spends ⅓ of its pages summarizing Smith, Ricardo, Malthus, Mill, Marshall, Keynes, Hayek, Friedman – you get the point. But if you have already read these ‘ancients’ you can skip these parts and get to the point, the constructive comments of the new author. And, why not, even use AI to identify up to which point the summary of the ancients lasts and where the fresh perspective begins

The ancients spent a lot of time in their thoughts

The ancients, perhaps limited by the lack of Ctr+Z, version control, editors and  the available quantity of actual paper or parchment they can write on, seemed to have put more thought in their language and words their use. To say this in a different way, their words are condensed and hide a lot of meaning. This is why ancient text is accompanied by notes and commentaries that try to connect the content of the writing with the context. Naturally, we think that context is needed because there is no journalistic reportage of antiquity or renaissance – at least by modern standards. But context, really, is as necessary today. What is the context we use to understand a modern text in yet another age of misinformation and disinformation? And isn’t reading through AI another ‘commentary’ that paints a context of its own? Then, maybe it’s better to have AI simply extract the data and leave the knowledge to us and the ancients.