Publishing has an AI problem

In late March, the New York Times dispensed with the services of one of their freelance book reviewers. Normally, such things go without any publicity or wider attention, but the rationale in this case was an unusually specific yet potentially wide-ranging one, not least because it taps into a growing sense that AI is not just coming for writers’ livelihoods, but that it has already taken up position in our sitting rooms and studies, and is laying waste to all around it.  

The well-respected novelist and critic Alex Preston had filed a review of the novel Watching Over Her for the newspaper in January, which an investigation revealed had plagiarised another review of the same book in the Guardian. When Preston was asked what had happened, he confessed to having used an AI tool to create his piece, which in turn had drawn on the existing critique without his knowledge. He was then fired and, thanks to an investigation by The Wrap, his name — and actions — were made public. Cue chaos. 

Full disclosure: while I am not a close friend of Alex Preston, he is someone who I’ve always liked and admired, and, I believed, a force for good in the literary industry, although I note that his byline has never appeared in this publication. I was also lucky enough to have had my book The Windsors at War reviewed glowingly by him in The Observer a few years ago, so feel a professional debt to him. Therefore my reaction — as well as that of many other members of the Anglo-American literary circle — was that of near-disbelief. 

As Sam Leith, the Spectator’s books editor, wrote on his Substack, “Why would a critic of his calibre want or need to use AI to help with a review? And why — lord, why? — would he risk doing so in a space as high-profile as the New York Times? This is one of the most prestigious book-reviewing spaces in the English-speaking world. It’s a showcase — and they pay.” 

Preston supplied an explanation to Leith — who, again in the interests of disclosure, calls himself “a friendly acquaintance” of the writer, and who has commissioned him to write for the Spectator — in which he said that “I had written a draft review of the book, but it was under length and I was rushing badly and drowning slightly. I made the stupid decision to use an AI tool to help expand and smooth it… it was absolutely an improper use of A.I., and a complete failure of judgment on my part. I was ashamed when it was put to me, and I admitted it straight away. It was the case of someone naively and clumsily using a tool they didn’t understand and this is where we are.”

Virtually every writer friend of mine has an opinion about the issue. Some (like me) are broadly sympathetic towards Preston, both in the knowledge that few of us are beyond reproach and that becoming a figure of international intrigue and discussion for the wrong reasons is a horrible experience. Others, perhaps spurred on by jealousy or high principle, are far more condemnatory, suggesting that his deep dishonesty must be a sign of his failings as a literary critic. While I would suggest, in passing, that reviewing books for a living — if it can still be called that — is hardly the same thing as brain surgery or rocket science, it is certainly the case that AI — which, as Preston openly admits, is something that he doesn’t fully understand — is increasingly gobbling up everything around it.

It currently seems likely that AI will lay waste to the literary industry in all its forms. AI-created novels already exist, and Mia Ballard’s Shy Girl was pulled from bookshelves last month over her alleged use of Chat GPT or other AI software in either the book’s editing or in its creation. Ballard denies that her book is AI-generated, but the suspicions that it was were sufficiently widespread for her publisher Hachette to announce, indignantly, that it “remains committed to protecting original creative expression and storytelling” and that “Orbit (in the US) will not publish Shy Girl, and Wildfire (in the UK) have decided to no longer continue publishing their edition.”

Scapegoating one critic or novelist, however misguided their actions may have been, seems unwise

While it seems unlikely that Shy Girl will be troubling the review pages of The Critic any time soon, it is inevitable that, sooner or later, AI will move into non-fiction, and do so efficiently. As someone who spent two years writing a biography of David Bowie and taking the effort to research it, it is galling to think that a powerful AI service will, before long, be able to produce a similar book in moments, and then not charge for the privilege. It won’t be as good, I hope, but faced with a choice between paying £25 for a book and receiving the same information gratis, many will opt for the latter, and so publishers will find it harder to make money. Which, in turn, means that fewer books will be commissioned, fewer authors will be able to make a living, and the industry will founder. 

Perhaps in response to this, some cynical and misguided “writers” are even using AI wholesale to craft pitches and articles. Any editor who has been in the business for more than five minutes will be wise to the tricks and tells of such techniques — the odd, often stilted structure, the endless use of em dashes and the tendency to make statements and conclusions go in threes — but it seems as if fake news has been superseded by fake articles, and the primacy of the written word will suffer in the process. Such is the fear in which AI is justifiably held. Scapegoating one critic or novelist, however misguided their actions may have been, seems unwise. Instead, these stories demonstrate the growing malignity of a service that appears to be helpful and sympathetic, but instead lays waste to all around it. Look on my works, ye not-so-mighty, and despair.