I’m writing this having recently returned home from the London Book Fair, where the theme of the moment – and, officially, of this year’s Fair – was AI, and what it will mean for the future of book publishing.

There were daily seminars on the subject from some of the industry’s biggest hitters – CEOs from the ‘Big Five’ (the five giant publishing conglomerates who dominate today’s global market), as well as key figures from the agenting world, other industry experts and delegates from the IT companies who are spearheading the development of AI for publishing.

Change is undeniably in the air, as even the configuration of the 2025 Fair seemed to confirm. There were noticeably fewer publishers exhibiting on the show floor, and a lower number than ever of small independent publishers displaying their wares. And this year the ‘Tech Centre’ was at least the same size as – if not bigger than – the IRC (the International Rights Centre, where literary agents and publishers gather to sell copyright licences for their biggest properties).  

Judging by the mood of the Fair, those working in the business of book publishing are (justifiably) far more apprehensive about the damage AI may do to the industry than they are excited about all the potential positives the new technology may bring (such as increased overall efficiency, greater streamlining of the process, reduced costs and shorter delivery times than ever before).

There is genuine and widespread trepidation that this latest iteration of the unstoppable engine of progress may mean the end of book publishing as we know it. And, as in most other sectors of human activity, there’s a recognition that as the technology advances, we’ll soon be in a place where fewer and fewer real people will be needed to manage and oversee the processes of editorial, design, production, distribution, marketing/sales – and now, thanks to AI software such as ChatGPT, the creation of content. (Who knows? We may get to a point where we can even dispense with human consumers – the ones who read the books – too…)

So, as a ghostwriter in today’s publishing environment, how do I feel about the arrival of AI in my professional world? Do I see it – as many publishers and other industry players do – as a credible threat to the future of my profession? Will software such as ChatGPT ultimately make the jobs of human ghostwriters like myself redundant?

My own feeling is that the ‘ostrich approach’ – burying my head in the sand in the hope that it’ll all somehow go away – won’t ultimately be very helpful. It seems better to try to face up to the prospect of inevitable change than to pretend it won’t happen, because at least this way you have a chance to weigh up your options. So for me, the most honest response to these questions is to acknowledge that radical changes are in the offing – and probably in the not-too-distant future, given how quickly technology in this field has already advanced. Though it’s very difficult (for me, anyway) to predict just how far-reaching the AI revolution will be, or the precise timeframe in which it’ll all happen.

I read a fascinating feature recently in which the journalist – a well-known and very successful columnist for one of the UK’s biggest newspapers – had agreed to allow ChatGPT to write an article under his name for a future column. This journalist had, as the best columnists do, a very distinctive style of writing, a highly opinionated take on the world, and an interesting back story to which he quite frequently referred in his regular pieces for this paper.

All of this was somehow fed into ChatGPT, and the end result was, by the columnist’s own admission… really quite astonishing. Not only had the programme managed to replicate his writing style really well – beyond this, it had come up with humour, word plays and cultural references very much typical of his usual approach – but which were at the same time completely original for this piece (i.e. without anything being borrowed verbatim from his already published work). He freely admitted that he was taken aback by the quality and plausibility of the ChatGPT-generated column! If he had any criticism to offer at all, it was only that in places, the ‘bot’ had gone to town with some of the humour and overly colourful language – that it had been a bit carried away with itself, so to speak. But this, he quickly conceded, was nothing that a good sub-editor couldn’t quickly and easily remedy.

The article in question was written several months ago – I’m certain that the chatbot has advanced even since then.  

However, my own feeling is (and I may of course be mistaken) that AI applications like ChatGPT aren’t quite up to providing a full premium ghostwriting service just yet. Doing a surprisingly good job of replicating the style and world view of a widely published professional writer in a 500-word opinion column is one thing. But producing an 80,000 word manuscript, on behalf of someone who doesn’t write habitually – and indeed will not be writing a word of text to be included – and has no experience of how to structure and present a full-length book so that it is both easily readable and highly compelling… That is, of course, a very different proposition.

For now, anyway, I can’t imagine a piece of software being able to interact with a client in a way which will bring out at every opportunity the most telling and relevant details of their life and the highly individual nature of their story. Where there is no existing written information provided by the client for the ghostwriter to draw on (and this is very often the case), the most effective, and in fact most feasible, way to gather the ‘data’ needed for the book is to engage that client in an intensive interview process – typically 25 – 40 hours in total.

Will a chatbot have the interactive capacities and interviewing skills to engage its human subject in this kind of process? Will it instinctively know when and how to pursue a line of inquiry, so as to encourage the other party to open up, to confide? Will a piece of technology be able to draw out the subject by recounting – and at just the right moment – similar life experiences of its own, thereby building the client’s trust with a reciprocal willingness to be vulnerable and completely honest?

Will an AI ghostwriter be capable of sustaining a two-way exchange – picking up all the nuances in the conversation; discerning underlying patterns the speaker may not be aware of ; taking into account any changes in their tone of voice in order to excavate other layers of the speaker’s meaning and intent?

Will the software know that the things an interviewee fails to mention, deliberately or otherwise, can often speak volumes beyond the words they have uttered? And that the subject’s body language, facial expressions and the kind of the energy they bring to the conversation are all crucial elements in the gathering of ‘data’ for the book?

There is so much about a human interaction that you can’t pin down to a formula or summarise in standard form, just as the best stories usually end up being so much bigger than the sum of their individual sentences. And – until this point anyway – the best, and most successful interactions between a client and ghostwriter are always a collaboration – a two-way rather than a one-way street – where the alchemy of the human connection adds something magical to the mix which, if the writer is good enough, will emerge in the words on the page.

So in the light of all of the above, ChatGPT and its other iterations may not be putting the best human ghostwriters out of a job just yet. But that’s not to say we won’t be given a good run for our money at some future point – and possibly sooner than I think.

For while I personally believe that in almost all meaningful scenarios, an interaction with a real person is preferable to one with a piece of technology, I’m very much aware that there are a lot of people in today’s world who wouldn’t necessarily agree with me.

Already, a staggering proportion of the population across the globe are spending countless hours of their daily lives interacting with electronic devices – often more intensely and certainly with more attention than they are with the people around them, including those with whom they supposedly have their closest and most important relationships! (I’m not of course immune to the lure of online browsing myself – virtual window shopping is my Achilles heel. . .)

But in all of this – and even in the mooted scenario where AI takes over the world and makes 99 per cent of the human race obsolete – I continue to think that we have some kind of agency, and choices, for the time being anyway. Realistically, my feeling is that in the immediate future, we humans will still be needed for some parts of the ghosting and writing process of producing a book – for the gathering and shaping of the ‘data’ at the very least. The obvious divide between a quality ghostwritten book and a hastily generated AI-only one will emerge even more definitively, at least for a certain window of time – and it will continue to be the case that anyone who wants to put a work of quality and longevity out into the world, something they can be proud of, will still have to invest time and money in the process.

But now more than ever before, we need to continue to take control of our own narrative – before AI is able to do that for us too.