ChatGPT may learn to write good stories... but will it be allowed to tell them? by Griselda Heppel

ChatGPT in action. Photo by Matheus Bertelli: https://www.pexels.com/ photo/laptop-office-working-internet-16094045/
Katherine Roberts’s experiment with ChatGPT got me thinking. As yet this AI writing tool can in seconds produce reams of prose, matching the prompts given, in a way that no one, adult or child, will ever want to read. So far, so good. (Or satisfyingly bad.)


But any complacency on the part of genuine flesh and blood authors is misplaced. As Katherine and others have pointed out, this bot learns quickly. We can’t rule out that, sooner or later, given enough original material from an inventive editor, say, it will be able to create a gripping, heart-warming tale with all the right kind of characters, and dialogue that doesn’t make your ears bleed. Something actually readable. Publishing houses will save heaps of dosh by eliminating authors altogether from the equation and we writers will, er, be written out of history. 

An alarming prospect, and not just for authors. However much ChatGPT improves, I find it hard to believe it will ever generate anything original or startling. Future generations are destined to be shortchanged by a stream of rehashes of the kind of stories and themes already out there. The most gripping, heartwarming tale will cease to grip or heartwarm as readers tire of the same elements presented in multitudinous different ways. I mean, how can a bot ever produce that fresh, original voice so desirable to an editor looking for new talent? 

But another aspect of this Disappearing The Author malarkey strikes me. As we know, it’s not enough nowadays for YA and children’s books to provide a cracking story with appealing characters. They must tick a number of other boxes as well. Characters - most importantly the hero - have to be from diverse backgrounds to reflect the richness of human experience, covering the spectrum of ethnicity, class, neurodiversity, identity, health, differently abledness etc. 

But - and here’s the conundrum no one as yet has been able to solve - if you, the author, don’t share your hero’s background, but have got there by research and imagination, you can stop right now. You are not allowed to write this book. It is not your story to tell. Since no one can live more than one life, it narrows down what an author is expected to write about to a very small canvas indeed. Which will not be published anyway because it isn’t diverse enough. QED.

American Dirt by Jeanine Cummins
A couple of years ago, a vast Twitter campaign attempted to prevent a novel centring the Mexican immigrant experience from being published, on the grounds that the author, Jeanine Cummins, wasn’t Mexican. In this case, the publishers took a commercial view, went ahead anyway and American Dirt became an instant bestseller. But in many other cases, online fury against an author achieves its aim and books are withdrawn. 

What then, if there is no author anymore? Rather hard to demand a bot be authentic to its ‘lived experience’ since by definition it doesn’t live and has no experience, only technique. 

So who will take responsibility for 1. Producing diverse books and 2. Ensuring their diversity matches their creators? Will it fall to the editorial staff in publishing houses to divide the story ideas among themselves, each feeding ChatGPT the criteria that matches their own background and experience, and if so, how will anyone know? I mean, how can we be sure that the story told in a ChatGPT created book is a story that particular book is allowed to tell? 

 Heavens. Here’s a whole new aspect of ChatGPT to worry about. 



OUT NOW 
The Fall of a Sparrow by Griselda Heppel
BRONZE WINNER in the Wishing Shelf Awards 2021 
By the author of Ante's Inferno  
WINNER of the People's Book Prize

Comments

Interesting! I wasn't aware of books being pulled from publication because the author doesn't have a suitable background - surely that is wrong in the first place? The author could easily have interviewed people with the right background and done enough research to make the story feel real, in the same way a fantasy/sf author makes their fictional world 'real' without ever having lived in a fantasy world or on an alien planet.

An AI like ChatGPT (being a robot) obviously has no life experience, BUT it can learn how to live millions of times over from us... millions of different lives... so theoretically could have the 'experience' to write any kind of book?

However, I still think AI can never be original or surprising/exciting in the way a new author's voice can be, because it hasn't come across that new voice yet... however, after enough of a learning process, it could probably be used to write decent enough sequels to original stories written by real authors, which (as we know) publishers love. It is definitely a worrying time for authors!
I always wonder what people who implement that kind of 'rule' make of historical novelists, who clearly can't have the background of someone who lived in the 17th century or whenever.
Umberto Tosi said…
I'm with you. The write-only-what's-safe-inoffensive-and-about-yourself crowd diminishes writers as much or more than the prospects of smart-Chat-GPI scripting... We don't need the first and we need roles for the second.
Griselda Heppel said…
Thank you all. It's ridiculous, isn't it? I'm sure the reason a huge proportion of children's books nowadays are either fantasy novels, or set in an anachronistic historical period (eg 19th century but with robots... or jousting), is so that the authors can't be accused of cultural appropriation. At the Oxford Literary Festival a couple of years ago, the established writer Candy Gourlay felt she had to justify setting her marvellous historical novel, Bone Talk, among a particular Filipino tribe - even though she herself was from the Philippines, just not from that tribe! I wonder if soon painters will only be allowed to paint the country they grew up in and not travel to Tuscany, say, or Rajasthan to paint the colours there.

Oh yes, Katherine, ChatGPT could make hay with sequels, simply piggy backing on an author's original story. Not a welcome prospect.
Peter Leyland said…
Just caught up with this now Griselda, having finally emerged from my Vietnam trip. 'Cancel culture' as I believe it's called when a piece of work doesn't suit the norm was a topic at the recent Trondheim conference where we discussed how Joseph Conrad could write a successful book (Heart of Darkness) that he did not have the experience of. I will now look to 'Apocalypse Now', the film of the book? in search of ideas for my next blog.
Just read this, Griselda, as I try to assemble some thoughts while feeling rather 'browned off' about 'writing'! It is lovely! I so enjoyed both the blog and its comments. The very udea of these wretched 'bots' is indeed worrying, as worrying as all the 'cancel culture'fuss.

It's also just so negative, so nit-picking, and so reminds me of the know-alls at school who bullied, teased, and made simply being at school appalling at times!So glad you decided to tackle the bots and the cancellers head-on! I am certainly NOT a robot!
Griselda Heppel said…
Thank you so much, Peter and Clare. Glad you enjoyed this! Let's see how the whole bot/cancel culture intersection works out. It might be amusing to watch.