The literary world needs to wake up!
The Commonwealth Prize
The response from the publisher of Granta, Sigrid Rausing, has essentially been to shrug and say, “Who knows?” We asked Claude, it wasn’t sure, so whatareyagunnado. Anyway maybe it’s racism. A separate statement, this one from the Commonwealth Foundation, said, confusingly, that they would never feed stories to an AI model because then the evil AI would steal the content. Sam Leith, literary editor of The Spectator, agrees with Rausing. Rausing is one of our greatest philanthropists and Leith one of our best critics. I find their attitudes bewilderingly passive.
From Ian Leslie, a very nice piece. I wasn’t going to write about the AI story that won the Commonwealth Prize. I wasn’t surprised by what happened. And I have written before everything I would have said. In short: this is what happens when you ignore AI and get your opinions from the newspaper rather than using the models yourself. Asking Claude if the story was AI generated is a giveaway that the people involved don’t quite know what they are doing. Some literary people are at risk of circling around themselves, maintaining a sense of political and moral acceptability.
Obviously that is naive and this prize is only the start of what will happen. But it is also unliterary behaviour! The world is changing and even if you hate the changes and think them wicked that is all the more reason to have some understanding of what is going on. It’s like opposing the internet without ever logging on. Good luck! Perhaps it is a principled stance, and for certain individuals it is surely the right one, but it is not possible to simply avoid AI anymore. More and more submissions will be AI generated. More award committees will be fooled. And more uses of phrases like “AI plagiarism” will be used in the hope that a moral attitude will be insulating against the real world.
Ian repeats one of the platitudes that has become common wisdom recently:
A story that isn’t written by a human has no purpose, and can only offer paltry satisfactions to a reader who knows this to be so.
Maybe! But we are running that experiment and I don’t think we should be so sure. Yes there is a study, but that is based on current AI, not the capabilities of the future. One day, perhaps soon, we will not be able to tell what is written with AI, and some of it might be so good that we will just have to live with the uncertainty. If a human creates a story using AI and that story is never “revealed” as being part AI, or if it is simply too hard to be sure, then we might have to revise our ideas. If the writing is good enough, some people will get over the fact that AI was used. The future is high variation!
When he died, Larkin was discovered to be a monster, and some people said they couldn’t read his poetry anymore. That didn’t last. We know that many famous lines and phrases we love in older poems were “plagiarised” (great artists steal, and all that) and it doesn’t diminish our pleasure. Think of all the readers who didn’t notice or care that the recently cancelled horror novel Shy Girl was AI written, or part AI written.
Right now, we have AI writing that all sounds the same. What happens when a great writer trains their own model? Or when they use a series of prompts that guide the LLM to a new sort of intertextual writing? Or when AI companies train their models to be better writers? Or when a young person no-one has yet heard of emerges with a strange new sort of writing based on their years of AI prompting since their childhood?
It is so easy to say that stories have to be human written. Just like it is easy to say “people love human generated slop too!” But those phrases are also—though not always—part of the “currently sensible and acceptable doctrine”. The really sensible thing is to wait and see. While the literati debate the future without opening an LLM, people like Gwern are writing the future.
Thanks for the article; some quick, non-exhaustive thoughts.
The electric guitar, especially when run through various effects layers, creates a machine-assisted sound, but the strumming and picking are still at root human. Jimi Hendrix could still play (bigly) at the bottom of the distortion.
The problem with AI "creative writing" is that the machine is doing the work (such as it is) and the "writer" is just inserting prompts or some sloppy draft that the AI "fixes" and (at present) processes into Velveeta.
You are correct that AI's capabilities will probably improve, yet when struggle and craft are avoided by would-be writers, they fail to be writers. Having AI finish one's stories and poems is like having the spotter finish the hoisting the weights at the gym. To claim that one has bench-pressed 200 lbs when the spotter saved one's neck and lifted 110 lbs worth is not to have lifted 200 lbs, and to remain weak.
The struggle and effort are the whole point--certainly for writers, perhaps less for readers; yet without writers creating the valid human work, the readers downstream are spiritually undernourished. (I'm talking here about more serious literature, not Hollywood pulpy junk, which has been machine-made for decades: a bean-counting script committee is not much different from AI.
Some thoughts:
(1) As someone who loves the early Naipaul novels (both Sir Vidia and Shiva) I would normally be thrilled to see a Trinidadian writer getting recognition.
(2) The story is absolutely ripe and rich with the deep Trini aura and feel that so many writers have explored.
(3) Even if its not AI, so much of the writing is overwrought and bad with the metaphors - “Her hair is midnight rain; her laugh is bright as zinc” / “Doing is a treacherous bridge: you step on and it carries you to a side you didn’t plan to reach.” / “Sita kept a cheap copybook with a red line that was less a margin than a joke.”
(4) The writing itself should have made judges reconsider if it deserved a merit award.
(5) If it is AI, then AI is capable of 'vibe pastiche' whilst being unable to actually distinguish what is good writing within that pastiche
(6) However, what if one day, to use the Trinidadian example, a writer emerges who in their early work is influenced by the Caribbean canon and so their writing resembles it? How do we distinguish between influence and people claiming they are AI enabled? If AI just reconstitues what it is scraped? That is worrying.
(7) At some point, writers who have not used AI will be accused of AI. This will lead to paranoia and witch hunts.
(8) I'm worried.
(9) Is this a dividing line between eras? Pre AI and Post AI? If all AI does is pastiche what was written before then perhaps literature from the Pre AI era will become viewed as a pristine canon that was not suspect.
(10) How will new voices emerge if they will be suspected of not being the voice of a true literary consciousness and soul?
(11) Olga Tokarczuk - Nobel laureate - has spoken of how she uses AI as a collaborative tool in her writing in some ways. Seems like she's saying it should be harnessed in some ways. But this makes us ask - a hybrid literature where human creativity is abetted by AI response to ideas etc. Is that literature.
(12) I dont know. But the dam has burst