The case against AI: writing isn’t meant to be easy

If artificial intelligence excels at anything, it is generating controversy – not least in the world of books. Over the past week or two, it’s already provoked two rather embarrassing scandals. First, Olga Tokarczuk, the Polish winner of the 2018 Nobel Prize for Literature, admitted that she uses AI when she needs help developing ideas. Then, Granta magazine, no less, was revealed to have published a prize-winning story which appears to have been written entirely by machine. Who knows what will be next.

Such scandals are nothing new. Since the launch of ChatGPT in 2022, there has been plenty of controversy about AI’s use in literature – and no shortage of hand-wringing about the fate of writers. Back in March, Hachette cancelled the publication of Mia Ballad’s novel Shy Girl after its author was accused of leaning a bit too heavily on AI. A few weeks later, the New York Times dropped Alex Preston for using the same trick to write a review. And before the dust had a chance to settle, the New Yorker went on to ask ‘Is it wrong to write a book with AI?’

These latest scandals seem unusually poignant, if only because of who is involved. We’re not talking about a hard-pressed writer scrambling to make a deadline, or a self-published author hoping to make it big. This time, it’s a Nobel laureate and the most self-consciously highbrow literary magazine out there. Reading the reactions online, you’d think that Olympus itself had fallen.

Outrage has centred on a series of rather shrill questions. Why did the Commonwealth Prize jury not spot that Jazmir Nazir’s story was probably AI-produced? Why did Granta not pick up on it either? As the magazine has been at pains to stress, its agreement with the Prize obliges the editors to publish the story; but how could they have missed the slew of tell-tale tics in the first paragraph alone? Couldn’t they see that the story was bad? Why, if these charges are true, did Nazir do it? More to the point, why does Tokarczuk? She clearly has enough talent, so why does she feel the need to ‘cheat’? Should she be blacklisted? Does this mark the end of ‘proper’ literature? Or should we just accept that books are going to be slop from now on?

There’s a lot to unpack here, but underlying these questions are a couple of fairly basic assumptions. First, that AI produces obviously bad-quality writing – and should therefore have no place either in literary magazines or publishing. Second, that literature is the product of an author’s own effort and experience – anything that isn’t is unworthy of serious consideration.

Each of these is perfectly respectable. I suspect that, if pushed, most of us would instinctively agree with them. But I rather think they miss the point.

AI-produced slop really is dreadful, but AI didn’t invent bad writing. It’s been around for as long as the written word itself. A little under 2,000 years ago, the Roman satirist Juvenal claimed that it was torture to listen to a certain Cordus recite his work. In the 1470s, Niccolò Perotti thought there were so many bad books being published in Renaissance Florence that it would have been better had printing never been invented. And in the mid-20th century, J.R.R. Tolkien and his friends held competitions to see who could read Amanda McKittrick Ros’ novel Irene Iddesleigh (1897) out loud for the longest without laughing.

Bad writing has never had much difficulty finding its way into print, either. While most magazines and publishers are generally highly attuned to quality, what matters most is the bottom line. If something sells, that’s all that counts – even if it’s objectively ‘bad’. Like it or not, bad books do sell. Just look at the best-seller lists. Almost every week, easily marketable pulp comes out on top, while more polished works end up in the remainders bin. It’s such a depressingly familiar situation that Balzac even satirised it in Les Illusions Perdues (1837-43).

What makes writing ‘good’ is a tricky question. I don’t think I could pin it down with any ease. If pressed, I’d probably have to say – pace Kenneth Clark – that ‘I think I can recognise it when I see it.’ But I don’t think you need a particularly clear sense of what a ‘good’ book looks like to ask whether you can really distinguish between a ‘bad’ book written by a human being and a ‘bad’ book written using AI – qualitatively speaking, at least. Are there different types of badness, just as there are different types of infinity? Either way, bad is… bad.

Whether literature should be an author’s own work is another matter. On the surface, the answer seems self-evident. When we open a tin of soup, we expect it to contain what it says on the wrapper – so it’s only reasonable to expect that a book should be by the person whose name is on the cover, too. And only them. Originality seems to be almost a sine qua non. But how robust is this? While plagiarism is undeniably a form of intellectual theft, it’s not so clear that borrowing is quite that bad.

All authors are influenced – consciously or unconsciously – by what they have read or seen before. In fact, this has often been seen as one of the defining features of literary creation. Seneca the Younger argued that, just as bees make honey and wax from the pollen of flowers they leave behind, so writers should craft new works from the materials they encounter in their reading (Ep. 84.3-5). As T.S. Eliot wryly put it, ‘immature poets imitate, mature poets steal; bad poets deface what they take, and good poets make it into something better’. This being so, can we really draw such a line between literary borrowing and AI-powered appropriation? It seems painfully difficult.

So no, I don’t think the outrage that has been generated by these latest scandals is really justified. It doesn’t change much if bad AI writing wins prizes or gets published. Nor does it seem to harm ‘literature’ that seriously if an author wants to use AI for inspiration. Put bluntly, there’s plenty of other rubbish already out there.

Yet I still think these scandals should be a wake-up call. Despite everything, I strongly believe that the encroachment of AI into literature is catastrophic for writers. Not because of how bad AI’s outputs can be, or how unoriginal it might make us – but because of what it does to the practice of writing. How it harms our relationship with our work.

It strikes me that, if you are going to be a writer, then the process is as important as the product – if not more so. That, after all, is what defines the writer. How you get the words down on the page, not that you get them down. And if you’re going to get anything out at all of writing, I don’t think it should be easy.

I can only speak for myself, and I don’t claim to have any particular authority. But the best way I can think of explaining what I mean is by comparison.

Whenever I sit down in front of a blank page, pen in hand – or, more often, fingers hovering over the keyboard – I feel as if I am looking out of the window at a howling storm. Inside, it is warm and cosy. Everything is comfortable and reassuring. I don’t want to go outside. No matter how much I might need to get started, there’s a part of me that just wants to curl up on the sofa and forget all about it.

When I eventually force myself out of the door – that is, when I force myself to start writing – it’s usually tough going. The rain lashes my face. A trickle of cold fear runs down the back of my neck. The fog is so thick I can’t even see where I’m going. And sometimes, I’ll be honest, I do give up. The mud is too thick. My feet get stuck in the mire. It’s hopeless.

But if I put my head down and push on, the storm gradually seems… not more bearable, but… different. Struggling against the terrible inertia of the words starts to feel less a labour to be shirked than a challenge to be met and relished. Out of the darkness loom unexpected landscapes of unimaginable vastness and wonder; and fear itself changes first to curiosity, then to adventure. Even the crash of thunder – the threat of failure – evokes a thrill of fragile excitement. Then, when the journey is finally done, and the last flourish of the pen is complete, it feels like I have really done something. I don’t say that every journey is perfect. I usually fall down a few times. There’s mud on my knees and twigs in my hair. More often than not there have been tears along the way. But I’ve arrived. I’ve made it. And I feel alive.

I hate it sometimes, but I love it, too. I couldn’t do without it. I like that it’s hard. Sure, if you’re going out in a storm, it’s much easier to take the car – just as when you’re writing, it’s much simpler to use AI. I can see that. It might win you prizes, or get you published that much more easily. But really – what’s the point? Where’s the wonder? Where’s the excitement? Where’s the humanity? What do you learn about yourself? If you’re going to do that, why not just stay at home and do something else?

As for me, I’m off out again. Come along, if you like. Grab your coat. Pick up a pen. It’s terrible out there – and I promise, it’s worth every moment.