Will AI Destroy Great Writing? 

Is a still image generated by artificial intelligence a photograph, a snapshot of what the AI tool ‘sees’ online, or is it something else entirely? Can the copyright of a film script drafted by AI be owned by the human who prompted it? Will graphic artists and creative writers soon find their labour is no longer a necessary part of our creative industries? Are AI tools just that – tools, some of which can be used to aid creativity, or something more worrying, another step toward the creeping corporate takeover of our lives?

Back in June at a Creative Bath event, I sat on a bench hoping it wouldn’t rain and talked to Richard Godfrey. Richard is one of the founders of Rocketmakers, a tech company that develops bespoke software solutions for companies big and small. A registered B Corp*, they have their own pro-bono initiative, the Rocketmakers Collaboratorium, and they invest in start-ups as well. They’re a key part of the thriving creative technology sector in Bath. Richard had a project he thought might appeal to me. 

In my role as Co-Director of the Centre for Cultural and Creative Industries at Bath Spa University, I’m currently working with MyWorld, a global centre of creative technology innovation. Funded by UK Research and Innovation, this project brings together computer scientists, psychologists, behavioural scientists, technologists, filmmakers, theatre-makers and humanities researchers to help find solutions for the most pressing problems in the rapidly developing field of immersive media technologies. When the funding bid for this project was written back in 2019, artificial intelligence was an important aspect of many researchers lives, particularly in the context of developing computational software tools, for instance, using AI to improve the quality of filmstock shot in poor lighting conditions. Non-computer scientists like myself were accustomed to the increasingly sophisticated use of AI in our everyday lives, through tools that could be useful, such as predictive text on our phones and in our emails, or annoying, like the chatbot your bank offers you in lieu of actual customer service. 

So when Richard sat down beside me that cool breezy evening and told me that Rocketmakers had brought together a set of seven AI tools, including speech to text, text to speech, voice, 3D modelling and lip-synching in order to create an interactive, holographic, conversational 3D representation of Elizabeth Bennett’s head, I could not wait to talk with her. 

Elizabeth Bennett is the main character in Jane Austen’s 1813 novel, Pride and Prejudice. She is one of the most beloved fictional characters of all time, rivalled in popularity by perhaps only Harry Potter, Dracula and Scrooge. The novel has inspired countless adaptations and spin-offs, from the transmedia powerhouse of Bridget Jones’ Diary to the Lizzie Bennett Diaries, one of the earliest YouTube blockbusters. Lizzie’s journey from loathing to loving Mr Darcy is the template upon which most of the genre of romantic comedy has been built. #BookTok, the booklovers’ TikTok which is, incidentally, the only social media platform that directly influences book sales, would lose its collective mind over the idea of being able to converse with Miss Bennett.

In the publishing landscape, the arrival of easily accessible AI tools for writing is causing considerable upheaval. Generative text-based AI platforms, like ChatGPT, have spawned a host of AI tools aimed at helping writers write more quickly. In some instances, this exacerbates problems that already exist. In 2014 American novelist Chuck Wendig published a blogpost that described the self-publishing platform Kindle Unlimited as a ‘shit volcano’, where thousands and thousands of books are published online with the hope of capturing consumers. Adding AI tools into that picture has turned that volcano into a publishing Fukushima – an earthquake plus a tsumani plus a potential nuclear meltdown – with bad actors churning out millions of AI-written titles across both ebooks and audiobooks. Journalist Jane Friedman recently came across a tranche of books written by AI attributed to her on her Amazon author page; it was only her position as a well-established publishing pundit that prodded the company into removing them. Amazon recently took a tiny step toward dealing with this problem by limiting the number of titles an ‘author’ can publish to three books per day. 

In addition, there is a huge wave of litigation, as authors concerned that their copyrighted works have been used without their permission take these platforms to court. At the time of writing, there are at least six major lawsuits against OpenAI and Meta underway. One recent filing, a class action complaint by the Authors Guild of America, includes Jonathan Franzen, John Grisham, George R.R. Martin, and Jodi Picoult. AI writing tools are based on LLMs – large language models – which draw upon vast datasets from across the internet, datasets that include books that are copyright-protected as well as sites that contain pirated books, known as ‘shadow libraries’. These tools are also at the root of some of the grievances behind the screenwriters’ recent strike in the US.

This article has also been published on MyWorld’s website.