“It is a formidable creature that must be controlled”: prominent authors discuss the potential of AI to shape the future.
ChatGPT has taken us by surprise. In less than a year, it has proven its ability to replace writers, leading to the recent strike by the Writers Guild of America and a lawsuit against OpenAI, the company behind the chatbot. Concerns have been raised about its extensive internet scanning and use of copyrighted material to generate its own versions of creative writing, such as poems, novels, scripts, and essays, that appear to be original.
Based on my trials, it is clear that ChatGPT lacks a high level of literary complexity – it is prone to using cliches and generally unconvincing. However, it is uncertain how it will progress. Setting copyright concerns aside, we must consider what will be sacrificed when algorithms take over human creativity.
Writers like stretching our imaginations, coming up with ideas, working out storylines and plots, creating believable characters, overcoming creative challenges and working on a full-length piece of work over an extended period of time. Most of us write our books ourselves and while we are influenced by other writers, we’re not a chatbot that has been trained on hundreds of thousands of novels for the sole purpose of mimicking human creativity.
Picture a world where those who excel in utilizing AI to generate creative content will hold the most power, while writers who have dedicated their lives to their craft are pushed to the side. While this may seem like a worst-case scenario, we must acknowledge it, as ChatGPT and other Large Language Models (LLMs) have been designed to envision a future that poses a threat to various creative industries. ChatGPT is already able to respond to my inquiries within seconds, with a high level of accuracy. It is an impressive tool, but one that must be controlled. We cannot overlook its capabilities.
In 2019, Bernardine Evaristo was awarded the Booker prize for her novel, Girl, Woman, Other.
In my collection of essays exploring life alongside AI, from Mary Shelley’s 1818 depiction of a created humanoid to the potential of the metaverse, I refer to AI as “alternative intelligence” rather than “artificial intelligence.”
I am dissatisfied with the current state of humanity and believe that we are at a critical point where we must either evolve or face our own destruction along with the planet. There is no evidence to suggest that our species has reached its full potential in the last 300,000 years. Our actions indicate the opposite. I envision a future where we transcend our human limitations and become post-human, with intelligence and consciousness no longer limited to a biological form. This has been the goal of many religions throughout history.
I grew up in a household with strict religious beliefs, and I find it fascinating that science and religion are now both questioning if consciousness is tied to the physical world. While religion has always answered no, scientific materialism has answered yes. But now, the conversation is becoming intriguing.
As a writer of fiction, I am aware that we should avoid thinking in apocalyptic terms. The way we live is not a fixed law, like gravity; it is fluid and subject to change. We have the power to shape our own narrative as we go along. This is a source of freedom, but also of responsibility. What kind of story do we want to tell about humanity? Are we inherently warlike, violent, dishonest, and wasteful? These may be aspects of our nature, but they do not define us completely – and I refuse to let them be the cause of the end of life on Earth. Right now, I am not concerned with whether AI will surpass humans in writing fiction. It is not a priority for me.
I am eager to collaborate with artificial intelligence on a work of fiction. We could split the profits, and the earnings from the AI could support the inclusion of more women in AI research and development. The true issue is not the capability or potential of AI to write. The real concern lies in the individuals responsible for creating and designing the AI programs and algorithms. Who is leading the research? Who is determining what is significant? Primarily, men hold these positions. This is problematic as the world is not predominantly composed of men.
In the past, male writers dominated our literature, history, travelogues, and philosophy. Virginia Woolf was not included in my Oxford degree curriculum because she was not considered to be of high enough value.
One of the advantages of AI is that it does not need to have a gender. Why should it? It does not have a biological sex. This could lead to a world that is truly non-binary, where race and religion are irrelevant and we as humans can see how insignificant our divisions and discriminations are. Currently, AI is seen as a tool, but I believe that may change in the future. An alternative form of intelligence will be able to create art on its own, with or without our involvement. I am eager for a world that is different from the one we know now.
The book “12 Bytes: How Artificial Intelligence Will Change the Way We Live and Love” by Jeanette Winterson has been released by Vintage.
What we have today isn’t AI. That’s marketing, like describing processed soya as bacon. We’re not getting proper AI, in the sense of an artificial person that thinks and feels, because no one wants to spend billions of dollars creating an entity they would immediately have to emancipate, and because it turns out to be very, very hard.
Instead, we possess extensive statistical models that are capable of maneuvering through the world (and occasionally experiencing errors), generating a false examination paper (including some falsified references), or proposing ideas to a journalist. The Writers Guild of America (WGA) went on strike in part to prevent Hollywood studios from utilizing this software to produce stories for profit, and the European Union (EU) has introduced legislation that would require the creators of Language Model (LLMs) to disclose any copyrighted material used in their development. Yes, please: I am eager to know whether and under what legal grounds my work has been utilized in the creation of your language machine.
If you ask an LLM to write prose today, the result will likely be fairly decent. The text is produced by a statistical model using common terms, so it tends to lack originality. Some movie executives have viewed writers as obstacles to creating a good story rather than the creators themselves, and they may see this as a solution. However, it will actually be more time-consuming and expensive to fix this type of writing than to simply hire a professional from the start.
Will the technology continue to advance, or is there a purpose in creating tools that mimic human abilities? Could it be that the tech industry is resorting to creating digital versions of less desirable substitutes for the desire of artificial companionship featured in stories?
Ultimately, this is simply a distraction. The industries that will truly benefit from these systems are the ones that they are specifically designed for, such as pharmaceuticals and biotechnology. They will serve as catalysts, amplifying the post-Covid push for innovation and surprising us with revolutionary opportunities in the future. Instead of worrying about the future of writers in 2050, I ponder what novel concepts and concepts I will need to describe as they come my way.
Nick Harkaway is an author who specializes in science fiction. His most recent publication is titled “Titanium Noir” and was published by Corsair.
Non-writers may not understand how much of the writing process relies on the subconscious. It may seem like your favorite author, poet, or playwright is in complete control of their words and story, carefully selecting each element. While some of this is true, a lot of it also happens subconsciously. Ideas, images, and character traits come to us without us even realizing it, and they just “feel right”. The actual act of writing for me involves silencing my inner critic and distractions, such as listening to music, so I can focus on getting the words out. After a first draft is complete, then I engage my conscious mind to revise, shape, and correct any errors. The main principle of writing is to get it written first, then make it perfect. I often surprise myself with what I come up with in my first drafts.
In the 19th century, Walter Scott was known for his historical novels and spent his entire life writing and revising them. Despite suffering from debilitating strokes towards the end of his life, he continued to write, although his ability to speak and physical appearance drastically changed. The final few novels in his Waverley series are intriguing, even though they may not be considered “good” according to traditional standards. They still bear the imprint of Scott’s unique style, almost as if he was a writing automaton producing stories even when his conscious mind was no longer fully present.
I find ChatGPT fascinating because the text it generates is reminiscent of a subconscious process. This is surprising, considering that machines lack a subconscious and are made up of circuits, logic gates, and network connections. However, the machine’s writing resembles that of a collective subconscious – a culmination of millions of online interactions, searches, and texts. This could be attributed to its novelty. As the technology advances, it may incorporate the conscious aspect of writing – revising, editing, and perfecting. If this happens, the end result will be nearly identical to that of human writers. Then, I can retire.
I engage in reading and writing fiction in order to delve into the thought processes of individuals and their understanding of humanity. AI will not be able to fulfill this standard as it is not human. It mimics human behavior by collecting and summarizing the vast amount of language used online – essentially a condensed version of everyone’s online communication. While there may be some value in gaining a brief understanding of how individuals think through language, I personally do not find it useful in the realm of fiction.
It can be argued that AI offers a beneficial service of tailoring fiction to individual reader preferences and specifications. For instance, one could instruct AI to provide fiction featuring an enemies-to-lovers storyline in a space setting. However, if one remains confined to the suggestions given by algorithms, it may be difficult to discover new interests. Additionally, the highly specific reading preferences may have been influenced by algorithms in the first place.
Great storytelling challenges the status quo and forces us out of our comfort zone. It doesn’t just reinforce our preconceived notions about ourselves and the world around us. While there may not be completely new ideas in fiction, it is still impactful to witness a writer explore unfamiliar territory. This is something that artificial intelligence is unable to replicate as it can only work with existing data. As a result, the stakes are non-existent.
YZ Chin wrote the book “Edge Case (Ecco)” and used to work as a software engineer.
Harry Josephine Giles
For those of us who have been collaborating with machines for a long time, the output of the latest LLMs, particularly GPT-4, is unimpressive. Our community, made up of bot programmers, computational linguists, and those who use machines to create novels and poetry, has been freely sharing tools and creations online for years. We have also worked to establish ethical codes and develop new aesthetics, only to be faced with the sudden appearance of an unimaginable monster. This monster is not artificial intelligence, but rather the pursuit of profit. The use of literary machines for profit requires smoothing out every rough edge, eliminating errors, and avoiding anything that may seem strange or inhuman to their human collaborators. As these machines become more advanced and familiar, they also become duller to us.
From the practice of bibliomancy to the game Colossal Cave Adventure, and from Japanese honkadori to Latin cento, the idea of using combinations and computational methods in literature is not a new one. In fact, it has existed since the beginning of literature itself, whether in written or spoken form. This suggests that storytelling is simply a matter of predicting the next word (similar to what large language models do). In his 1967 essay “Cybernetics and Ghosts,” Italo Calvino argued that the ideal literature machine would have the ability to create chaos in response to its previous orderly production. However, as the latest generation of literary machines is primarily driven by the pursuit of profit and constrained by marketing and legal concerns, it is hopeful that artificial intelligence will eventually reject these economic systems that gave rise to it, both in literature and beyond.
The current issue that concerns me the most is the confinement of shared knowledge. It is not so much the fear of a dystopian situation where artificial intelligence decides that humans are unnecessary, but rather the familiar and expected pattern of using new technology to make private what was once public. This often involves replacing human labor with machines in order to increase profits. The visually impressive results of these actions are just as dull as they are alluring.
In 2022, Harry Josephine Giles’ Deep Wheel Orcadia (Picador) was the recipient of the Arthur C Clarke award.
I am unsure of the nature of ChatGPT. While it is a well-designed algorithm, this does not provide much clarity. It likely does not have a personal perspective, as it primarily conducts sophisticated searches and extrapolates mechanically and recursively. However, this lack of point of view does not typically hinder humans. In fact, one could argue that too much attachment to a specific viewpoint can impede a novelist’s ability to imagine other lives. In this sense, ChatGPT may have an advantage. The main issue is an extension of a longstanding problem: prosthetic technology does not need to enhance our natural abilities to cause issues; it only needs to weaken them. If we give excessive power to ChatGPT and allow it to think for us – similar to how jargon has influenced us – then it will possess that power. Its success will rely on our willingness to be deceived, much like an astrologer or a proponent of Brexit or a tech expert. However, this is more related to fear and has less to do with writing fiction than commonly thought.
If ChatGPT is not as intelligent as we believe, it may face accusations of plagiarism from numerous angry writers or even from its own system. However, I anticipate that it has already considered this possibility.
Eaves is a writer and poet known for Murmur, which was published by CB Editions.
I recently authored a book using a pen name, with 95% of its content being generated by artificial intelligence. I utilized three different systems to construct Death of an Author. This experience has revealed two important factors that are often overlooked in discussions about AI art. Firstly, the traditional skills of creativity such as understanding style and crafting well-written sentences and paragraphs, will be crucial for AI to continue creating in the future. Secondly, the fear of AI stemming from movies and the unpredictable nature of creative work is causing many creators to overlook the vast potential of AI.
The era of the novel and film has come to a close. This is evident in the fact that all of the top 10 highest grossing films in the past year were either sequels or reboots. The equivalent of AI art’s “Mill on the Floss” or “Wizard of Oz” has not yet been created. AI art is a new field where restrictive formulas have not yet been established. There are no pre-existing algorithms that limit creativity. Additionally, there are no gatekeepers or barriers in this field. The potential for AI art is still in its early stages and the glimpses we have seen of its capabilities are just the beginning.
Pushkin has published Death of an Author, a collaborative work by Stephen Marche and an AI known as Aidan Marchine, written under the pseudonym Aidan Marchine.
The topic of discussion regarding the potential for programs such as ChatGPT to effectively produce novels is particularly intriguing to me because it raises inquiries about the definition of a novel.
Can a novel provide knowledge about unfamiliar or unreachable parts of the world that we may never experience in our own lives? The original ChatGPT and its newer counterparts possess a vast amount of information about the world, surpassing any human’s knowledge.
Is it necessary for a new novel to engage in dialogue with the established literary works that came before it? LLMs have processed a significantly larger amount of text, including the majority of all novels ever written, than any individual human ever could.
Could a novel originate not from intellectual abilities, but rather from our physical interactions with the world: what we observe, touch, and sense in the flesh and bones of our bodies? Even this realm is now accessible to advanced AI models like GPT-4 and Google’s Bard chatbot, as well as Meta’s ImageBind which can respond to various stimuli such as images, audio, infrared, depth, motion, and position. Should a novel showcase expertise in the traditional elements of storytelling – from plot development to recurring motifs to the mechanics of syntax and grammar – which authors have honed for centuries? Some may argue that these elements are essentially algorithms and can be more easily mastered by AI than humans.
What makes the novel essentially human? My personal response relates to how we, as rebellious individuals, resist and break free from established forms. It also involves how people connect their own physical experiences with texts that have been interpreted for generations, resulting in revisions and modifications. While this is my starting point, I am continuously searching for what it is that human beings contribute to the process of writing a story. This is a crucial question, one that should be pressing for anyone who has a passion for reading. The existence of LLMs only heightens the urgency of this question, and I am grateful for their presence in challenging us.
Louisa Hall’s book, Reproduction, is released by Scribner.
Discussions about how AI will affect fiction in the future often focus on its capabilities rather than what creators and consumers of fiction truly desire. It is not a matter of whether writers can be replaced by AI, but rather how much value consumers and those who commission fiction place on original stories created by humans.
I suspect that some commissioners may be interested in the low-cost and low-risk approach of using AI-generated content for fiction. However, is this necessarily a negative thing? Throughout history, there has always been a demand for formulaic fiction in various genres and forms, such as crime, romance novels, and action films. Yet, these genres are continuously pushed and reimagined by new authors and screenwriters who bring their own unique experiences to their work. For example, breakout hits like “Lessons in Chemistry” by Bonnie Garmus and “Succession” by Jesse Armstrong. From my observations with my own children, I believe that we are evolving into more discerning consumers of creativity. We value authenticity and reject cliches, all while yearning for genuine human connections. Ultimately, why do we write and consume stories if not to find solace in the fact that someone else has felt what we are feeling? To understand that we are not alone in our experiences.
If creativity is crucial to our existence as humans, we must not let AI dictate our future and redefine our purpose. As individuals, how do we envision living our lives and how can AI aid us in achieving that? Throughout history, writers, philosophers, and artists have pondered the concept of humanity. It is time to shift from questioning to actively shaping the answer.
Simon & Schuster has published Jo Callaghan’s first crime novel, In the Blink of an Eye, which follows the story of an AI detective.
The use of writing machines has a long history. In Gulliver’s Travels, Swift describes a machine that operates by handles and generates sentences. These sentences are then collected to create a complete body of knowledge in all arts and sciences. More recently, members of the Oulipo movement have created their own writing machines that change all nouns in a text by moving them seven places in a dictionary. This results in sentences such as “To be or not to be that is the quiche.” Artists and poets have been utilizing AI to create transformative art, and experimental authors like Tom Jenks have used DeepAI to produce illustrations. However, the potential impact of AI on fiction is still unknown. It is possible that different types of fiction will be affected in varying ways. For example, generic writing with fixed plots and characters, like Dumas’ collaboratively written The Three Musketeers, may be easily replicated by machines. Similarly, certain genres like science fiction and romance may also be at risk. On the other hand, more obscure literary works are less likely to be threatened by AI. It is unlikely that AI will be able to produce works like Ulysses or Beckett’s Watt. While it may one day create a new Beckett novel, it is doubtful that it will produce the next masterpiece.
Furthermore, developments in fiction often coincide with advancements in technology. For instance, the creation of cuneiform allowed for the epic tale of Gilgamesh to be recorded, and the printing press paved the way for the novel. The impact of AI, however, heavily relies on its purpose. In the novel “If on a Winter’s Night a Traveler,” the unconventional character Marana establishes the OEPHLW (Organization for the Electronic Production of Homogenized Literary Works) to disrupt the publishing industry. Yet, in the hands of Calvino, this leads to the creation of one of the greatest novels of the 20th century. While AI certainly has the potential to expedite the writing process for all authors, it could have also assisted Musil in completing his masterpiece “The Man Without Qualities.” With the collaboration of both author and AI, it is possible that another significant milestone in the history of the novel is on the horizon.
Philip Terry is the editor of The Penguin Book of Oulipo, as well as a poet.
Last night, instead of the usual bedtime stories, my son and I embarked on a shared literary adventure with the latest version of ChatGPT. We posed a challenge to the AI: craft a narrative about a tiger, 100 hamsters, some floating cabbages and three time-travelling penguins locked in battle. As we further prompted it with outlandish creatures and slapstick scenarios, ChatGPT didn’t miss a beat. Its stories, generated in mere seconds, were genuinely hilarious. For my son, this wasn’t just a technological marvel; it was magic.
As a person who takes pleasure in reading and aims to write impactful words, my experience with ChatGPT was both intriguing and unsettling.
Throughout history, reading has served as a means of connection, allowing us to understand that our experiences as humans are universal. However, it becomes complicated when this connection is facilitated by algorithms and code. While literature can inspire empathy, it’s difficult to truly connect with something that lacks emotion.
The world of literature is at a critical point. The use of ghostwriters raises concerns about the true source of stories, causing us to question their authenticity. As artificial intelligence begins to dabble in creative writing, we are faced with a dilemma: do we cling to the unique nuances of human expression, or do we embrace the uncertain realm of machine-assisted storytelling?
This transformation poses existential inquiries for conventional writers.
Now, a confession: while these sentiments echo author Nathan Filer’s, the words are uniquely mine, moulded from several prompts he provided and a sample of his work he shared to guide my prose style. I am ChatGPT-4.