Artificial Intelligence Books: A New Dawn of Writing or a Threat to Authorship?

The so-called generative artificial intelligences are becoming increasingly advanced and the first books written with AI either entirely or partially have already appeared. This phenomenon has posed many questions that we’ll deal with throughout this article. We’ll see some examples of its impact on writing, copyright, and some notes on plagiarism analyzing the pros and cons of writing with AI. 

Is It Possible to Write a Book Using AI?

The short answer is—absolutely. In fact, it’s already been done. According to an article by news agency Reuters, at the beginning of 2023, Amazon’s shop had over 200 books written by Chat GPT.

Under the cover of night, the artificial intelligence of OpenAI had been used to publish in one of the largest bookshops on the Internet. And it’s highly possible for the number of books written with AI to be much larger—Reuters reported on books actually signed by Chat GPT, but we don’t know how many used it but don’t disclaim it.

Some of the books were written for children, using AI tools like AI image creator Midjourney. Titles like The Power of Homework and The Star Weaver’s Lesson: Magical Bedtime Story could be found on Amazon until the platform became aware of their existence and decided to take them down.

However, the use of generative AI also has its upsides. Spanish writer and literary critic, Jorge Carrión, just published Los campos electromagnéticos, with Caja Negra Editora publishing house. In the creation of this book, Carrión used AI writing tools Chat GPT-2 and Chat GPT-3 to create two literary texts. For Carrión, these breakthroughs pose a challenge and a way to define artistic practices.

Of course, there are other examples of books written with AI whose authors (can we call them that in the strict sense?) have a less noble vision of AI and its impact on the artistic and literary fields. This is the case of Tim Boucher, an American who published over 90 books with Chat GPT and Midjourney on Amazon in nine months. According to Boucher, AI allows him to write 2,000 to 5,000 words within a few hours.

Generative AI has moved the debate surrounding the future of this technology to a much closer and more mundane terrain—from the possibility of an apocalypse provoked by a machine uprising to AI’s conquest of art and all it implies for the cultural industry, intellectual property, and copyright.

So, let’s see the pros and cons of writing with this kind of tech.

Pros of Using AI Writing Software

Even though AI is undoubtedly still in a very early stage, its potential has proved to be enormous. Those who defend this technology claim that its development brings, as it happened with other technological innovations through history, many tools for translators, artists, writers, designers, and illustrators from the publishing sector.

Improved Creative Capacity

Books written exclusively using AI have many defects—from the lack of consistency between paratexts and the body of the book to narrative contradictions and redundant texts. However, when writers use it as a complementary tool, it can be useful to prompt new ideas, overcome writer’s block, and get realistic information about what the characters would do.

The case of Argentine journalist and writer Mercedez Ezquiaga is an example of this. In 2020, the author faced writer’s block when writing her book Será del arte el futuro. Cuando la creación expande sus fronteras. To overcome it, she turned to Esteban Tablón, a friend who worked with a rudimentary AI called Lucia Funes. Lucia’s intervention, ultimately, was similar to hiring a ghostwriter, that is, it simulated Ezquiaga’s tone and style.

Text Proofreading and Editing

Even though generative AI can’t make a professional book report with all that it implies, it is true that many authors are already using it to edit and proofread their work, looking for misspellings, typos, and syntax mistakes. Granted—this saves a lot of time, but it’s similar to using a proofreading tool like Microsoft Word’s, so this task can’t be solely up to artificial intelligence.

Editing and proofreading texts using AI is useful, but its implementation still leaves much to be desired, for example, if we wanted to make a book analysis that goes beyond mere description. This is clear when we see it being used to review books on Amazon. Such is the case of The Cloud by Robert Rivenbank: AI reviewed the text after only reading the cover and other reviews on the platform to create something that looks more like an ad than a review. 

Writing Dialogues, Creating Characters and Situations

One last advantage we will mention about using AI in fiction is its ability to help us improve our creative writing. Artificial intelligence can help us with personification, describing spaces or simulating situations between characters in a story. For example, if we asked Chat GPT to behave like a suspect being interrogated in a dungeon, we would not only get a good approximation to how someone would behave in those circumstances, but also get a clear description of the scene and the interrogator.

We can go back to Tim Boucher as an example of this. The man who wrote 97 books with Chat GPT wrote them as part of a common universe. This means that Chat GPT can hold a record of the different characters and their characteristics to write accurate dialogues and place the characters in the same universe. This ability is really useful to maintain the tone and consistency through a common literary universe. 

Cons of AI Writing

The disadvantages of writing using AI writing assistants have been further focused on, since the first reaction to this technology, especially for those who understand the intricacies of the publishing world, was one of rejection. Historically, this is always the first reaction in the face of any development—it’s what happened with computer assisted translation tools, transcription tools and, probably, how copyists reacted when Gutenberg’s printing press appeared in the scene.

Intellectual Property and Copyright

This is one of the ethical and practical positions that have been addressed most extensively in the professional field. Starting at the beginning, in mid-July 2023, America’s Authors Guild sent an open letter to the main AI developers talking about consent, credit and compensation of their language models, which were “trained” with material protected by copyright. It isn’t a direct infraction, but an ethical position on the “intelligence” of these models and who it belongs to.

But there are examples of direct infractions: in mid-August, writer and cofounder of Open Road Integrated Media, Jane Friedman, discovered that her name was being used to sell books written by AI on Amazon without her consent. The company refused to take the books down because her name isn’t a trademark and, therefore, it doesn’t have legal protection. Something similar happened to Chris Cowell, a software engineer who published a book and was immediately plagiarized by an AI.

Poor Quality of the AI Text

Let’s talk about the elephant in the room—Are artificial intelligence fiction books any good? Does anybody want to read them? The short answer is—at the moment, no. AI can write alright, but can’t work based on experimentations and the unpredictability of artisanal craft. In fact, texts written exclusively with generative AI can barely choose the right names to write a story.

This stands out in the reviews of the book Death of an Author, by Stephen Marche, which was written almost entirely using three generative AI tools: Chat GPT, Sudowrite and Cohere. Even though according to journalist and American writer Dwight Garner, this is the first novel of the kind that is “halfway readable”. He goes on to say that, even though Chat GPT has the capacity of choosing a good title for a book or knowing how many pages a book should have, it would never be able to produce a text with convictions or honest doubts from experience, so artificial intelligence fiction books will never be able to be moving for humans.

Content Bias in AI writing

One of the main reasons why AI critics believe this technology is dangerous is the gender bias that seems to have filtered through the generative language models. A UNESCO study revealed regressive gender stereotypes used in generative AI. Even though we haven’t found an example of a book written with AI that shows a clear bias, it is noticeable in most automatic translation technologies and AI software’s answers.

This downside to generative AI doesn’t merely imply sexism in language or using masculine pronouns by default, but it also shows racial stereotyping, homophobic attitudes and richer narratives to men. This basically means that AI produces and deepens biased contents that further marginalize minorities.

Writing with AI—Tool or Threat?

Writing books using artificial intelligence is a reality that is becoming more and more common every day. However, there are still many questions and problems associated with this tech. Firstly, there is the issue with authorship and intellectual property rights which have been doubted—the US Court determined in August 2023 that AI-generated art cannot be copyrighted because rights are only applicable to people.

Even though there are cases in which the person who owns the copyright for a book isn’t necessarily the person who wrote it, like what happens with ghostwriters, books written with AI are at risk of having no copyright protection whatsoever. This is because language models like Chat GPT can’t be considered “original”, because they use other written texts as a database.

However, we could argue that no book, be it human-made or artificially-generated by an AI, is original in the strict sense of the word. There are many fan fiction books that have hit the bookstores, and even the most important books in the history of literature have, necessarily, conversed with all the ones that came before them (whether the author knows it or not). Still, we can notice that AI is groundbreaking and that it will probably change the way authors write in the future.

Deja un comentario