Back in high school, Mariam Galloway, now a 19-year-old Melbourne University student, took part in a popular trend going around. Using a website called The Hunger Games Simulator, you could change the names and photos of the original tributes to names of your choice, and the website would recreate a similar plot of the original story but with the new names. For Mariam and her friends, it was all for “fun” in dull school moments.
“Like you’re bored in class … you’re already thinking of your classmates. So, it is something silly and it’s fun because it’s people you know,” she tells upstart.
After some research online, upstart found on the popular platform Reddit, people were discussing using AI to craft fictional narratives or to replicate existing fictional stories, featuring real people. This can consist of friends, family, celebrities, or colleagues as the characters in the story. As well as inserting themselves in existing stories like The Hunger Games, users can input prompts such as “write a romantic story about [insert best friends’ name] and me” or “create a mystery plot starring [insert teacher’s name] as the detective”.
Dr Judith Bishop, a Doctor of Linguistics, started her research by developing training-data for AI ethics in developing training-data, she grew concerns about the ethical representation of individuals within AI. She tells upstart that the appeal of activities like this lies in the accessibility and creativeness of seeing yourself or others in various fictional scenarios.
“Generative AI makes doing things with digital information so much easier than it ever was before,” she says.
Anuki Pandithakoralege presses enter. The Artificial Intelligence (AI) chat generator’s three bubbles dance along her screen. The generated response, chapter 34 of Jane Austen’s classic novel Pride and Prejudice with her name in place of protagonist Elizabeth Bennet’s, slowly reveals itself on the screen.
Pandithakoralege, a 20-year-old psychology and criminology student, has been a reader from a young age, consistently consuming different genres. These days she gravitates more towards books discussing political and social issues.
We have tasked Pandithakoralege with putting herself or people she knows into a romance novel. She chose the 19th century classic novel Pride and Prejudice, and to tell us how it feels to see yourself inside a text. For Pandithakoralege, it was an intriguing and immersive.
“As soon as I got through the first paragraph, I got enthralled in it, I got hooked. And then it just felt like maladaptive daydreaming,” she tells upstart.
Pandithakoralege found herself “more invested” because it felt like it was her life when she read the AI-generated Pride and Prejudice.
“When it’s my own name, the thing about the maladaptive daydreaming, it blows the lines between reality,” she says.
While fun, this kind of creative play can raise ethical concerns, particularly regarding consent and potential for misuse. Approximately 57 percent of Australians aged 14 and over have used AI tools for various purposes. It’s crucial to consider the implications of using real identities in a creative work, especially if it is shared online, as well as issues of intellectual ownership and copyright for the authors whose work are being used. A 2023 study about copyright challenges within AI by Nicola Lucchi says that “generative AI systems challenge traditional concepts of authorship and ownership, since creative intent no longer rests solely with a human.”
Bishop says that while these tools can be imaginative or entertaining, they also bring up important social and cultural issues related to consent, trust, and digital rights.
“The key concerns are a combination of representation, consent, identity and the public or private nature of the work,” she says.
For example, issues would arise if users were to share these versions with real people’s identities online.
“Whether it is private or public, is this being shared in a place where it could have an impact on reputation or bigger psychological impact on the person who is named?” Bishop says. “You can think of some cases where this is totally fine. Like your friend, you make a birthday present of a story, and you put them in it, you just share it with them, they might like it, they might hate it, but it’s just between you two.”
“It might not be OK to share that same story online, because in doing so they might be revealing something about their friend that they wouldn’t want revealed to anyone who might see it, Bishop says.
These concerns connect to larger digital ethics questions. When posted publicly, an AI-generated story can reveal personal characteristics or ruin one’s reputation, even if it appears harmless. Drawing comparisons to freedom of speech, Bishop continues by saying that there are limits to creative freedom.
“If someone is going to get hurt that’s not ok,” she says. “The human rights commissioner put it nicely, ‘with freedom also comes special duties and responsibilities’ and that is the case with creative expression as well as freedom of speech of other kinds.”
When asked to repeat the activity she’d undertaken back in high school, replacing Katniss and Prim’s names with those of herself and her friend in a scene in the novel, Galloway found the results more unsettling.
“I feel like that for me is worse, they are way more invasive,” Galloway says. What once seemed like a playful classroom game now carried a different weight.
“I felt less emotionally connected to it. That scene in the book is pretty good, now it just came across as strange,” she says.
Another issue is the potential for incorrect information about real people being produced and shared online. Popular platforms like ChatGPT or Character.ai use advanced natural language processing to generate stories,. ChatGPT can produce a short story in seconds from a simple prompt, while Character.ai allows interactive dialogues with AI-generated characters. All of these AI platforms, while effective, also still have holes in their knowledge where they will miss something or make up something due to a lack of knowledge.
Bishop notes that generative AI blends the line between reality and fictional scenarios in different ways, “including its tendency to hallucinate information, which itself has led to defamation cases.”
During the small experiment, our participants spotted small errors after AI had generated the new version of the stories. While ChatGPT has been used for more complicated things, the system does make some mistakes. When Pandithakoralege used ChatGPT, she noted some issues around a significant characters name.
“It kept Charlotte as Charlotte at the very end. The last mention of Charlotte. I thought that was interesting, the algorithm missed it somehow,” she says.
Before the caution arises, however, there’s the excitement, laughter, and curiosity in the act. Not everyone who joins in is purposely trying to cross a line. For many, dropping a friend or celebrity’s name and characteristics into a wild, AI-generated story is all about laughter. People tend to do this based on curiosity or entertainment, since the story is usually shared between friends or kept private.
Murasaki Kobayashi, a 25-year-old master’s student in business, asked to generate a story with her own idea by using ChatGPT. She created her own prompt, opting to use a traditional romance trope, a love triangle, including two of her friends, inspired by pre-existing romance novels. Kobayashi described reading her AI-generated story as if she had “entered another dimension and saw another version of her friends’ life”.
“It [was] too hilarious, I couldn’t stop laughing. I can’t imagine if I send this to them [her friends],” Kobayashi tells upstart. “I’ve known them for years.”
Utilising AI to explore each individual’s creativity is complicated. Legally, Australia’s Online Safety Act (2021) and AI Ethics Principles provide a starting point, but regulation hasn’t caught up with potential representational harms like these. Until clearer laws emerge, experts urge users to pause and think before generating such content.
Bishop says that finding the line between intrusion and fiction is something “we really need to be thinking about, having more clear regulations around digital rights.” There is currently still no map that has been clearly laid out as AI is evolving and developing so rapidly.
Perhaps the question isn’t can we do it, but should we? And if we do, how do we do it wisely?
Article: Nykita Van Den Berg, Milla Webster and Quynh Nguyen.
Photo: MacBook Pro by Amelia Bartlett found HERE and used under a Creative Commons licence. This image has not been modified.
