July 15th 2024.
A recent study published in the journal Science Advances has found that using ChatGPT, a generative artificial intelligence tool, can greatly enhance the creativity and engagement of stories. The researchers compared the work of writers who utilized this tool to those who did not, and the results were significant. The stories created with the help of ChatGPT were found to have more plot twists and were overall more captivating for the audience.
However, the study also revealed that there was a downside to using this AI tool. It was observed that there was a lack of diversity in the stories created by writers who relied on generative AI. This could potentially lead to a decrease in what the researchers refer to as "collective novelty". In other words, the stories may become less unique and more similar to each other if the publishing industry were to embrace this technology.
So, how does this generative AI work? Essentially, it is based on large language models which are trained on vast amounts of text data. This enables the AI to process, interpret, and respond to requests in natural language, just like humans do. This means that it can generate content in various forms, such as text, images, audio, or video.
The study also looked at how this AI tool affected the creativity of writers. Surprisingly, it was found that the writers who were already highly creative did not benefit as much from the ideas suggested by ChatGPT, whereas those who were less creative became more creative with the help of this tool. The researchers noted that this resulted in a sort of "equalization" of creativity among all the writers.
However, while the individual creativity may have increased, there is a potential risk of losing collective novelty. This means that if more writers start using generative AI, their stories may become less unique and more similar to each other. The lead author of the study, Anil Doshi, a professor at the University College London, explained that this could be a potential concern for the publishing industry.
The study involved 300 participants who were asked to write short, eight-sentence stories for a target audience of young adults. These stories were then judged by 600 recruited individuals. The writers were divided into three groups - one without any AI help, the second with one AI-generated idea and three sentences of the story, and the third with up to five AI-generated story ideas.
The researchers found that the stories written by writers who took the AI's help were significantly more novel (8-9%) compared to those written without AI. The stories were also evaluated for their usefulness - whether they were engaging enough for the audience and if they could potentially be published. It was found that the stories written with AI assistance were not only more novel but also more useful.
Interestingly, the study also showed that the less creative writers improved significantly when using the AI tool. Their stories were found to be 10.7% more novel and 11.5% more useful compared to those who did not use AI. The AI made their work up to 26.6% better, 22.6% more enjoyable, and 15.2% less boring, according to the judges.
To measure the inherent creativity of the writers, the researchers used a psychological test called the Divergent Association Task. This test evaluates a person's ability to think of multiple solutions to a problem spontaneously, a key aspect of creativity. From this test, it was observed that the writers who used AI's ideas had stories that were 10.7% more similar compared to those who did not use AI.
In conclusion, the study showed that using generative AI, specifically ChatGPT, can greatly enhance the creativity and engagement of stories. However, there is also a potential risk of losing collective novelty if this technology becomes more widespread. It is clear that AI has the potential to greatly impact the world of storytelling, and further research is needed to understand its long-term effects.
[This article has been trending online recently and has been generated with AI. Your feed is customized.]
[Generative AI is experimental.]
