Log In
New research suggests that as more writers use AI to help create stories, their writing “would become less unique as a whole and more similar to each other,” the study’s authors wrote.
Valerie Sumner, founder and principal at VRS Meetings & Events, Inc., is an enthusiastic user of AI platforms, including Spark AI and ChatGPT, for a long list of event-related analyses and tasks. But along with experiencing AI’s power to boost productivity, she’s encountered its limits, Sumner told Convene.
For example, when Sumner used AI to help her draft event marketing emails for a client, she created a template using language tailored to her client’s needs and asked ChatGPT to generate 15 different versions. It generated instantaneous results, including suggested fixes and alternate language. But something was missing, Sumner said: Her voice. The grammar and sentence structure created by generative AI might be perfect, but generative AI creates content that sounds a lot like … AI. Sumner said that you can tell that it’s AI-produced copy “after you use it enough — it doesn’t sound like me.”
Valerie Sumner
Sumner’s observation about AI-generated writing’s tendency toward homogeneity tracks with ongoing research about the tradeoffs we may be making when we turn more of our writing tasks over to ChatGPT. On the one hand, AI not only runs circles around humans when it comes to minimizing the time it takes to create content, it objectively improves most people’s creativity and writing, according to research conducted by Oliver Hauser, an economist at the University of Exeter, and Anil Doshi, a professor of strategy and entrepreneurship at the University College London School of Management. In their study, they found that writers who had access to five ideas generated by AI wrote stories that were seen to be an average of 8 percent more novel and 9 percent more useful than the writers that worked without it. Using generative AI “professionalizes” writing, Hauser and Doshi wrote in a paper that was published in July in Science Advances.
But there’s a catch: When the stories that were written with the help of AI were considered as a group, the researchers found that they lacked diversity compared to the writing produced by the group who worked without AI. Their findings suggest — as Sumner observed — that as more writers use AI to help create stories, their writing “would become less unique as a whole and more similar to each other,” they wrote.
In a recent Instagram post, content creator and software engineer Abi Bouhmaida listed a few dozen of ChatGPT’s most over-used words and phrases, including “innovation,” “integration,” “thought-provoking,” “exemplary,” “align,” “augment,” and “transformation.” “It’s as though you asked a million people to respond to your prompt, and ChatGPT gives you the ‘average’ of all their responses. Its output will be generic and vague,” Bouhmaida wrote in the post. “ChatGPT won’t give you the best word for each scenario, it will just give you the most commonly used one.”
Generative AI is a rapidly evolving technology, and its full potential is yet to be explored, Hauser and Doshi pointed out in the paper. The advances are coming so quickly that ChatGPT-4, the generative AI model they used in their research, “may soon become obsolete.” And their research, they added, is just a “first step toward understanding the relationship between generative AI and human creativity.”
Meanwhile, Sumner has developed a working relationship with generative AI and her own creativity — whenever AI waters down her voice, she just puts it back in.