Since the release of ChatGPT in 2022, there have been questions about how it will affect creative industries. One study demonstrates what generative AI can do in regard to literary editing.
The release of generative AI shook the creative industry, factoring into the start of a writers’ strike. Ever since, there have been questions about whether human creators will soon be replaced by AI. This worry is shared by editors around the world. Understanding what generative AI can do as an editor can help human editors anticipate their future job market and can help writers and publishers to know what the best method for editing is.
THE RESEARCH
A group of professional literary editors and writers in Australia tested generative AI’s skill set in developmental and substantive editing on a short story and published their findings in the article “Ethics and the use of generative AI in Professional Editing.” Researchers Dr. Renée Otmar, Dr. Rose Michael, Sharon Mullins, and Dr. Katherine Day had a literary short story edited by both a human editor and ChatGPT at three different stages in the editing process: before any revisions by the author, after some revisions, and after acceptance for publication. Importantly, ChatGPT was prompted the same way in all three stages.
At the first stage, the human editor offered specific advice on what elements of the story could be improved. Examples include advice on the main character’s motivations and the author’s methods of foreshadowing. ChatGPT also offered advice on the elements of the story. Examples included advice on the story’s pacing and symbolism. But when asked to edit the story, ChatGPT misinterpreted some literary devices and changed others into clichés.
The human editor in the second stage offered less advice, but still gave specific suggestions. ChatGPT produced a list of broad advice, similar to the first stage, focused on elements of the story, but did not provide specific suggestions. Further, when editing the story, ChatGPT padded and lengthened phrases and clauses with unnecessary words.
In the final stage, the human editor made no substantive edits, only changing some words. ChatGPT, however, once again produced a similar list of story elements to work on and replaced the literary title with a generic phrase. Something that concerned the researchers was that ChatGPT did not flag the story as already published at this stage, and thus possibly plagiarized—something that would be innate to a human editor.
“We also recognize the rich collaborative potential for writers and editors in the early brainstorming stages of developing story ideas, when GenAI can provide suggestions to help clarify their thinking on rough drafts prior to in-depth work with a professional editor. But it is no substitute for ‘the real thing.’”
Otmar et al. (2024)
THE IMPLICATIONS
This comparison between the editing capabilities of ChatGPT and human editors indicates that generative AI in its current state is good at offering broad, general writing advice but is not consistent in providing specific suggestions for the literary genre. In fact, when ChatGPT was asked to edit the story, it made the writing worse. The researchers concluded that ChatGPT has “rich collaborative potential for writers and editors in the early brainstorming stages” but “is no substitute for ‘the real thing’” (Otmar et al. 2024, 1729).
This conclusion may look different in other editing fields, such as academic, technical, or journalistic, which may use AI differently in their editing processes. Further, the results may look different depending on the prompt used. For example, if ChatGPT were informed that the story had already been published, perhaps it would have provided less invasive edits. However, based on the results of the study, generative AI is still being developed and its editing capabilities could improve in the future, but for now human editors may still be the best bet for literary editing.
To learn more about generative AI’s editing capabilities, read the full article:
Otmar, Renee, et al. 2024. “Ethics and the use of generative AI in Professional Editing.” AI and Ethics 5: 1719–1731. https://doi.org/10.1007/s43681-024-00521-7.
—Elise Green, Editing Research
FEATURE IMAGE BY BRIAN PENNY
Find more research
Check out Parker Cook’s Editing Research article for more on how AI can enhance the editing process: “ChatGPT, Editing, and You.”
For more information on how human editors can collaborate with generative AI, read Clarencia et al.’s (2024) conference article, “The Impact of Artificial Intelligence in the Creative Industries: Design and Editing,” In 2024 International Seminar on Application for Technology of Information and Communication (iSemantic), pp. 440–444. https://doi.org/10.1109/iSemantic63362.2024.10762015
To learn more about how AI may affect the academic publishing industry, read Rachel Baron’s (2024) article, “AI Editing: Are We There Yet?” Science Editor 47, no. 3: 78–82. https://doi.org/10.36591/SE-4703-18.