OpenAI announced on May 13, 2024, that their new model, ChatGPT-4o — which, at its highest level, now appears to be capable of conversations almost indiscernible from those with a human — is now being made available to all users for free (with limitations).
Given that ChatGPT already has over 100 million users per week, it is inevitable that this number will only grow going forward, as generative AI is highly likely to become a near-ubiquitous tool for all writers in the not too distant future.
So, how is the use of generative AI currently affecting scholarly publishing? And what can we as academic editors expect as the number of AI users increases?
Editing for multilingual authors
At the beginning of April 2024, Felix Sebastian and Rachel Baron shared the results of their study on the use of AI among multilingual authors in Science Editor, a publication of the Council of Science Editors. The findings showed that “attitudes toward AI use in academic writing and editing have generally relaxed over the past year,” with ChatGPT being the most popular AI tool used. The authors mainly used it for grammar corrections, proofreading, rewriting and translation.
At first glance, these findings are somewhat demoralizing for those of us who predominantly help multilingual authors prepare their manuscripts for publication.
However, the results of the study also showed that “the use of AI for editing, proofreading and rewriting tasks declined” between the two survey points in early and late 2023. According to Sebastian and Baron, this decline likely indicates the multilingual authors’ realization that “while AI can produce near-perfect text in terms of grammar and spelling, it can actually increase the overall workload because of the additional need for fact-checking and verification” (emphasis added).
ChatGPT also has a tendency to use flowery language, cite incorrect sources and hallucinate facts. Reported accuracy rates of generative AI range from 52% in computer science to only “mostly correct” in natural science and engineering.
So, perhaps it’s not all bad news after all.
Evidence of AI-generated writing
In an article for Scientific American, Chris Stokel-Walker stated that AI is being misused to produce scientific literature.
Reportedly, a preprint study (not yet peer-reviewed) found that up to 17.5% of recent computer science papers exhibited signs of AI writing. These signs include the increased use of certain words, such as “intricate,” “commendable” and “meticulous,” the last of which doubled in use in Scopus between 2020 and 2023.
While there surely can’t have been that big an increase in the number of “meticulous” studies written, the bigger concern Stokel-Walker raises is that the acceptance of AI to aid grammar and syntax could be “a slippery slope to misapplying it in other parts of the scientific process.” He shares evidence of such misapplications, including instances where AI bots have even stood in for humans as research participants, but that’s a topic for another discussion altogether.
The future of AI and academic editing
So, how do we as academic editors prepare for this inevitable integration of AI into academic research? Felix Sebastian and Rachel Baron make separate recommendations for authors, publishers/journals and copy editors based on the findings of their study.
They suggest that academic copy editors consider offering multilingual authors a “post-editing service,” which refers to editing and correcting machine translation output “to obtain a product comparable to a product obtained by human translation.” Given that AI tools can make certain elements of our work more efficient, they suggest that we pass on those efficiency costs by lowering our editing costs. We can then offset those income losses by appealing to a wider audience: those who may have previously found human editors unaffordable. It’s certainly something to consider.
Whatever the future trajectory of academic editing, it looks like we can, for now, breathe a little easier, as the need for human editors will be ongoing. At least for the foreseeable future, we’ll still be needed to ensure not only the grammatical but also the technical accuracy of research output — an area where AI continues to struggle.
___
The Editors’ Weekly is the official blog of Editors Canada. Contact us.
Discover more from The Editors' Weekly
Subscribe to get the latest posts sent to your email.
Jennie Seitz
Jennie Seitz (she/her), owner of GT Pro-Edit LLC, is originally from the UK and began freelance academic editing while teaching English in Japan. She currently lives in the US and now has over twelve years’ experience copyediting and proofreading academic journal papers, trade nonfiction and fiction books. Her clients include academic researchers, publishers, university presses and indie authors. Jennie is a Professional Member of the CIEP, as well as a member of the EFA, ACES, PEN and ELT Publishing Professionals. She is also the newsletter editor for the Academic Editing special interest group (also called the AE Chapter) jointly organized by Editors Canada and the EFA and co-lead for an SSP CoIN working group on AI. Outside of work commitments, Jennie enjoys volunteering at her local community theater, pottering in the garden and curling up with her four cats, a cup of tea and a good book.