These days, it seems like everyone wants to ask me the same question: Are you afraid that artificial intelligence (AI) is going to make human editors obsolete? My answer, in short, is a resounding no.
Many times, this question is posed by someone in academia, where there’s been a lot of concern in recent months about the threats and opportunities offered by generative AI and chatbots like ChatGPT in the classroom. While I certainly agree that generative AI poses serious ethical issues for students and writers alike, it has arguably less applicability to most manuscript-editing-related tasks.
Instead, editors primarily use other AI-assisted tools to increase their speed and efficiency, including those that perform simple copyediting and proofreading tasks, such as identifying and correcting grammar and punctuation mistakes.
In this post, I’ll briefly identify how some editors are using AI in their work and more fully explain why I’m not concerned that these AI editing tools will replace human academic editors at any point in the near future.
How do editors use AI?
While there’s been much ado about the negative impact of AI and the threat of singularity, it isn’t all bad and shouldn’t be treated as such. There are benefits to using AI-assisted editing tools for both editors and authors.
First and foremost, there are programs that employ AI to perform simple proofreading and copyediting tasks, including identifying and correcting grammar, punctuation, and spelling errors, by comparing your text against a database of English-language rules. These function like Word’s Spelling and Grammar Checker, which has long annoyed draft writers and editors by incorrectly identifying errors.
Editors might employ AI and automations (for example PerfectIt or Grammarly) to ensure that they’ve caught all of the errors in a document either before or after they begin editing in order to increase their speed and consistency.
There are also editorial companies who use AI to pre-edit or as the sole editor on manuscripts in order to cut down on costs for authors. The price of using such programs is, after all, much cheaper than paying an actual person to do similar tasks.
Generative AI can also be helpful to editors in performing business tasks, including drafting emails, blog posts, and common messages or queries for authors. It can also assist in breaking down and analyzing a manuscript, including reverse outlining, summarizing or creating an abstract, and in brainstorming either content or titles.
If you’re interested in learning more about the many ways editors can use AI in their work, check out Erin Servais’s course, AI for Editors.
Why I’m not scared that AI will replace human editors
While there are many reasons to fear AI, and it certainly will trim certain tasks performed by human editors, there are also numerous reasons why we shouldn’t fear its recent expansion into our industry.
Academic authors are discerning and want a human touch
One of the key ways that academic editing differs from other kinds of nonfiction editing is how discerning our authors tend to be. They set high expectations both for themselves and for those who perform services for them, and often have a long list of needs and expectations that are best handled by a human with good customer service (author handling) skills.
Furthermore, although an advanced degree isn’t necessary to perform most academic editing services, many of our clients often want their editor to have subject-area specialization in order to better understand and make suggestions for their texts. Although some AI models can “learn” from the text it reads out on the internet, it cannot discern misinformation nor can it obtain a PhD degree in any discipline (yet). This means that they cannot yet identify nuances in your argument or suggest accurate alternatives for technical jargon as a human might.
AI editing is an imperfect science
As anyone who’s used Word’s Spelling and Grammar Checker already knows, a computer can point out errors that don’t follow a set pattern, but it often makes mistakes. This is the same with AI programs: a human needs to check to ensure that the AI hasn’t inadvertently introduced errors into the text.
AI also has limited capabilities in terms of the types of changes it can suggest. AI isn’t the best choice for those who are interested in improving their writing style, as evidenced by the bland writing style that is the hallmark of the chatbot. It won’t help you break the “rules” of a language in order to add flair to your writing or help you sort out what you’re trying to say in a particularly verbose piece of prose. Only a human with the ability to query the author using comments or meet on Zoom can do that.
AI is better for short-form documents
If you haven’t yet fooled around with a generative chatbot like ChatGPT, go do it now and come back to this post. It’s great for drafting short pieces of writing that you can tweak to your heart’s content: fundraising letters, job descriptions, abstracts for a paper, the kinds of emails or messages you dread writing, you name it. There are endless possibilities for the ways that this technology can help us cut back on the hardest part of writing, which is just getting something down on the page.
That said, it’s terrible at working with long documents, like those academic writers tend to produce. ChatGPT cuts you off after about 1,000 words, meaning that they cannot write to the length of even a journal article. Many programs, Word included, also tend to glitch or get bugs with extra-long manuscripts like academic books, which can be upward of 80 to 100,000 words (or 400 pages).
Again, because of the bland nature of the writing they produce, generative AI text also needs to be heavily line edited by a human editor in order to produce authorial voice, particularly that of a specific person.
AI can’t do brain-heavy services like line editing, developmental editing, or coaching
While AI can be useful for small tasks and drafting text, it’s not so useful for higher-level edits, including those that require the editor to use critical thinking skills, such as line editing, developmental editing, or coaching. Proofreading tools, for example, cannot interpret your work or help you establish an authorial voice or drastically shift the tone of your writing, as a line editor might.
Generative AI might assist with certain tasks of a developmental editor or coach, who works with authors on the content of their writing, including summarizing or establishing a schedule. However, since the changes it suggests are based on established rules, it cannot perform many of the bigger-picture aspects of writing an academic book, for example, establishing a strong argument that flows through the entire manuscript.
AI poses security issues
Academic authors are also understandably worried about the privacy issues related to AI software. If a document is uploaded onto a website or trolled by a tool, can that text become a part of the database that the machine uses to learn and then applies to future manuscripts it reads? What consequences might this have for academic integrity and for authors who are concerned about authorship rights?
In a field where many of our authors are concerned about the ownership of their ideas and the sequence in which their ideas are published relative to other authors, the use of AI to analyze texts might add an added layer of complexity and the possibility that one’s intellectual property rights might be breached.
Conclusion
While the integration of AI into academic editing has raised concerns about its potential to replace human editors, there are numerous reasons why this fear is currently unwarranted. The use of AI-assisted editing tools, such as those for proofreading and copyediting, has undeniably increased efficiency for editors and authors alike. However, the nuanced nature of academic editing and the discerning expectations of academic authors suggest that a human touch is not yet replaceable.
