More About AI: Guidelines

AI art creation of my cat, “Benjamin.”

But if you think your job as a researcher or publisher is obsolete, the great news is that you couldn’t be more wrong. I use ChatGPT all the time now- it can be of immense help. But, for starters, the more I use it, the less inclined I am to use it. I’m confident that I do a much better job, and even if you might dispute that, I can at least say with certainty that it would be toilsome and really boring to use AI all of the time over my ‘native’ human skills. AI can replace tasks, but it can’t replace human beings. There’s a lot to unpack there, I realize, but let’s just start with the proposition that AI will take over human tasks but will never replace humanity. That’s science fiction.

AI has been of great help to me as an academic writer, and I have a simple analogy to share: a recipe is not a meal. Someone has to follow, interpret, and elaborate on recipes. It can be very subtle and subjective. In a nutshell, that’s what ChatGPT or Claude can do for your paper: give you instructions and facts. They cannot give you inspiration, imagination, or ingenuity. And why would anyone want them to? AIs are great historians- they don’t come up with the next great thing. That’s our job.

Feel free to use and experiment with AI. If you are writing a paper or about to embark upon one, please don’t plagiarize AI. The recipient will know it’s AI, and you will become deflated and dejected. But if you’re feeling writer’s block or are perhaps utterly confused about the academic task ahead of you, by all means refer to AI for help.

How AI Can Help Authors Before Editing

Remember that AI is similar to a search engine- a really big and complex search engine. But it did not come out of nowhere: as we said, AI uses the same technology that computers have always used. Instead, it is the latest iteration of ever larger waves of computing power that produce amazing results. There are revolutions within the digital revolution, and AI is the newest one. There will be more.  Where it transcends a search engine, though, is in its ability to ‘evaluate.’ AI can’t actually evaluate or intuit, however. It uses machine learning to process very large amounts of data patterns and, mathematically, generate probable outcomes (just like your phone does when you text). While it seems uncanny and familiar to us, AI is not thinking in any way, manner, or form that humans would recognize as thinking. Instead, AI analyzes user content and offers suggestions for expansion.

Every writer has trouble getting started at times or even experiences writer’s block. The best thing a writer can do to escape this trap is ‘brainstorm,’ that is, jot down every idea or phrase that comes to mind, even if some of them seem completely irrelevant. AI is an excellent tool for generating brainstorming ideas, and in the case of ChatGPT, for example, it will always offer a “would you like me to?” suggestion for expanding your topic (it’s often clickbait). This is extremely helpful for the writer to establish a direction. Be careful, however, to only allow your AI to generate example text that is strictly within your parameters for the topic. In human terms, AI can be too ‘eager to please,’ diluting your construction, or just plain sounding odd or out of place. Any prose generated by AI should be evaluated skeptically before drawing on key ideas, researching further, and filling out a project. I never copy-paste or plagiarize AI-generated text- that is a sure-fire way to degrade your writing’s quality.

We are of the opinion that the best use by far for AI in your project is to allow it to organize and produce a well-structured layout for you before getting started. Whatever you may think of this blog, that is what I have done here. And that is my point- a framework is something that a writer can use to fit their objectives. With an AI-produced frame, the writer can meet narrow and succinct criteria, or they can expound. I’m expounding big time here- some would say I’m being verbose, but I like to think that I have a lot to say. While on the topic of ‘me,’ I’m not good at drafts and outlines. I love to wander out into the weeds. AI has really helped rein me in. I don’t need AI’s prose and poetic articulation very much as I have plenty of my own. But some writers may be the opposite: they may be excellent at deductive frameworking but not as good at creative imagery. AI is a universal tool that can accommodate all types of writers.

Finally, for this section, I’d like to touch upon an issue that is, for me, controversial, because, after all, we want our writing to come out sounding in the style of the human who wrote it, not a boring robot (sorry, Chat ol’ boy). But we need to check our grammar and ease of reading (ChatGPT calls this ‘readability’, which feels awkward to me, but you be the judge). Not out of choice, but I have always used Microsoft Word for editing (I find the program to be atrocious, to be honest, but I’ll end up with lots of formatting problems with my client and publishers if I don’t use it). I digress, but what I want to stress here is that for the last 30 years, I’ve had that thing throw its grammar suggestions at me (and my concerned clients), and I have rarely used its suggestions. I have made plenty of grammar mistakes on my own, which I fix one by one. Sometimes, no, often, I prefer the original language marked as ‘wrong.’ What really drives me batty is these grammar checkers’ hate for the passive voice. Just say no! The passive voice is a fine voice to be written in.

I still distrust grammar checkers, and although ChatGPT and Claude do an excellent job in milk-toasting your writing into correct but boring grammar, I am very careful with their use and usually only correct the stuff with red lines, not the interesting phrases with yellow lines. Here is my main point: You are a human being, and you have style. No one in this world has the same style as you. You need to let your style shine through. I’m not sure how well AI can help you with that. AI can give you great ideas, but it can’t give you the voice you need to express those ideas. No matter what genre I write in, be it a sorciolgical study, epidemiology (I do a lot of that), review of a cancer study, a political analysis, a work of fiction (haven’t done that in many years), a social media post (usually my dark side), or a birthday card, I always allow my style to show through. So should you.

How Editors Use AI Responsibly

Like a Swiss Army Knife, AI gives editors a solid array of useful tools in a small package. It should be remembered that AI is for fixing and polishing prose, not for ghostwriting. One example of this is having AI perform terminology consistency checks. It’s surprising how many interchangeable terms can confuse readers if not used consistently.

AI is a master at detailing drafts for readability- sometimes excessively so! Editors need to use extra caution, therefore, when proofreading text with AI. The submitted draft should sound human, and most of all, contain the elements of the original style of the author. ESL academic work invariably contains examples of awkward/unclear sentences and missing transitions. This is a great role for AI to play as it uses its powerful generative search engine to isolate such transgressions in phrases, and makes it easy for the editor to smooth them out while preserving the author’s voice. Finally, while preserving the author’s intent, AI is a natural guide to prevent, not enable, violating ethical boundaries such as ghostwriting and plagiarism. (On the note of plagiarism, editors have tools, including but not exclusive to AI, to check for plagiarism. This is for the protection of the author as well as the editor, and a good editor that encounters plagiarism would never do anything to harm the author, but would rather work out a means for revision.)

There are Many Things AI Cannot do (as of this writing!)

Just as you wouldn’t flay your Swiss Army Knife’s blade willy nilly in the dark, so too does AI require human supervision to be useful. Every author writes for a discipline, and it is the editor’s job to make sure that the author’s content is discipline-specific, as AI cannot always determine this. It’s probable that sometimes the editor may not be able to determine this either, but the editor will know when to seek guidance, whereas AI won’t (AI finishes its program and stops, while a human editor can and does revisit their work). Human editors have a great deal of experience and can recognize nuance specific to the discipline. Unless instructed, AI in general cannot pick up on cultural tone; the editor can adjust for this and maintain objectivity. AI has trouble discerning levels of formality and can often come across as ‘cold fish’ neutral in tone. Conversely, when AI does adopt a tone (humorous and light, for example), it often sounds exaggerated or even quirky. So yes, we editors have to edit AI as well. Another concern, obviously, is fact-checking. My advice is to never take facts directly from AI without checking them. Best practices are to use your (meaning the author’s) own sources so that you know that they’re correct within the confines of the journals being cited. It is fine to use facts generated by AI, just as long as they’re double-checked by research. Now, editors aren’t responsible for fact-checking as such, though they may be familiar with the topic. It is up to the author to make sure that they got their facts straight, which is worth bearing in mind and warrants a good deal of caution. It’s conceivable that an author would request fact verification for an extra fee, though it’s not common practice.

Practical Tips for ESL Researchers Who Use AI

At Planet Proofreading, we encourage researchers whose native language is not English to use AI, as AI generative text can be very helpful in getting you started in writing, especially for your transitions, introductory paragraphs, and conclusions. If English is not your native language, it can be very tempting to rely too heavily on AI for the bulk of your prose. Don’t do it! Revisit your paraphrasing skills and edit your work before sending it to your editor. It’s OK if there are many mistakes: that’s the editor’s job to fix! Here are some strategies for using AI in tandem with grammar/spell checkers and machine translation.

1. 1. Write in your native language, and have it translated into English using Google Translate. This would never be your final draft, but it is a good way for your editor to get to know your personal style and the subject you are working on. Communicate frequently with your editor as together you build a draft acceptable to send for submission or publication.
   2. Use AI to build a skeletal framework on which you’ll build your paper. Designate this as your ‘template’ (don’t worry- this is not plagiarizing- you’re using a modern tool to organize your project.)
   3. ESL researchers are almost always great readers, and I’m sure you are too! Use your reference texts both judiciously and abundantly, being careful to use quotation marks when quoting word-for-word,  and using in-text citations wherever you integrate information into your text from outside sources. Bear in mind that Chat GPT, Claude, and other AI can ‘speak’ your native language as well as English and can be a great consultant for getting your work clear and organized.
   4. 4.Back-translation: If you do use machine translation to translate your work from your native language to English, go ahead and edit your work and incorporate material from your references. When you’re satisfied, translate your English back into your native language. It won’t be perfect, but you’ll be able to confirm whether what you’re writing lines up with the meaning you’re trying to convey.AI grammar tools such as ChatGPT or Grammarly require human feedback to pass the Turing test, which gives readers assurance that the text is genuine and not just generated by a machine. This is important because while AI can sound polished and articulate (not consistently enough, however), the worst thing about relying on AI is that it can make factual errors that could end up having real-life consequences. AI stagnates growth and stifles ingenuity, which wastes the researcher’s time, to be sure, and the lesson from that is that AI serves its best purpose as a spotlight on the past, not as a forward thinker. The writer (you) must always be checking for originality. Every writer has a unique and authentic voice and style. It is definitely helpful to generate text from AI- but it must be infused with the authenticity of the author’s style. This is true even for the driest of scientific journals. At Planet Proofreading, we realize that many authors are now using generative AI to ease the writing process. As editors, we have always been aware of the need to correct in the author’s voice, be true to their intended meaning, and not to ghostwrite. In other words, if you use AI, there are two layers of separation between you and your work when it comes back from the editor. It’s our job to make those layers as seamless as possible, true to your original intent. You can feel assured knowing that we have been doing this for years, and we’re pretty good at it. At Planet, we always encourage responsible and collaborative use of AI. If we know how the author is using AI, we can stay ahead of any surprises that could come up in the final draft (such as ‘made up’ bibliographies with fictional titles) and unintended plagiarism. AI has not made editors like us redundant; if anything, it has increased our workload and has certainly raised our learning curve.

      Why not sample our work and see for yourself? We offer a 300-word copyediting sample absolutely free, with no pesky account forms or personal information required. Send in your sample today!