top of page

Using AI in Academic Writing

Writer's picture: KathyKathy

As an Academic English instructor, I’ve had the privilege of working with thousands of International Business students over the past 18 years, mainly non-native speakers.


My tutorials focus on the characteristics of academic writing and on helping students improve their writing skills through reading, language analysis, targeted exercises, and personalized feedback on drafts. For years, it was an effective, established process that helped students improve their final papers and build their writing skills.


But then, ChatGPT arrived and suddenly, everyone had access to a 24-hour assistant capable of writing pretty much anything, including seemingly perfect texts. Emphasis on seemingly.


After reading as much as I could and testing AI on various aspects of writing (e.g., language generation, language analysis, paraphrasing, English proficiency, revision and feedback, summarization, and plagiarism),


I experienced firsthand the potential benefits and pitfalls. So, I began supplementing my tutorials with AI Do’s and Don’ts and created an AI in Academic Writing Workshop.

“When used properly, AI can be an exceptional learning tool, empowering students to develop a range of genuine human skills such as critical thinking, reading comprehension, paraphrasing, tone, grammar, vocabulary… and the list goes on!”

Do’s


1. Use AI to augment your revision process, not replace it. When you ask a question, have a real conversation with the AI and ask it to explain its suggestions.

Always aim to understand the rationale behind each suggestion to enhance your understanding and learning. If you disagree with something, don’t be afraid to push back and ask for an explanation or for alternative ways to say something.


2. Be very selective in how you use AI enhanced text. Use AI to refine your natural voice rather than overshadow it. You want your writing to sound like you and not like a generic, artificial version of you, one that uses overly advanced vocabulary or terms that are commonly associated with AI-generated texts. 


Don’ts


1. Don’t be tempted to a) copy/paste AI written texts into your paper or b) run very large chunks of text. First, write your text yourself, presenting your ideas and how they fit together in your own words (using techniques we discussed in class). Once you've got that foundation, then turn to AI to help tweak sentences that you feel need improvement. Don’t overuse it and sacrifice your own voice.


2. Don’t forget that AI can produce errors and misinterpretations. AI may not always understand context or nuance in your text or in your prompts, which can lead to inaccuracies in your academic writing. I have seen this happen many times! 


3. Don’t forget to cite your AI sources (check with your university for specific guidelines). Also, be aware that AI may suggest fictional academic sources in its responses. Be sure to verify authenticity and reliability of information.


AI and Academic Writing Workshop


I designed a workshop to offer students firsthand experience in 1) exploring the result of using various prompts, 2) opportunities to enhance their academic and general English skills, and 3) insights into the limitations and potential risks of AI.


Workshop Activity 1


Prompt Experiment

Everyone in the group logs into an AI (ChatGPT, Copilot, Gemini etc.) and asks for assistance with one particular sentence. We don’t share our prompts until we get the result.


It’s an engaging and insightful exercise to see what happens to the same piece of text when everyone asks their own questions, from general requests to improve overall English to detailed questions about specific grammar constructions, more appropriate synonyms, nuanced word choices, etc. 


You can try this yourself with friends and colleagues. But it's the next step that's critical. What do students do with these results? My role is to assist them in learning how to fine-tune their prompts and how to be selective -- to pick and choose what resonates with them, understand their preferences in terms of formality and language level, and to become familiar with their own voice. This self-awareness is a key aspect in truly harnessing the power of AI for language development. 


“AI can make your English better, but I’d also argue the opposite: good English makes AI results better.”

Workshop Activity 2


Text transformation: AI is good but it can’t read your mind (not yet, anyway)

Below you’ll find three versions of the same text with notes below each. The first text is a real-life example from a student’s paper. The second was written by AI when it was asked it to improve the student’s text. The third is a result of a human reviewer who took a few of the AI’s suggestions but decided to add her own text to enhance credibility and clarity.


1. Student’s Original Text 


It is examined what words employees use to describe male and female leaders and about if man or women score higher in leadership skills.


Feedback: The original sentence is awkward, lacks clarity and contains a question word phrase (“what”) instead of a noun-based phrase as well as incorrect vocabulary.


2.   ChatGPT’s Rewrite (prompted with “improve the academic English”)


The study endeavors to scrutinize the linguistic patterns employed by employees when characterizing male and female leaders, alongside assessing whether men or women attain superior ratings in leadership competencies.


Feedback: The AI version somewhat improves clarity and flow, but much of the vocabulary is overly advanced and doesn’t match the student’s own voice.


3.   Human edit (for example)


In his 2018 study, Smith conducted an analysis of the vocabulary used by employees to characterize male and female leaders. The research scrutinized whether there were disparities in performance between men and women in a standardized assessment of leadership aptitude.


Feedback: In this version, the writer is beginning to infuse the text with her unique voice, adding a human touch that the AI didn’t suggest. By incorporating her own ideas and style, she effectively emphasizes credibility and clarity, complementing the suggestions from AI (e.g., the term scrutinize). Her decision to divide the sentence into two parts provides a clearer presentation of the study's distinct objectives. Her inclusion of a source (Smith, 2018) further enhances the trustworthiness of the information.


This is just the beginning…


As we look to the future, it's clear that AI's impact on education will grow, requiring us to stay one step ahead of technology. A blend of traditional and AI-driven methods is already enabling us to provide students with a learning environment that's more personalized than ever. 


Through engaging and interactive writing activities, students not only gain the experience they need to better navigate AI but also learn how to reflect on their own particular skill sets. This empowers them to enhance their writing ethically and responsibly, choosing AI results that align with their academic goals without compromising their uniquely human voices. 


 _____________________________________________________________________________________

If you’re interested in seeing a demonstration of the workshop on Using AI for Academic Writing, or if you’d just like to have a cup of coffee and share ideas about this subject, feel free to contact me!

26 views0 comments

Comments


bottom of page