The widespread use of generative AI models has ushered in a transformational era in the field of academic writing. As its use becomes more and more prevalent amongst students and academics, educational institutions have little choice but to adjust their teaching methods to facilitate the use of such powerful mechanisms.
During my bachelor programme I attended a course called Digital Business. One of the most intriguing aspects of this course was the fact that the coordinators have adjusted the syllabus to adapt to students using generative AI for assignments. When it came to the assignments we were presented two options, to use generative AI for writing and receive part of the assignment grade based on the quality of our interactions with the AI, or to refrain from using it altogether but be subject to AI detectors and potential plagiarism claims. In this article, I wanted to present some pro and contra arguments to this topic and link it to my experience.
So, should students use AI for their assignments?
According to Fok & Weld (2023), there are five main ways of interacting with AI to improve writing. First, students can use large language models to generate ideas for their writing. This can be taken further by students guiding the AI through the generated ideas and requesting an initial write-up of the text. Second, students can always rely on its ability to take existing writing from them and add to the beginning, the middle, or the end of a paragraph writing sentences according to the prompter’s writing style. Third, given that students have trouble fully comprehending some written information, AI can assist them by elaborating on the given text and its content. Fourth, students who do not feel confident with the level of their English proficiency can request AIs to recommend alternative words and writing structures for a given segment of text. Lastly, one of the most important features of student-AI interaction is the capacity of writing assistants to present critical feedback on all aspects of the text written by students and to ask questions that can further improve the quality of the final output.
Or should they be wary of using it?
Opponents of using AI for academic writing mainly focus on its questionable nature with regard to the originality of paper and referencing. Some argue that since AI was trained on a a large data set of writings by other authors, it is inevitable that the generated content will be to some degree based on someone else’s work, without attributing credit to the original writers. Therefore, they believe claiming the output of AI as your own work could be viewed as an infringement of intellectual property (Frye, 2023). However, others have claimed that this is incorrect, as AI does not generate precise duplicates of pre-existing creations, they just rely on NLP to create something that builds on the structure and patterns of human writing, and thus is not significantly similar to the works of the others it has been trained on (Frye, 2023).
How did I do it in my bachelor programme?
As I have mentioned above, during my Digital Business course we were instructed to use generative AI to guide us through the writing process. We were first requested to ask the AI to write the initial text solely based on the title of our topic. Then we had to proceed with interacting with the AI after being given a set of prompts the teachers recommended we use. It mainly included prompts asking for feedback or overall improvement concerning the content or the writing style. What I found the most efficient way was to break down the text into paragraphs, and then one-by-one have the AI write the text given what I thought would make sense to incorporate. Once the whole raw text was done, I proceeded with asking for critical feedback for each paragraph and incorporating the chatbot’s recommendations. Unfortunately, I no longer have access to the exact outcome of the research our teachers were conducting with this assignment, however, they claimed that the assignment grades have significantly improved compared to last year’s grades.
References:
Frye, B. L. (2023). Should using an AI text generator to produce academic writing be plagiarism? Fordham Intellectual Property, Media & Entertainment Law Journal, 33, 947. https://ssrn.com/abstract=4292283
Fok, R., & Weld, D. S. (2023). What can’t large language models do? the future of AI-assisted academic … https://cdn.glitch.global/d058c114-3406-43be-8a3c-d3afff35eda2/paper4_2023.pdf
Interesting! I agree with the cons like working on top of a initial write-up of the text to improve it. I also have written a blog about if ChatGPT is affecting the critical thinking, which is close to your topic I believe. Reading your blog I did wonder if the fact that a written text with the help of AI will create a huge discrepancy with the way you speak. For example, someone who is not very good in English, is it okay for that person to have high quality papers since the speaking end writing part are not in terms of quality close to each other? I don’t think it wont be a problem for an assignment for school, but the question does rise for me if it will be challenging when you are for example when you give classes during your PhD.
I really enjoyed your structure and the topic itself in the blog post, as I think it is a super relevant topic. My first few interactions with ChatGPT were indeed of more academic nature, e.g. prompting to write some things about my Bachelor thesis topic or highschool assignments. I really think it depends on how you use these tools, similarly how we learned how to google properly and critically evaluate the sources and reliability. There is a huge difference between simply taking ChatGPTs text and copying the result and thinking through structure and content and extending it using ChatGPT. Especially since we prompt further and further to create our desired result, which means we actively evaluate and think critically, what makes sense to incorporate and what not. I was shocked, how well and fast ChatGPT could come up with key concepts that I have found through a lot of work. For me, ChatGPT is a great information gathering (similar to google but different in its approach) and idea generating tool, for which I will continue to use as I feel like it has increased my efficiency and quality by a lot. I also agree with Ayse, that there are drawbacks e.g. relying on it but also the missing credit, which is why i would use it in research to expand on what concepts i could be looking for.