Utilizing Artificial Intelligence in Academic Work

We have been talking a lot lately about the advancement and proliferation of artificial intelligence (AI) technologies. While we initially used them for entertainment, socializing, or work, we have since begun to use them for educational purposes in the academic field as well. 

At first, we saw AI as a helpful servant. Now, we are saying that this servant could be the killer of humanity. We are afraid of it. The reason we say this is that, unlike mechanical machines, AI has reached a point where it can perform the “intellectual tasks” that humans would otherwise do. When you use ChatGPT-4o, one of the most advanced examples of AI models created by developing Large Language Models (LLMs), you can chat with the AI as if you were talking to a friend. At the same time, you can converse as if you were speaking with a scientist, a business person, a philosopher, or a politician. You do not need to know these individuals personally, nor do they need to set aside their time for you. 

AI has rapidly become one of the most important and transformative technologies of our time, with applications in almost every field and industry. Among these applications, academic writing stands out as one of the fastest-growing areas where AI-based tools and methods are most widely accepted. I am not saying this out of a sense of convenience or practicality, as if it were useless to stand against a rushing flood, but I believe that the use of AI-based tools in scientific writing should be widely adopted. 

So why? In their June 2023 article in the journal Nature, titled “Academic Writing with Artificial Intelligence: A Paradigm-Shifting Technological Advancement,” Roei Golan and his colleagues categorize the use of AI in academic writing into two broad categories: tools that support authors during the writing process and tools used to evaluate the quality and validity of written work. Tools capable of understanding and generating human-like language, such as natural language processing, can help authors write and prepare their articles. Tools like plagiarism detection software and automated peer-review platforms can provide support to reviewers and editors in the process of evaluating the quality of articles. Furthermore, automated peer-review platforms can evaluate large volumes of articles quickly and objectively, which has the potential to reduce workloads.

One of the most significant advantages of AI-based tools in academic writing is that they save time and increase efficiency. For example, natural language processing algorithms can help authors identify and correct errors in their work, allowing them to focus on the content of their writing. These algorithms can also assist with tasks such as language translation and text summarization, making comprehensive literature reviews more efficient. Additionally, natural language processing algorithms can generate specific drafts for articles, research protocols, grant applications, informed consents, emails, medical necessity letters, reports, and other written documents. These drafts can be used as frameworks that ensure important components of the text are included. Finally, language processing can provide specific suggestions to strengthen an article or abstract. For example, AI algorithms can suggest previous relevant studies that should be included in the introduction or limitations that should be mentioned in the discussion sections. However, current limitations of AI algorithms include restricted access to all publications and an inability to recognize the most up-to-date studies. Furthermore, AI can suggest an appropriate sample size based on previous studies or recommend the most suitable statistical test based on the distribution of the study. Therefore, it can act as an editor, reviewer, and even an advisor in academic work. The point to remember here is that it is still the researcher's duty to create the command sequences (prompts, or we could call them text-based algorithms) that explain to the AI how to perform editing, reviewing, and advising, and to develop and improve these for each study. Because there is no single prompt that is suitable for every researcher or every study. 

Finally, I would like to share a few notes that I consider important with academics who use, do not use, are thinking about using, or are not thinking about using AI tools:

1. Students now have easy access to advanced AI-based tools like ChatGPT. These tools use Large Language Models (LLMs) and can generate original written content that students can use in their assessments.

2. These tools are often accessible through commercial services aimed at ‘helping’ students with their assessments.

3. Outputs generated by LLMs are consistent enough to be undetectable by academics or traditional text-comparison software used to detect plagiarism, although unaltered fake references may indicate the use of these tools.

4. The use of these tools may not necessarily be considered plagiarism as long as students are transparent in their submissions; however, it may violate the academic integrity policies of Higher Education Institutions (universities).

5. There are legitimate uses for these tools to support student education, which requires universities to carefully consider and determine their policies regarding students' use of this software.