Artificial Intelligence Generated Student Essays – Should we be worried?
Oliver Thomas, School of BioSciences, The University of Melbourne
The increasing prevalence of artificial intelligence (AI) generated essays in tertiary education is a cause for concern. These essays are often of a poor quality and can contain plagiarism. AI generated essays also have the potential to be used for contract cheating, which is a serious academic offence.
In this blog post, we will explore the side-effects of AI generated student essays in a tertiary education environment. We will also look at the potential solutions to this problem.
As more and more schools move to online learning, there is an increasing need for automated essay generation. However, there are a number of problems with this solution. First, artificial intelligence (AI) generated essays often seem robotic and lack the creativity of human-written essays. Second, AI essays can be easily detected by plagiarism checkers, which can cause problems for students who are trying to cheat. Finally, AI essays often contain errors that need to be corrected by a human editor.
In addition to this, AI generated essays are often repetitive and lack originality. This can lead to plagiarism issues, as well as to students being less engaged with their studies. Additionally, AI generated essays often contain errors, which can lead to students receiving lower grades. Finally, and perhaps most importantly, the use of AI to generate student essays can lead to a decreased understanding of the material being studied. If students are relying on AI to do their work for them, they are likely to have a poorer understanding of the concepts being taught. This could have a negative impact on their later studies, as well as on their future.
There are a few ways to combat the problem of AI generated student essays. One way is to design your assessment tasks in such a way that it is difficult for AI to generate a response. For example, you could ask open-ended questions that require critical thinking and analysis. You could also use essay topics that are current and relevant, as AI is less likely to be able to generate a response on these topics. Another way to combat AI generated essays is to use plagiarism detection software. This software can help to identify essays that have been generated by AI and can flag them for review. Finally, you can also talk to your students about the issue of AI generated essays and let them know that you are aware of the problem. By raising awareness, you can help to combat the problem and ensure that your students are not using AI to generate material.
Plot twist
OK, so the above writing was pretty coherent, if a little boring. It’s definitely a little repetitive. I generated the above text using an Artificial Intelligence blog content generator. Although the text is a little “robotic”, it still managed to hit the major issues of academic misconduct through AI generated prose. I generated the above text by going to www.simplified.com, and entering a few prompts (i.e. words/sentences to base the AI generated content on) for blog sections. The whole process took me less than 5 minutes and was possible using their free version. If I was prepared to shell out the ~30 AUD a month, I’d have access to far more “bells and whistles” – better control over AI content generation, a larger number of words spawned and so on. Just think, if I can do this for free, within 5 minutes…what can our students do when preparing assessment tasks that they have little content knowledge about.
Let’s try an example. I’ve used as a prompt, from a 1972 Zoology examination question, “Discuss ciliary locomotion in Metazoa”. Here is one of the responses it gave me;
Metazoans are a subgroup of animals characterized by having cilia. Cilia are fine, hairlike structures that protrude from the cell surface. These structures are used by the cell for locomotion, feeding, and sensing the environment.
The ciliary locomotion of metazoans is achieved by the coordinated beating of cilia. This process is known as metachronal wave locomotion. In metachronal wave locomotion, the cilia on the cell surface are arranged in rows. The cilia in each row beat in a coordinated fashion, creating a wave that propels the cell forward.
While ciliary locomotion is most commonly associated with metazoans, there are a few other cell types that use this mechanism for locomotion. For example, some plant cells, such as those of the moss Physcomitrella have ciliary locomotion.
A bit worrying, no? Would you have scored a student zero if they produced this answer? Consider this; we devote a large amount of resources to using plagiarism checking software (e.g. Turnitin), which is effective at picking up direct copying from sites (incidentally, I ran this whole blog post through Turnitin, and it returned a 0% similarity score). Our ability to determine collusion and/or contracting cheating is far more difficult. To be able to confirm that another person has written part, or the entirety, of an assessment item requires proof that there are two bad actors at play – the cheater, and the colluder. If, however, the “colluder” is an artificial intelligence program, and the content generated is “unique”…the problem becomes excruciatingly hard to control.
I don’t have a lot of answers to this problem, but I do have a whole lot of concern. To cheer you up, the banner artwork above is an AI artwork that I generated using midjourney, employing the prompt “thylacine resurrection”.
Subscribe to the BioSciences Educator Blog – BRITE Ideas Blog