In a world where Artificial Intelligence is advancing rapidly and it becoming more difficult to tell human made to AI generated, we never wonder how students might be affected by this. How do we know if students won’t end up being wrongly accused of using generative AI?
For this article I wanted to have more background knowledge on AI and how it works, specifically AI detectors, after a while I found that AI identifies machine written text by inspecting patterns and measuring the perplexity and burstiness of a sentence or text. (how predictable the text is and how the structure varies)
AI detectors are not always correct and oftentimes can make mistakes, like for example when a piece of text is too short it can often lead to a mistake with the AI detector, due to lack of context the piece of information can be interpreted wrongly and lead to it being marked as AI made. Not only can lack of context cause AI to make a mistake but also, different writing styles. For example AI detectors can see the writing of a student still developing their writing skills or a creative writers work and it might confuse the detector. This would be because these detectors look for patterns that they were made to look for, so when they can’t find one they will automatically flag it as AI.
So for students that might just have different writing styles or students that have trouble sometimes with writing they could end up getting a zero on something that they did spend time working on. Even though AI plagiarism is a serious problem that teachers have to face there has to be ways to check for AI plagiarism that has a better accuracy rate than using an AI checker.