Just a few days ago, I read an article about how to manipulate AI Grading Algorithm by students. I think that is an interesting topic because what I am sure is that most of us at least once thought or wanted to change our scores throughout our education. I remember a moment that a new system (unfortunately I do not remember the correct name) entered our lives when I was at secondary school and it was possible to show our grades to our families without lies. Well, mostly I never worried about my grades until university – the score of the first exam of statistics class was terrible in the first year of my university and I wish I had a chance to change it 🙂
As you know the advent of technology transformed our lives and students in the United States have taken advantage of using this technology to cheat an algorithm marking their work. According to it, the student’s grades have risen to almost good scores. Educational institutes have been started to involve algorithms in their systems to gauge their students’ progress and to do exams. Even though the system evolves into technology-based education more than before during COVID, also it was used before COVID. Last year when I was studying in Canada we had a lecture called Data System Architecture and our exams were computer-based. After finishing exams, we were able to get our score immediately and also to see our correct and false answers. I still have doubts whether this way works well for everyone and every class because at the end of the day it is an algorithm and it’s developed by data, so there is a risk for being biased. Therefore, it might not be suitable for every lecture.
If I turn back to how this issue discovered is that one child received a 50 percent grade on an assignment and the thing is this child was talking about how much he loved his History teacher and how it is possible while loving that class having low scores.
So, how this algorithm works then? Of course, they looked for the answers, and if activities include keywords that are used for determining a system-assigned score, the student will earn a zero percent if none of the keywords are included in the response, and will earn 100 percent if at least one keyword is included in the response. On this website, you can find more detailed information about how it works: https://help.edgenuitycourseware.com/hc/en-us/articles/360042418854-Auto-Grading-Method-Comparison
What is a big issue? As I said before algorithms use data and this data should be a large dataset to analyze better. Basically, they are supposed to be educated by the quantified and qualified dataset. Otherwise, the system works biased and for example, if the data has niche subjects such as Latin characters, the grade might be biased for people who take the exams without Latin characters. Also, according to the news, rather than analyzing keywords as Edgenuity did, the algorithm mediated grades based on the historical success of the school – this is also the unwanted result.
Another significant point is economic and racial biases in society. No matter where you live or what you do, in daily life people suffer from these topics. Moreover, it is an inadmissible thing that having it in the educational life from my perspective. According to Stephen Curran, a teacher and education expert, “children from a certain background may find their assessment is downgraded” – this is really sad.