The use of artificial intelligence in education has led to a new problem. A recent report has revealed that students are using OpenAI’s natural language model, ChatGPT, to cheat on their homework. The model has the ability to generate human-like responses to questions, making it easy for students to fool their teachers.
The AI model, which is designed to answer follow-up questions and challenge incorrect premises, allows students to ask it to write an essay on a given topic. This essay can then be presented to the teacher as the student’s own work. The AI model can even be asked to write in a way that avoids detection by anti-cheating software.
The issue has been acknowledged by OpenAI, which has created an AI Text Classifier to detect whether a piece of text was generated by ChatGPT or not. However, the tool has been found to be only 26% accurate in identifying AI-generated text. This makes it difficult for teachers to approach students when they suspect plagiarism, as they do not want to accuse innocent students of cheating.
To address the problem, OpenAI is updating ChatGPT’s ethical responses to prevent it from being used for cheating. The company may issue warning responses or even refuse to answer a question if a student tries to manipulate it to avoid detection. Meanwhile, anti-cheating software company Turnitin is working on a new service that will accurately determine whether a student has used ChatGPT to complete an assignment.
According to experts at Turnitin, the service will rely on the fact that ChatGPT writes in a very average way, while human beings have idiosyncrasies that make their writing unique. This will enable the service to differentiate between work done by a student and work generated by an AI model.
The issue of AI cheating in education is a serious one, and it is crucial for educators to be aware of it. As technology continues to advance, it is likely that new solutions will be developed to combat this problem and ensure that students are held accountable for their own work.