Utilization of AI Detection Application Software Packages for Quality Research Integration and Content Assessment in Mathematics Education

Abstract

The effectiveness of AI-detection technologies in identifying variations in content generated by AI and humans is assessed in this study. In particular, fifteen human-paraphrased and fifteen AI-paraphrased human-generated and AI-generated contents were analyzed. To rate the documents, eight (8) free AI-detection tools were chosen at random. The findings show that, depending on the kind of information being examined (human-paraphrased or AI-paraphrased), there are significant variations in the effectiveness of AI-detection methods. While some tools, like OpenAI Text Classifier, are better at identifying human-paraphrased content, others, like Copyleak and Sampling, are excellent at identifying AI-paraphrased content. The study further shows that manual paraphrasing of human-generated content can reduce the accuracy of AI detection tools. Conversely, AI-generated content remains detectable even when paraphrased using AI, with no significant impact on detection tool performance. These results highlight how crucial it is to use the appropriate AI tool depending on the particular kind of content that needs to be examined to get reliable detection outcomes.