Currently, AI-generated writing assistance offers several benefits, including time savings in preparing teaching materials, ease of incorporating specific objectives into instructional guides, and the creation of academic activities that develop skills. On the other hand, its use in the creation of academic texts can sometimes represent a lack of creativity or authorship, even plagiarism. However, in certain situations, human-generated texts are identified as AI-generated. To understand these cases, this article uses a Large Language Model (LLM) to perform different tests and address the following objectives: 1) identify the AI patterns used to generate texts; 2) analyze the AI patterns to detect whether a text is AI-generated or human-generated; and 3) create a text with the AI patterns and corroborate its origin. Among the results, it was found that statistical and stylistic patterns are used to detect whether AI-generated text can be distinguished from human-generated text. On the other hand, although there are general AI patterns for creating texts, there are also specific AI patterns depending on the area of knowledge being studied. This work was conducted during the Machine Design course, held from February to June 2025, for Mechanical Engineering students, when they consulted an LLM to develop their design projects. To understand students¿ general perceptions of using LLMs to detect AI-generated texts, a survey was administered at the end of the course.
