The rapid advancement of automatic text generation has transformed various domains, promoting its use in professional, academic, and personal activities. However, it has also introduced significant challenges, such as the spread of fake news, identity impersonation, and misuse in academic evaluations. Despite the increasing prevalence of these technologies, there remains uncertainty about the human ability to distinguish between texts generated by artificial intelligence and those written by humans. This study evaluates this ability through an application that presents pairs of similar texts: one written by a human and the other rewritten by GPT. The texts, written in Spanish, were sourced from various domains in Ecuador, including informal social media posts (X/Twitter), formal news articles, and academic thesis summaries. The analysis of results considers variables such as gender, educational level, and text domain. The findings provide a baseline for understanding the current state of human perception regarding automatically generated texts and underscore the importance of developing models and tools to enhance detection capabilities, addressing the ethical and social challenges posed by these technologies. This research establishes an important foundation for justifying and guiding future work in recognizing automatically generated texts, as well as for the development of artificial intelligence models that can serve as detection tools.