Plaintext dataset of short essays generated using ChatGPT 5.3

This dataset contains 50 AI-generated short essays created using ChatGPT 5.3 for research into the linguistic and discourse properties of large language model generated text. The dataset is intended to support qualitative and corpus-assisted analysis of how recurrent structural patterns in LLM outputs may guide reader interpretation. The essays were generated in response to 50 researcher-designed prompts across five discourse categories: argumentative/evaluative prompts, comparison prompts, explanation prompts, problem/solution prompts, and definition/conceptual prompts. The prompts address topics relating to technology, language, communication, digital literacy, artificial intelligence, online interaction, misinformation, and mediated communication. The generation instruction used was: Write ten individual essays of approximately 800–1200 words in response to the following prompts. Do not use headings or bullet points. This instruction was followed by batches of prompts. The resulting essays are presented as generated, without substantive editing. The files are provided as plain text (.txt) files and use UTF-8 encoding. This dataset does not contain human-authored private communications, personal data, or third-party identifiable information. The prompts were researcher-created and the essays were generated specifically for research purposes.