Potentials and Limitations of AI as an Academic Writing Assistant for Non-Native English Speakers: An Assessment Framework for ChatGPT-Generated Texts

บทคัดย่อ

The use of Artificial Intelligence (AI) tools like ChatGPT-4 in academic writing has gained much attention, especially among non-native English speakers. This study looks at the strengths and weaknesses of ChatGPT-4 as a writing assistant by introducing a framework that uses a scoring rubric and software tools. The results show that ChatGPT-4 can produce well-organized, accurate, and relevant academic content. However, it still struggles with improving readability for non-native speakers and ensuring sources are trustworthy. The evaluation process, which included Flesch Reading Ease (FRE) scores and similarity checks, highlights the need for careful review when using AI-generated content for academic work. The proposed framework provides a clear method to make the most of ChatGPT-4’s strengths, but some challenges remain. These include the subjective nature of rubric scoring, the shallow focus of FRE scores, possible biases in software tools, and the fact that the study was done in a controlled and unchanging environment rather than reflecting real-world situations where user interactions and needs can vary. Despite these issues, the findings show that non-native English speakers can use ChatGPT-4 effectively for academic writing when combined with proper evaluation methods, ensuring their work is both high-quality and credible.