AI-Human Writing Divide: Pedagogical Considerations

Abstract

The emergence of Artificial Intelligence, or AI, has rapidly transformed the panoramic landscape of academic writing by defying the traditional ideas of authorship and creativity. This study examined the linguistic distinctions between AI-generated and human-written persuasive essays by analyzing five texts from each category. Utilizing computational tools like AntConc and UAM Corpus tool, along with qualitative methods, the study investigated key features of writing: lexical diversity, sentence complexity, nominalization, and the use of modals, epistemic, and discourse markers. Results indicate that human-written texts (HWT) demonstrated significantly higher lexical diversity, as measured by Type-Token Ratio (TTR), suggesting richer and varied vocabulary. Contrastingly, AI-generated texts exhibited a lower TTR but showed higher lexical diversity in keyword analysis, stressing methodological differences in assessing vocabulary richness. Sentence complexity analysis revealed that AI-generated essays tend to have longer sentences with more complex syntactic structures, while human texts contain a greater number of shorter sentences, indicating  a more direct communication. Nominalization patterns differ between two corpora: human writers used process nominalizations, highlighting focus on actions and processes relevant to the nature of persuasive essays, while AI-generated texts preferred quality nominalizations, stressing attributes and descriptions. Furthermore, this study also found out that AI-generated texts overused the two modals ‘can’ and ‘must’ compared to human texts. In contrast, human writers used a broader range of epistemic and discourse markers, which made their essays natural and had a nuanced expression of certainty, source attribution, and argument coherence. These findings emphasize the need for language teachers to emphasize the flexible and context-appropriate use of linguistic markers, expand students’ modal verb repertoire, and deepen their understanding of nominalization types to enhance sophistication in academic writing. This study also highlights the importance of nurturing creativity and critical thinking in writing instruction in light of AI’s growing role. In conclusion, the AI-Human writing divide presents both challenges and opportunities for academic writing instruction.