Are we reading the researchers or the artificial intelligence? Transformations in academic writing after the expansion of generative AI
TL;DR
This ex post facto study examines how AI has transformed textual quality in education research, and indicates a discursive shift: AI enhances accessibility and fluency but may compromise rigour and authorial distinctiveness.
Abstract
The rapid expansion of generative artificial intelligence (AI) has intensified debates on authorship and authenticity in academic discourse, yet empirical evidence remains limited. This ex post facto study examines how AI has transformed textual quality in education research. A corpus of 1,000 open access articles indexed in Google Scholar was analysed, comparing papers published up to 2021 (pre-AI) with those from 2024 onwards (AI era). Textual quality was assessed using a validated 25‑item instrument covering five dimensions: orthographic and grammatical accuracy, cohesion and coherence, adequacy to academic register, style and readability, and formal conventions. Results reveal significant differences. Pre‑2021 articles scored higher in accuracy, cohesion, register adequacy, and formal conventions, while post‑2024 articles excelled in style and readability. Findings indicate a discursive shift: AI enhances accessibility and fluency but may compromise rigour and authorial distinctiveness. These results highlight the need to reassess academic writing standards in digitalised contexts.