Writing without borders: AI and cross-cultural convergence in academic writing quality

English has become the dominant language in global academic publishing, facilitating cross-border collaboration while reinforcing structural barriers for non-native English-speaking researchers. This study examines the evolution of academic writing quality in social sciences abstracts from 2012 to 2024, focusing on disparities across linguistic, regional, economic and gender-based classifications. Using over one million English-language abstracts retrieved from the Web of Science, the study evaluates writing complexity through readability metrics such as the Flesch-Kincaid Grade Level. A mixed generalised linear model (GLM) is employed to identify key factors influencing writing quality, with particular emphasis on internet access. To assess the potential impact of large language models (LLMs) such as ChatGPT, the analysis incorporates a lexical tracking approach that monitors the frequency of adjectives and adverbs commonly associated with AI-generated content. The findings reveal a global improvement in writing complexity, with non-native English-speaking countries showing notable advances. China, initially lagging in English writing standards, has surpassed traditional leaders such as the United States, signalling a shift in global academic communication. Enhanced digital infrastructure and the adoption of AI-assisted writing tools appear to play a contributory role in this convergence. These results offer empirical insights into how technological advancements are reshaping scholarly expression and mitigating long-standing linguistic and structural disparities. The study provides evidence-based guidance for policymakers, educators and research institutions seeking to enhance the accessibility, inclusivity and quality of academic writing across diverse global contexts.