AI-assisted language assessment in the post-ChatGPT era: A systematic literature review

The present review synthesizes peer-reviewed journal articles on AI-assisted second language (L2) assessment published between 2023 and July 2025. Following the PRISMA process, 81 articles were analyzed to examine the key characteristics of AI-assisted L2 assessment research, the employment of validity frameworks, and the implications of the findings. The synthesis indicates that studies published after the emergence of ChatGPT (OpenAI, 2022) have largely relied on generative AI, particularly ChatGPT, rather than other supervised AI technologies. Many studies focused on English as the target language, assessed the writing construct, examined foreign language contexts rather than second language contexts, and were conducted in postsecondary settings. The studies primarily investigated agreement between AI and human performance, followed by the impact of AI on language learning and comparative evaluations across AI approaches. AI was most frequently used to score assessments, provide both scores and feedback, and generate test items. AI was employed for both assessment for learning and assessment of learning. The vast majority of studies did not explicitly refer to any validity framework. Although many studies acknowledge the potential of AI-assisted L2 assessment, they rarely argue for its use beyond local contexts. Research gaps and implications for future research are discussed.