This exploratory analysis investigates the integration of ChatGPT in self-directed learning (SDL). Specifically, this study examines YouTube content creators’ language-learning experiences and the role of ChatGPT in their SDL, building upon Song and Hill's conceptual model of SDL in online contexts. Thematic analysis of interviews with 19 YouTubers and relevant video contents reveals distinct constructs of ChatGPT-integrated SDL, suggesting a reconceptualization and refinement of the SDL framework in the consideration of generative artificial intelligence (AI). This framework emphasizes critical aspects of utilizing ChatGPT as an SDL tool on two distinct levels: 1) the interactive relationships and interplay between learners’ personal traits and their ongoing learning processes (local) and 2) the evolving nature of SDL in the rapidly advancing landscape of generative AI, with socio-political-cultural foundations of AI constantly shaping the learning environment where SDL occurs (global). The study highlights the potential of ChatGPT as a tool for promoting self-directed language learning (SDLL) and provides implications for the development of learning technologies and research on AI-facilitated SDL.

Since late 2022, dozens of YouTube channels focusing on a diverse array of topics related to language learning with generative AI tools such as ChatGPT have rapidly emerged. This study explores the implementations and perspectives of YouTube content creators who now constitute an increasingly important segment of the ecosystem of language teaching and learning. A mixed methods netnographic approach was employed, combining qualitative and quantitative techniques. A total of 140 videos were identified and analyzed, and an in-depth content analysis was conducted to uncover underlying themes. Four main categories of creators were identified: educators, learners, technology professionals, and e-learning providers. Educators, especially English and Japanese teachers, were the majority, followed by learners and technology field professionals. This study highlights the benefits, drawbacks, and concerns associated with the integration of AI tools in language learning. By examining this rapidly evolving phenomenon, the study contributes towards an understanding of the role and impact of generative AI tools in language education.

ChatGPT is a state-of-the-art generative artificial intelligence (AI) chatbot released by OpenAI in 2022. It simulates human conversation and has the capability to generate different texts at various levels of sophistication in near real time depending upon the user’s skill in creating prompts. While concerns have been raised about academic dishonesty and cheating among students, ChatGPT has significant academic potential for education, particularly in the field of language learning. This research explores the potential of ChatGPT in supporting and empowering Chinese language learners (CLLs) whose first language is English to enhance their writing skills, mainly focusing on the research question: Is there a functional relation between Chinese language learners from low‐income families using ChatGPT after school twice a week and improvements in their Chinese writing? Four participants with varying language proficiency levels were recruited, and their data were analyzed using an ABA design. Over three weeks, they utilized ChatGPT twice a week for approximately 20 min each after school. The students’ writing scores, writing samples, and learning reflections were used to triangulate the data and enhance the data’s trustworthiness. The findings indicate that (1) each participant made a noticeable improvement in their Chinese writing scores during the intervention and reversal phases; (2) ChatGPT played a crucial role in correcting errors and facilitating the development of complete sentence structures; and (3) the students expressed a sense of empowerment through their interactions with ChatGPT. These findings highlight that ChatGPT shows promise as a supportive tool for CLLs from low-income families, reducing educational inequality and promoting equitable access to language learning opportunities.

ChatGPT's ability to realistically mimic human conversation and its high level of ability to handle linguistic ambiguity opens new and exciting avenues in language learning. Building upon the technical affordances of ChatGPT, this study explores the perceptions of educational affordances when incorporating ChatGPT across languages discussed by language communities on YouTube and identifies best practices for its effective use in language education. Through inductive content analysis, this study discussed 18 languages categorized into four groups: (1) romanized languages with high resources, (2) non-romanized languages with high resources, (3) languages with medium resources, and (4) less frequently used languages. The findings reveal consensus that (a) ChatGPT is a valuable and remarkable tool for language teaching and, (b) learning and it cannot fully replace teachers, as humor, wit, and sympathy cannot be programmed. Two potentially significant issues or two gaps were identified and discussed: namely, the learning optimization gap and the knowledge comprehension gap.