Physics Chatbots into Blended Learning: A Bibliometric Analysis (2020–2025)

Abstract

Physics learning, particularly on Newton’s laws, often faces challenges due to the abstract nature of the material and the limited interactivity of traditional media. Advances in artificial intelligence (AI), especially generative chatbots, provide new opportunities to strengthen blended learning approaches that combine face-to-face instruction with online environments. This study aims to map the development of physics learning media based on blended learning with chatbot assistance during 2020–2025, while examining their effectiveness and limitations in supporting students’ understanding. A descriptive bibliometric analysis was conducted using data from the Scopus database, with the keywords “blended learning,” “chatbot,” “physics,” and “Newton’s laws.” Bibliographic data were exported in BibTeX and CSV formats, then analyzed with the bibliometrix package in RStudio through the Biblioshiny interface. The analysis covered annual publication trends, average citations, relevant keywords, and keyword co-occurrence networks. The results show a significant increase in publications since 2023, driven by the emergence of generative AI technologies such as ChatGPT. Although the number of studies remains limited, their distribution across journals and international proceedings highlights the interdisciplinary nature of the field and the strength of global collaboration. Keyword analysis reveals a balance between technological aspects (chatbots, AI, machine learning) and pedagogical dimensions (students, education, learning), while also indicating the absence of standardized vocabulary and consolidated theoretical frameworks. In conclusion, the integration of physics chatbots into blended learning is still relatively new but holds great potential, particularly for the underexplored topic of Newton’s laws. These findings provide an original contribution to the literature on physics education and learning technology, while opening pathways for future global research on adaptive AI-based learning models. A limitation of this study lies in its reliance on Scopus alone, suggesting that further research should broaden database coverage for a more comprehensive perspective.