Keyword: «large language models»

The use of artificial intelligence technologies and neural networks for educational purposes is a current trend. Since the launch of the Chat-GPT chatbot in November 2022 by OpenAI, which is a large language model as a type of generative artificial intelligence, educational researchers, psychologists and representatives of other fields of humanities and natural sciences from different countries have turned to the analysis of technological functions, didactic potential, advantages, disadvantages and risks of its practical application. The number of studies on this topic is steadily increasing every year. The aim of the publication is to identify pedagogical, linguodidactic and psychological conditions for the effective and safe use of Chat-GPT in the higher education system. The leading methodological approaches that the authors relied on are a systemic approach, an express review and critical analysis of scientific sources, as well as elements of bibliometric analysis. 150 publications of Russian and foreign authors covering a variety of issues and the scale of use of this chatbot in the field of education gave the material for the review. As a result, the strengths and weaknesses, opportunities and threats of using Chat-GPT as an educational technology were identified. The novelty of the study lies in the fact that for the first time, based on the analysis of a large volume of the latest scientific sources (2023-2024), the opinions of foreign and Russian educators regarding the risks of using neural networks in education were systematized, and recommendations were proposed for optimizing educational activities through the use of chatbots with generative artificial intelligence in university practice. The theoretical significance of the study lies in the generalization and classification of key problems associated with the implementation and testing of Chat-GPT in the educational process, as well as in the description of a number of significant pedagogical, linguodidactic and psychological factors that can prevent or eliminate the negative consequences of using chatbots and other neural networks in higher education. From a practical point of view, the article is of interest to educational researchers and representatives of the management level in the field of education, as well as to a wide pedagogical and academic audience interested in the implementation of the latest information and communication technologies and the promotion of the digital transformation of education and science.

In the context of the digital transformation of education and the mass spread of foreign language learning, it is important to use artificial intelligence technologies in such a way that they enhance rather than replace the activity of teachers and learners. Without a well-designed methodology and clear rules of use, there is a growing risk of reduced learner autonomy, loss of academic integrity, and a widening gap between the theory and practice of language teaching. The aim of the article is to substantiate and describe a model for the effective use of artificial intelligence technologies to enhance classroom-based acquisition of a second foreign language, implemented as an adaptive methodological ecosystem that strengthens learner activity and preserves the primacy of human contribution. The study is based on a theoretical and methodological analysis of works on the theory of second language acquisition, intelligent computer-assisted foreign language learning, formative assessment, and self-regulated learning. These strands are synthesized into practical tools that take into account Russian educational standards and multilevel descriptions of communicative competence. The article proposes an adaptive ecosystem of scaffolding support using artificial intelligence technologies, built around five interrelated components: sequencing of lesson stages, scaffolding support, target skills, safeguards, and self-regulation, as well as a synergy mechanism that structures iterative interaction cycles between the human user and the digital assistant. The model introduces an intervention threshold for artificial intelligence technologies that does not exceed one fifth of the volume of auxiliary materials and presents a ready-to-use toolkit: an alignment map of the digital assistant’s roles across lesson stages, a policy for the use of artificial intelligence with a prompt log, formative assessment rubrics, lesson scenarios, a risk and resilience matrix, an implementation checklist, and monitoring indicators covering vocabulary retention, fluency of oral speech, error frequency, text coherence, teacher checking time, and the footprint of artificial intelligence use. Theoretical significance lies in clarifying the role of artificial intelligence technologies as an instrument of metacognitive support that does not substitute for genuine language use. Practical significance consists in providing a methodological toolkit that helps save teacher time, improve the quality of formative feedback, protect academic integrity, and adapt the model to different disciplines, proficiency levels, and classroom formats.