Research of the artificial intelligence generative function in creating foreign languages lesson plans

The growing demand for artificial intelligence (AI) in various spheres of societal and human activity is driving the expansion of its functions. One such function in the field of education has become generating foreign language lesson plans. This raises the question: how effective are such plans in terms of their compliance with widely recognized conditions for achieving methodological goals, which are based on the principles of fundamental research in educational psychology, psycholinguistics, and the methodology of foreign language teaching itself? The article aims to identify problematic aspects of foreign language lesson planning and the potential for using AI in the professional education of future teachers. The authors employed a method of comparative analysis of foreign language lesson fragments developed by senior-year students-future teachers-and several AI aggregators (DeepSeek, Perplexity, GigaChat, and Qwen). The results, analyzed against a range of significant criteria for forming foreign language grammar skills (Task 1) and speaking skills (Task 2), show that the greatest difficulty for both the neural networks and the students was meeting the condition of "sequencing actions from simple to complex". Regarding the criteria of "vocabulary inventory", "situation individualization", and "activation of learner oral interaction", all fragments demonstrated sufficiently high scores. In Task 1, the quality of the fragments from three neural networks was significantly lower (Perplexity, GigaChat, and Qwen scored 65, 70, and 61 points respectively), while the students and DeepSeek scored 91 and 97 points respectively. The DeepSeek neural network demonstrated high final results, while GigaChat showed the lowest results in both fragments. The students overall showed higher results in Task 2 (96) than in Task 1 (91). In general, it is noted that completing Task 1 posed greater difficulties for both the students and the neural networks; both students and neural networks solve relatively simple tasks more effectively, while those requiring the consideration of a complex set of actions prove challenging. Although the lesson plans proposed by the neural networks exhibit a number of discrepancies with common requirements, it is unreasonable to reject their use entirely. Teachers gain the opportunity to partially save time and utilize teaching techniques and tools that differ from standard teaching materials or habitual experience. In training future foreign language teachers, lesson fragments generated by AI should be used as instructional materials for methodological analysis and for fostering a better understanding of the criteria for evaluating the effectiveness of various methodological solutions. The authors declare no conflicts of interests.

Keywords

Authors

References