Introduction. The article presents the results of a study on generating high-quality educational content in mathematical literacy for 5th-grade students using generative AI. The problem stems from the lack of adaptive assignments that meet educational standards and the limitations of AI (hallucinations, non-reproducibility). The aim of the study is to develop, test, and evaluate the effectiveness of an original prompt-engineering strategy for generating pedagogically relevant and age-appropriate math problems. Materials and Methods. The study employs systemic and activity-based approaches. Methods include analysis of AI applications in education, experimental task generation using a hybrid prompt-engineering strategy (Few-Shot Learning + Chain-of-Thought + Role Prompting) based on ChatGPT-4o, expert evaluation (10 mathematics teachers with ≥12 years of experience), and statistical data processing (Cohen’s κ, mean values µ). Verification involved generating tasks in a new context (airports) and assessing them based on adequacy, student-appropriateness, and complexity criteria. Results. Key findings demonstrate the successful implementation of the strategy, enabling the generation of structurally consistent tasks (κ = 0.82). The critical role of Chain-of-Thought prompting in creating multi-step problems is emphasized. The authors highlight the dual functionality of tasks (learning and assessment). The experiment confirmed high expert ratings for adequacy (µ = 4.81), format compliance (µ = 4.77), and descriptive completeness (µ = 4.82). A limitation in terminology complexity for some tasks was identified. Conclusions. The study concludes that the combined prompt-engineering strategy is highly effective for generating standards-aligned tasks and has strong potential for integration into digital learning platforms. Further optimization of linguistic adaptation and the development of a validation pipeline are required for implementation.
Information about the paper titled "LINGUISTIC AND INFORMATION COMPETENCE OF FOREIGN MA STUDENTS STUDYING RUSSIAN AS A FOREIGN LANGUAGE: STRUCTURE AND CONTENT" at IATED Digital Library
Introduction. Modern education faces a contradiction between the active integration of generative artificial intelligence and its underexplored potential for providing evaluative feedback in development students’ mathematical literacy. The purpose of the article is to identify the potential of using a generative language model as a teacher’s tool for generating expert-level evaluative feedback when assessing open-ended mathematical problems Materials and Methods. The research is based on systemic-activity, criteria-oriented, and comparative approaches. Methods employed included theoretical analysis of scholarly literature, criteria-based assessment combined with prompt engineering techniques, as well as quantitative and qualitative analysis to determine the agreement between the evaluative feedback generated by the language model and that provided by a human expert. The sample consisted of 51 students Results. The research experimentally confirmed the feasibility of using generative artificial intelligence for providing evaluative feedback in mathematics education. An effective strategy for automating the assessment of open-ended mathematical problems was developed and substantiated, based on criteria-based assessment and prompt engineering techniques using GigaChat Pro language model. Empirical data revealed a moderate agreement between the evaluative feedback generated by GigaChat Pro and that provided by an expert teacher: accuracy reached 73%, Cohen’s coefficient (k) was 0,57, and the semantic similarity of textual comments (BertScore F1) was 0,614.
The study explores a pedagogically driven prompting strategy for generating AI-assisted tasks to enhance mathematical literacy of Russian 5th-grade students. Grounded in international frameworks (NCTM, PISA, TIMSS) and AI-ineducation research, the authors designed a hybrid Few-Shot + Chain-of-Thought approach using ChatGPT-4o. The methodology involved defining metaparameters, generating context-based tasks, and conducting expert evaluation. Results demonstrate that the strategy produces tasks comparable to textbook models, with teachers rating them highly for relevance (M=4.81 / 5) and age appropriateness (M=4.55 / 5). Limitations were observed in AI-generated terminology, indicating accessibility issues for younger learners. The validated approach offers a scalable method for task generation aligned with PISA and national standards, supporting integration into platforms such as the Russian Electronic School and Spherum, and expanding opportunities for teacher-AI collaboration in instructional design.