In this article, the development of in-service teachers’ digital competence is examined within an AI-supported learning environment designed to enhance professional modeling skills and the creation of interactive AR-based instructional content. The study investigates how such an environment supports teachers in developing contextual digital skills that enable not only the use of emerging technologies but also their meaningful adaptation to pedagogical goals and instructional needs. A training program involving 916 in-service teachers from Kazakhstani secondary schools was implemented, and survey data were collected to assess changes in digital competence and readiness to integrate AI and AR tools into teaching practices. The findings demonstrate high levels of interest and engagement: 96% of participants expressed readiness for further learning, 86% reported satisfaction with the course content, and 84% showed contextual maturity in applying newly acquired technologies in their instructional processes. These results highlight the potential of AI-supported professional development to strengthen teachers’ capacity to design interactive learning environments, promote equity and quality in digital education, and enhance student engagement.

The object of this study is the analysis of questionnaire data using ontological modeling. The task relates to the fact that conventional methods for processing questionnaire data are often insufficiently effective when working with large volumes of information and do not make it possible to automate many analysis processes. As a result of the study, an ontology was designed that structures and analyzes questionnaire data, which allows for a more accurate identification of hidden relationships between variables. Using these theoretical provisions, an information system for assessing the quality of assimilation of preschool children's competencies was built. 150 children from various preschool organizations were involved in the study as respondents. The data integration method proposed in this paper significantly facilitated the process of data analysis both for a group and for an individual respondent. The key difference of the proposed methodology is the automation of routine data analysis operations based on the ontological structure, which significantly simplifies the processing of large volumes of information. This makes it possible to solve the problem of limitations in conventional analysis methods and makes data analysis more scalable and reproducible. The practical application of the results is possible in marketing for analyzing customer satisfaction, market segmentation, and evaluating the effectiveness of advertising campaigns. In the educational domain, the ontology could be used to evaluate the quality of programs and analyze respondents' opinions, and in sociology – to analyze public opinion and conduct research on social phenomena. Thus, the proposed ontology provides an effective tool for analyzing large volumes of questionnaire data, allowing organizations to make more informed decisions and improve their efficiency

Educational equity and access to quality learning opportunities represent fundamental pillars of sustainable societal development, directly aligned with the United Nations Sustainable Development Goal 4 (Quality Education). Student retention remains a critical challenge in higher education, with early disengagement strongly predicting eventual failure and limiting opportunities for social mobility. While machine learning models have demonstrated impressive predictive accuracy for identifying at-risk students, most systems prioritize performance metrics over practical deployment constraints, creating a gap between research demonstrations and real-world impact for social good. We present an accountable and interpretable decision support system that balances three competing objectives essential for responsible AI deployment: ultra-early prediction timing (day 14 of semester), manageable instructor workload (flagging 15% of students), and model transparency (multiple explanation mechanisms). Using the Open University Learning Analytics Dataset (OULAD) containing 22,437 students across seven modules, we develop predictive models from activity patterns, assessment performance, and demographics observable within two weeks. We compare threshold-based rules, logistic regression (interpretable linear modeling), and gradient boosting (ensemble modeling) using temporal validation where early course presentations train models tested on later cohorts. Results show gradient boosting achieves AUC (Area Under the ROC Curve, measuring discrimination ability) of 0.789 and average precision of 0.722, with logistic regression performing nearly identically (AUC 0.783, AP 0.713), revealing that linear modeling captures most predictive signal and makes interpretability essentially free. At our recommended threshold of 0.607, the predictive model flags 15% of students with 84% precision and 35% recall, creating actionable alert lists instructors can manage within normal teaching duties while maintaining accountability for false positives. Calibration analysis confirms that predicted probabilities match observed failure rates, ensuring trustworthy risk estimates. Feature importance modeling reveals that assessment completion and activity patterns dominate demographic factors, providing transparent evidence that behavioral engagement matters more than student background. We implement a complete decision support system generating instructor reports, explainable natural language justifications for each alert, and personalized intervention templates. Our contribution advances responsible AI for social good by demonstrating that interpretable predictive modeling can support equitable educational outcomes when designed with explicit attention to timing, workload, and transparency—core principles of accountable artificial intelligence.

One of the important elements of the information infrastructure of educational institutions is the information and educational environment. The information and educational environment of preschool education organizations has its own characteristics, which must be taken into account when building a model for managing information processes. The object of study in this work is the information and educational environment for preschool education organizations. The problem to be solved is the need to develop a model for managing data and information processes, which will allow determining the learning outcomes of preschoolers and adjusting individual work with them. The introduction of the developed model allowed to reduce the time spent on adjusting individual work with students by 30 %. These results are explained by the optimization of information processes, as well as improved monitoring of the formation of skills of preschoolers and a reduction in the time for its implementation. When monitoring for each child, 211 indicators are examined. There are 633 indicators per year for three monitoring, in aggregate, per one child. By default, the data is entered into Microsoft Excel and processed manually. However, a large number of entries slows down the processing of Microsoft Excel data and increases the chance of errors. The use of this model will make it possible to carry out calculations automatically, save data and generate reports for each child or group of children. The developed model can be used in information and educational environments for preschool education organizations in order to improve the efficiency of monitoring and managing educational processes

Development of a decision­making method to form the indicators for a university development plan

Modern linguistic steganography faces the fundamental challenge of balancing embedding capacity with detection resistance, particularly against advanced AI-based steganalysis. This paper presents DeepStego, a novel steganographic system leveraging GPT-4-omni’s language modeling capabilities for secure information hiding in text. Our approach combines dynamic synonym generation with semantic-aware embedding to achieve superior detection resistance while maintaining text naturalness. Through comprehensive experimentation, DeepStego demonstrates significantly lower detection rates compared to existing methods across multiple state-of-the-art steganalysis techniques. DeepStego supports higher embedding capacities while maintaining strong detection resistance and semantic coherence. The system shows superior scalability compared to existing methods. Our evaluation demonstrates perfect message recovery accuracy and significant improvements in text quality preservation compared to competing approaches. These results establish DeepStego as a significant advancement in practical steganographic applications, particularly suitable for scenarios requiring secure covert communication with high embedding capacity.