A model for organizing students» project activities using artificial intelligence technologies in a digital educational environment

Abstract

The article proposes and empirically substantiates an organizational and technological model of students» project activities that integrates artificial intelligence technologies into the digital educational environment of the university. The purpose of the research is to improve the quality of projects, the objectivity of assessment, personalize student participation and reduce the unproductive burden on teachers. The experiment was conducted in 2021-2024 on the basis of the Institute of Information Technology with the participation of 520 IT students and 42 teachers, divided into experimental and control groups.; EG used a developed Python/Django and React platform with AI microservices, while KG used a traditional project organization. The methodological framework included system modeling, pedagogical design, and statistical analysis (t-criterion, χ2). The model implements modules for intelligent team building based on clustering of digital portfolios and psychometric data, predictive risk analysis based on Git and Jira data using LSTM networks, as well as an NLP module for automated feedback and evaluation of texts and program code. The results showed a more than twofold increase in the proportion of projects rated «excellent» (up to 45,66%), the complete disappearance of unsatisfactory results and a significant reduction in the proportion of «satisfactory» compared with the control group. There was an increase in engagement: the time spent working on the platform increased almost 1.8 times, while the bulk of typical requests shifted to the AI assistant, which allowed teachers to focus on meaningful support. The workload of curators per project has almost tripled due to the automation of plagiarism testing, formatting and primary analytics. Independent assessment testing revealed a statistically significant increase in teamwork, critical thinking, and problem solving skills in the experimental group. It is concluded that the developed model acts as an effective sociotechnical system that provides synergetic interaction between human and artificial intelligence, and can be scaled to various areas of training; development prospects associated with the use of more complex neural network architectures and the creation of a fully adaptive trajectory of project-based learning are outlined.