Аннотация
Increasing the adaptability and effectiveness of university management as complex sociocenoses requires new approaches combining systemic methodology and modern artificial intelligence tools. The purpose of the research was to develop and test an intelligent decision support system that integrates the formal apparatus of the theory of inventive problem solving (TRIZ) with the generative capabilities of large language models for analyzing and optimizing managerial situations in the educational environment. To achieve this goal, a prototype of the chatbot "Management Solutions Consultant" was developed. Its core is an original algorithm that implements a sequence of classic TRIZ methods based on the licensed ChatGPT 4.0 API: from the formulation of an administrative contradiction and the construction of a problem situation model (including the operational area and the ideal end result) to the selection and adaptation of universal principles for resolving contradictions (techniques, standards) to the context of the university. The system architecture includes a request preprocessing module, a TRIZ logic application module, and a contextual response generation module. The approbation was carried out on a series of cases modeling typical management tasks at the university (resource allocation, conflict of interest resolution). A conceptual model has been developed that describes the integration of such a tool into the university's digital ecosystem at the levels of the educational process, administration, and strategic planning. The scientific novelty lies in the proposed algorithm and architecture, which provide a structured, systematic application of the language model for creative solutions to management problems. The practical significance of the work lies in the creation of a prototype tool that can be implemented to support decision-making, training, and scenario modeling. Further research will be aimed at expanding the knowledge base of the system, quantifying the effectiveness of its recommendations and adapting the algorithm for other types of sociocenoses.