Artificial intelligence in administering the expenditures of the Social Fund of Russia: Concept of a modular control system

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Aim. To develop a conceptual model of a modular artificial intelligence (AI) system for controlling the expenditures of the Social Fund of Russia (SFR) and to assess its potential effectiveness. Objectives. To analyze the expenditure structure of the SFR and identify the areas that are most critical in terms of misuse risks; to systematize existing practices of AI application in the budget process and to assess their applicability for controlling the expenditures of extrabudgetary funds; to develop an architecture of a modular AI system adapted to the specifics of the SFR’s control procedures; to perform a calculative and analytical assessment of the proposed system’s effectiveness. Methods. The research is based on a systems approach, employing methods of structural-functional analysis, business process modeling, statistical analysis of data from the SFR and the Ministry of Finance of Russia, as well as the expert assessment method to identify the potential effect of AI implementation. The calculation of the labor intensity of control procedures is performed taking into account the standard labor costs established by departmental regulations. Results. A modular AI system architecture is developed, comprising three modules: automatic procurement verification (comparison with balance sheet accounting), risk scoring of social payments (a graph model of applicant connections), and control of medical services (matching the service code with treatment outcomes). The optimal system is based on the principle of combining automated selection of suspicious transactions while preserving the expert role of employees. The probable reduction in misuse of funds is estimated by the authors in the range of 1,2–1,8 % of the SFR budget. The reduction in labor costs for desk audits will allow reallocating freed resources to more complex cases requiring expert assessment. Conclusions. The proposed modular AI system for administering SFR expenditures makes it possible to reduce the time spent on control procedures while preserving the expert role of employees for decision-making in non-standard situations. The results obtained confirm the hypothesis of the expediency of implementing a modular AI-based SFR expenditure control system. Further research may be directed toward pilot testing of the system and the development of a methodology for assessing prevented damage from misuse of budgetary funds.