Economic entities as integral parts of the social system have an impact on it. The complexity of structures and uncertainty of behaviour which are also conditioned by incorporating the human factor are the typical characteristics of economic entities and the social system. The lack of precise measurement data as well as precise information is their typical feature. Methods of creating computer models of such systems must therefore be based on uncertain, incomplete or approximate data and hypothetical assumptions. The paper deals with the synthesis of the abstract model of the expert system for determining the level of corporate social responsibility of an enterprise (CSR) with the use of methods of artificial intelligence. The linguistic rule model is built on the basis of the expert determination of the level of CSR based on the level of care for employees, level of supplier‑customer relations, level of its ecological behaviour, and compliance with legal obligations. The linguistic modelling method is based on the theoretical approach to fuzzy set mathematics and fuzzy logic. The aim of the paper is the presentation of the system for determining the level of CSR with the use of non‑conventional non‑numerical methods as well as simulative presentation of the efficiency of its functions. The above‑mentioned expert system is a relevant module of the built hierarchical structure aimed at the research of impacts of activities of economic entities on the social system.