A fuzzy multicriteria model for assessing social risk based on public attitudes toward artificially generated content

Abstract

Relevance. Artificially generated content can undermine information trust, increase vulnerability to manipulation, and create group-level social risks that are difficult to represent by crisp indicators Purpose. To develop and empirically test a fuzzy multicriteria model that transforms heterogeneous survey responses into a group-level social-risk assessment Tasks. To formalize social indicators and criteria, risk-orient linguistic responses, compare socio-demographic groups, and define decision rules for interpreting the resulting index. Methods. Anonymous survey data from three hundred eighty-one respondents in Ukraine were processed using ordinal recoding, Z-shaped membership functions, equal criterion weights, TOPSIS ranking, and dimension-normalized conical aggregation Results. The group indices were 0.693 for respondents aged eighteen to thirty-three, 0 for respondents with a bachelor's degree, and 0.307 for the 6,700–10,000 UAH income group. After dimension-normalized aggregation, the selected scenario produced 0.276, corresponding to high social risk Scientific novelty. Social risk is formalized as a formative fuzzy multicriteria construct integrating trust, concern, verification behavior, regulatory expectations, and socio-demographic grouping Practical significance. The model can support public information-security bodies and media-literacy program managers in prioritizing target groups, selecting interventions, and planning repeated monitoring; it is not an individual diagnostic instrument Conclusions. The model provides interpretable group-level screening, but its outputs require validation using the sample composition, subgroup intersection sizes, and sensitivity analysis.