Objective: This article introduces the Russian Constructicon—an open-access electronic database documenting over 2,200 fixed syntactic constructions of the Russian language—as a novel linguistic resource for the analysis of causative semantics in human-computer dialogue. The study presents the theoretical foundations, structure, and functionality of the Constructicon, demonstrating how it systematically catalogues stable syntactic patterns with their semantic, syntactic, and pragmatic properties. The aim is to show that fixed syntactic constructions recorded in the Russian Constructicon can serve as an effective tool for identifying hidden causative meanings in this new type of discourse.

Method: The study examines the Russian Constructicon's architecture, semantic classification system (53 semantic tags including Assessment, Attitude, and Intensity), and search functionality. The empirical material consists of an author's corpus of 1,000 Russian-language dialogues extracted from public online sources (VK, 4pda.ru, ixbt.com). The study employs contextual analysis, functional-semantic analysis, and comparison of empirical data with the semantic construction types identified in the Russian Constructicon.

Results: The analysis reveals that users interacting with AI regularly reproduce three types of fixed constructions documented in the Constructicon. First, constructions with negative evaluation and fatigue semantics ("хватит с меня") account for 15.7% of all causative utterances, where users demand the assistant stop repetitive or irrelevant behavior. Second, constructions expressing prohibition through contrast ("не умничай, а ответь") constitute 12.7%, where users attempt to regulate the manner of the assistant's response by rejecting anthropomorphic imitation. Third, constructions denoting abrupt change of state ("меня прорвало") represent 8.0%, where users describe loss of emotional