The research investigates the possibilities of artificial intelligence (AI) in interpreting allegorical texts in linguistic and methodical aspects. The research material includes 70 proverbs of the concept sphere “Teaching/Knowledge/Literacy/Science.” The text material is difficult to investigate because of the genre specifics of proverbs and their genre characteristics (i.e., allegorical nature, factuality, conciseness, and instructiveness). The research aims to evaluate the possibilities of AI to interpret allegorical texts as texts with mainly didactic content and texts used in the learning process (educational texts). To evaluate the degree of interpretation adequacy, the authors set the following tasks for neural networks: (1) translate the selected proverbs into a foreign (English) language, (2) model educational-communicative situations for practicing the use of proverbs in speech, and (3) compose a linguistic-cultural commentary. The research has shown that neural networks can facilitate the integration of proverbs into the learning process. The research proves that AI can be used to compose training dialogs and commentaries to texts. The errors of neural networks made while interpreting the content of proverbs can be divided into three groups: (1) ignoring the form of the text (alliterations, puns), (2) misunderstanding of the general content due to incorrect choice of keywords semantics (replacing the meaning of archaism with modern semantics of the word, allegorical meaning with direct one, and ignoring the national-cultural component of the meaning), and (3) misinterpretation of the hidden meaning of the text (moral prescription).