This thesis is aimed at identifying the errors that arise in the acquisition of English by Uzbek-speaking students. In addition, this work proposes an AI-agent-based diagnostic model. The study relies on Contrastive Analysis and Interlanguage theories and attempts to predict the types of errors arising from the typological differences between the two languages. The purpose of the research is to compare the error-detection accuracy of the AI agent with human-expert evaluation, and to show in which error categories it is strong and in which it is weak. The results of the proposed model can serve as a practical basis for adapting English-language teaching methodology and AI tools to local conditions.