CRITERIA FOR LINGUISTIC EVALUATION OF GENERATIVE ARTIFICIAL INTELLIGENCE TEXTS IN THE UZBEK LANGUAGE

Lola Abdurashitovna Uzaqova

Authors

Lola Abdurashitovna Uzaqova Acting Associate Professor, Uzbekistan State World Languages University, Doctor of Philosophy (PhD) in Philosophical Sciences Tashkent, Uzbekistan

Keywords:

Artificial intelligence, generative text, Uzbek language, computational linguistics, large language models, morphology, semantics, pragmatics, linguistic evaluation, natural language processing.

Abstract

This article examines the theoretical and methodological foundations for evaluating Uzbek texts generated by artificial intelligence systems. The relevance of the study stems from the growing use of Uzbek in digital environments and the challenges of accurately representing its morphological, lexical-semantic, syntactic, and pragmatic features in automatically generated texts. The article argues that linguistic evaluation should extend beyond grammatical correctness to include semantic coherence, contextual appropriateness, stylistic consistency, and culturally grounded meaning. The agglutinative structure of Uzbek, suffix ordering, lexical relations, and sociolinguistic conventions are considered important dimensions of evaluation. A multidimensional framework for assessing AI-generated Uzbek texts is proposed. The discussion highlights the need to account for the structural, contextual, and cultural characteristics of Uzbek when developing and evaluating language technologies.