Від «галюцинацій» до «помиїв»: семантичний аналіз девальвації штучного інтелекту

This article provides a comprehensive sociolinguistic analysis of the lexical evolution in the field of artificial intelligence (AI) from 2015 to 2025. The author investigates the radical transformation of nomenclature describing the processes and results of AI work, documenting a shift from anthropomorphic, psychological metaphors to metaphors of biological waste and degradation. The theoretical framework of the research is based on conceptual metaphor theory and critical discourse analysis, allowing for the interpretation of language changes as indicators of public trust and the “loss of awe” toward technology. The first stage of development (2015–2022) is examined through the lens of anthropomorphism, with the term “hallucination” serving as a key concept. It is established that the use of clinical terminology granted machine models an illusion of agency and an “inner world,” justifying their status as “intelligence”. The central part of the article is devoted to analyzing the contemporary era of “slop”. The author argues that the term “slop,” historically referring to liquid swill, represents the lowest point of semantic devaluation of AI output. Particular attention is paid to metaphors of biological degeneration, such as “incestuous training” and “Model Autophagy Disorder” (MAD). These terms appeal to cultural taboos and pathologies to describe the processes of recursive model training on their own synthetic content, transforming the image of a “logical machine” into a “sick organism”. The conclusions emphasize that the shift in linguistic framing toward environmental metaphors reflects a maturing public perception, where AI appears not as a “savior” but as a source of digital garbage requiring “sanitization”. The use of terms of disgust is seen as a mechanism of abjection, allowing humans to reclaim their hierarchical superiority over machines in the context of automation.