?
Искусственный интеллект и профессия переводчика: критический интегративный обзор изменений рынка труда, качества перевода и профессиональной реконфигурации (2022 - 2026)
Background. The rapid diffusion of large language models has intensified claims that professional translation is approaching technological redundancy. The available evidence, however, is dispersed across labour economics, computational linguistics, translation stu-dies, industry research, and translator education, and it does not support a uniform conclu-sion for the profession as a whole.Purpose. This article examines how artificial intelligence has affected demand for translation labour, identifies the language pairs and communicative domains in which ma-chine-generated translation may support substitution, and analyses how professional practice and translator education are being reorganised in response.Materials and Methods. A critical integrative review was conducted using studies published from January 2022 through December 2025, together with earlier methodological and theo-retical works needed to interpret current developments. Searches covered Scopus, Web of Science, Google Scholar, ACL Anthology, and eLibrary.ru. Bibliographic metadata, publication status, and the fit between cited findings and the claims attributed to them were checked against official journal pages, proceedings records, institutional reports, or working papers. The evidence was appraised according to design fit, outcome validity, and transferability, and was synthesised by distinguishing technological exposure, translation quality, organisational adoption, workflow change, and realised labour outcomes. Results. Causal and quasi causal labour studies identify measurable pressure in online freelance markets and in United States local labour markets with greater machine translation adoption, but these findings cannot be generalised to all forms of translation employment. Quality studies show strong performance for well resourced language pairs and general content while also documenting persistent variation across languages, linguistic pheno-mena, document contexts, literary style, specialised domains, and evaluator expertise. The evidence therefore indicates that technical quality is a boundary condition for substitution rather than a sufficient cause of job loss. Professional work is being segmented among routine content with a low consequence of error, expert workflows supported by artificial intelligence, and human led translation where contextual judgement and responsibility remain indispensable. No Russian longitudinal or econometric study comparable to the strongest international labour evidence was identified. Conclusion. The review develops the concept of conditional substitution. Technical adequacy produces an occupational consequence only when organisations redesign workflows, clients accept the relevant quality threshold, responsibility can be reassigned, and market arrangements permit costs and risks to be transferred to language professionals. Translator education should therefore integrate machine translation and data literacy without weakening advanced language competence, domain knowledge, research ability, revision, ethical judgement, and responsibility for the final text.