Researchers at Volgograd State University (VolSU) have come up with a methodology for correcting artificial intelligence (AI) errors in text translation. This was announced by Professor Vera Mityagina.
A representative of the university explained that modern machine translation systems and language models in the form of AI-based chatbots can translate texts with varying degrees of accuracy. The study analyzed translations from Prompt Neural, Chat GPT, Google Translate, Yandex.Translate and other programs.
Some of them conveyed information verbatim, while others took into account an adaptive approach, including the search for idioms and speech equivalents when translating from one language to another.
In addition to the spelling and punctuation errors identified when working with AI translators, semantic and stylistic errors were also found.
According to her, experts have developed a methodology for preliminary editing of texts to simplify working with AI translators. This involves segmenting sentences, breaking complex constructions into simpler ones, and deciphering abbreviations. Thanks to this approach, the accuracy of translations has significantly increased, Mityagina concluded.