The article is devoted to the issue of translation of artificial intelligence terms. The manual “Make Your Own Neural Network” by T. Rashid, which describes in detail the process of creating a neural network, its design and functioning, was chosen as the material for analysis due to the accessibility of the information for a non-specialist in this field.
Artificial intelligence is a rapidly developing field, and therefore the terminological field of this science is also expanding and diversifying. One of the distinctive features of artificial intelligence as a scientific field is its cross-disciplinary nature as it deals with mathematics, physics, robotics as well as biology and medicine. The terms were divided into two major groups: terms directly related to the field of artificial intelligence and terms borrowed from adjacent fields. The terms were then categorized according to whether they had one, two or more equivalents or no equivalents in order to identify the pattern of equivalence in the target language and the difficulties encountered when the required variant was not available. Often, due to the novelty of AI terms, the translator does not find an equivalent in the target language and therefore resorts to calque, transliteration, transcription or their hybrid. In this article, the author suggests the variants of translation of the terms that have caused the greatest difficulty in translation, which, in our opinion, will better meet the requirements of the term as a lexical unit.