'Training' and 'Learning'

This paper analyses the terms training and learning, examining their historical evolution and usage as recorded in English and Italian monolingual and bilingual dictionaries. The analysis then focuses on two specialised domains, namely machine translation (MT) (both through Neural MT and generative AI technology) and Second Language Acquisition (SLA). We examine how the meanings of training and learning are adapted in each field, highlighting overlaps between their human-related and machine-related uses. English and Italian have been selected as instances of a Germanic and a Romance language to compare their respective encoding of the concepts of training and learning, analysing when and how these terms began to be applied to machines rather than exclusively to humans. This contribution to the glossary of AI terms discusses how linguistic choices shape perceptions of machine capabilities. The paper underscores the role of language in mediating technological change across the evolving meanings and uses of training and learning, and invites reflection on the consequences of anthropomorphic discourse in AI-related communication.