Animal translations: AI and the intelligibility of non‐human worlds

Amid the general sense of worry that large language models will soon drown out human voices, some researchers are optimistic that machine learning will allow humans to listen to and understand animal voices to an unprecedented extent. As part of a broader project aimed at interspecies communication, a loosely connected set of animal behaviourists, AI engineers, and public‐facing animal advocates are working to comprehend ‘animal languages’ and thus to improve the lives of domesticated and wild species. Work in this field not only assumes a novel set of speakers (which can include, for some, plant as well as animal species), but it also contributes to a redefinition of what language is. This article examines how the expansion of language to non‐human animals is helping to produce posthuman worlds in which linguistic expertise is partially transferred from humans to machines. These new ways of defining language are especially visible in the forms of translation that people in the world of interspecies communication engage in that link animals, humans, and artificial intelligence. Repeating earlier moments in translation history, a new kind of linguistic expertise makes a new kind of translation possible.