Abstract
Across law and science, interspecies translation has emerged as a consequential aspiration: from legal efforts to grant rights to animals and rivers, to AI-powered attempts to decode cetacean communication through bioacoustic machine learning. I examine the Cetacean Translation Initiative (Project CETI) as a case study in the constitution of such aspirations, arguing that AI-assisted interspecies translation constitutes the parties to the relation it purports to mediate. Situating Project CETI alongside the ‘rights of nature’ movement, I trace the linguistic presuppositions underlying approaches to language as an engineering problem divorced from semantics, an inheritance running from Shannon and Weaver to contemporary Large Language Models. I then demonstrate how the Generative Adversarial Networks Project CETI deploys perform communication in ways that are normatively constitutive. The technical novelty of AI and the normative novelty of Rights of Nature converge in a single operation: AI-assisted interspecies translation performs the parties whose encounter it mediates.