Dr. Doolittle uses AI: Ethical challenges of trying to speak whale

摘要

Artificial intelligence (AI) technologies are increasingly used in conservation practices, e.g., to prevent poaching or inventory wildlife. Another area of application is using AI to decode animal vocalisations to understand better—and subsequently better protect—the animals. It has already been applied to different species, including various whale species. Whales have complex vocalisations that are used for sexual selection, to coo their young, for echolocation, and as a form of communication. Scientists are deploying underwater microphones (hydrophones), robotic fish, and tags to record whale vocalisation. AI is used to identify whale vocalisation patterns, understand their meaning, and digitally recreate these sounds to communicate with them. Understanding and translating whale vocalisations into something humanly understandable aims at helping to identify their movements to protect them from ship strikes and bycatch and prevent or reduce sonar that interferes with their echolocation. Using AI holds potential benefits, but it also comes with several risks. We describe current projects that use AI to decipher the vocalisations of humpback and sperm whales (Section 3). We introduce six ethical challenges of applying AI to decode whale vocalisations and highlight what needs to be addressed to establish these practices responsibly (Section 4). These challenges are anthropomorphism, privacy rights, cultural and emotional harm to whales, technological solutionism, ineffectiveness for whale conservation, and gender bias. This paper critically evaluates the use of AI to analyse whale vocalisation, concluding that using AI to decode whale vocalisations holds many benefits for whale conservation; however, using AI to try to speak with whales is ethically problematic because of the potential emotional, physical, and cultural harm caused to whales.