How Scientists Are Using AI to Talk to Animals

Abstraction: AI and digital bioacoustics decoding complex nonhuman animal communication

Key points:

• Karen Bakker (UBC/Harvard Radcliffe) argues the field has shifted from teaching animals human language to decoding animal communication on its own terms—a more "biocentric" approach using the concept of umwelt (organism's lived experiential world)
• Digital bioacoustics uses miniature portable recorders deployed everywhere from Arctic to Amazon, on backs of turtles and whales, recording 24/7 without human disruption
• Yossi Yovel (Tel Aviv University) used voice-recognition AI to classify 15,000 Egyptian fruit bat vocalizations, finding bats argue over food, distinguish genders, use individual names, and mothers speak "motherese" (at lower pitch, not higher)
• Tim Landgraf (Freie Universität Berlin) built RoboBee—a robot that successfully decoded and transmitted the honeybee waggle dance, directing bees to a previously unvisited nectar source
• The same NLP algorithms used in tools like Google Translate can detect patterns in nonhuman communication; deep learning extracts patterns invisible to human ears (bats communicate ultrasonically and faster than humans)
• Bakker compares digital bioacoustics to the invention of the microscope: "a planetary-scale hearing aid" opening questions about the human/nonhuman divide and new conservation approaches

Connections: Karen Bakker · Digital Bioacoustics · Interspecies Communication · Natural Language Processing

Source: https://www.scientificamerican.com/article/how-scientists-are-using-ai-to-talk-to-animals/