Conversational or generative, artificial intelligence is becoming more and more a part of our daily lives every day, and the extent of its possibilities never ceases to surprise us. While chatbots like ChatGPT or Le_Chat are now part of our environment, current research around AI is sometimes much more original. This is the case, for example, of several scientific projects in the field of animal communication. What hopes can we soon have of understanding animals using artificial intelligence, to better protect them and promote their well-being?
AI to the rescue of research on animal communication
machine learning, at the heart of modern artificial intelligence, excels in detecting relevant patterns within large data sets. Algorithms can analyze vocalizations and cries, facial expressions, body movements, and even chemical information such as pheromones to identify potentially meaningful signals, radically transforming the way scientists analyze data relating to animal communication.
Various research projects currently being conducted around the world illustrate the full potential of AI in this field. The Earth Species Project, for example, is dedicated to decoding non-human communication with the ultimate goal of achieving two-way communication with other species. The Project Ceti (Cetacean Translation Initiative) focuses on translating sperm whale clicks and humpback whale songs, using artificial intelligence to analyze these “structured sequences” based on models inspired by those behind ChatGPT. DeepSqueak attempts to interpret rodent sounds to help detect pain or disease.
Deep learning tools aren’t magic, however, as Kevin Coffey, neuroscientist and co-creator of DeepSqueak, points out. Translating animal languages requires a deep understanding of animal behavior, which requires biologists to observe and contextualize. Subtle and nuanced, animal signals can be difficult to interpret, and some animals may simply not communicate in a way that their signals are directly translatable into human language. Biais anthropocentric, you said?
Source: Tweaktown
0 Comments