Why researchers think we’re close to getting interspecies chatbots.
Support our work. Become a Vox Member today: http://www.vox.com/memberships
AI researchers and biologists are teaming up to detect patterns in animal communication that are difficult, if not impossible, to observe with the human eye alone. The studies that have come out so far are steps towards what some AI companies see as a larger goal: building large language models to decode animal communication that is beyond the reach of human comprehension.
This video explains how they might do that, using the same tools that gave us text, image and translation tools for human language.
Here is the study about elephant names co-authored by Joyce Poole and Mickey Pardo:
https://www.nature.com/articles/s41559-024-02420-w
Karen Bakker’s book The Sounds of Life is a great read if you want to learn more about how biologists are taking advantage of advancements in sound technology to study animals:
https://bookshop.org/p/books/the-sounds-of-life-how-digital-technology-is-bringing-us-closer-to-the-worlds-of-animals-and-plants-karen-bakker/18252919
We didn’t cover it here, but there was an interesting breakthrough using AI to detect more sperm whale codas than ever before:
https://www.nature.com/articles/s41467-024-47221-8
Here is a link to Yossi Yovel’s bat study:
https://www.nature.com/articles/srep39419
Here is the macaque monkey study:
https://royalsocietypublishing.org/doi/full/10.1098/rsos.150432
Earth Species Project can be found here:
https://www.earthspecies.org/
Vox.com is a news website that helps you cut through the noise and understand what’s really driving the events in the headlines. Check out http://www.vox.com.
Watch our full video catalog: http://goo.gl/IZONyE
Follow Vox on Facebook: http://goo.gl/U2g06o
Or Twitter: http://goo.gl/XFrZ5H
source
date: 2024-07-31 12:00:07
duration: 00:09:06
author: UCLXo7UDZvByw2ixzpQCufnA
How AI could help us talk to animals
Researchers are using artificial intelligence (AI) to decode animal communication, with the goal of one day facilitating true interspecies conversation. The field of animal research is facing a "cocktail party problem," where recordings of animal vocalizations are often mixed with other sounds, making it difficult to identify individual calls. However, AI models have been trained to separate out individual calls, such as in a study on African savanna elephants, which suggests that they may have a way of directing calls to specific individuals.
Machine learning is being used to analyze large datasets of animal vocalizations, allowing researchers to identify patterns and relationships that were previously unknown. For example, a study on Egyptian fruit bats used a statistical model to identify the emitter of a call, the context of the call, and who the call was addressed to.
The next step is to develop self-supervised models, which can learn from unlabeled data and detect patterns and categories on their own. This approach is being explored by organizations like the Earth Species Project, which aims to build models that can translate animal communication into human language.
While there are challenges to overcome, such as the need for large datasets and the potential for human bias in labeling and validating the models, researchers are hopeful that AI can help us better understand and communicate with animals. By studying animal communication, we may gain insights into their social structures, behaviors, and even their emotions.
Ultimately, the goal of interspecies communication is not only to improve our understanding of animals but also to promote conservation and protection of the species we share the planet with. As one researcher noted, "They also have a right to be here and a reason for being here."
Error: unable to get links from server. Please make sure that your site supports either file_get_contents() or the cURL library.
