Does your loyal Labrador ever give you a telling off among all of the unconditional love or raise an objection that it is getting bored with the lack of variety in its diet? Well, those sentiments can be occasionally expressed and as any dog owner will know it tends not to be through verbal signals. While we usually understand what animals we have a close bond to are expressing to us through their barks, at least on a very general level, how great would it be to have that confirmed or understand a little more detail?
The latest AI technology in the world of language processing could, in fact, mean we will soon have a significant insight into the oral communication of animals. Two tech gurus, Britt Selvitelle, Twitter’s founder engineer, and Aza Raskin, a former head of user experience at Mozilla Labs and lead designer at Firefox behind numerous tech features and software part of our daily online lives, like endless scroll, are working on animal communication translation software.
The pair have already built a deep learning algorithm based on an analysis of 70 human languages. Based on the identification of a universal ‘shape’ to languages, it has demonstrated relative success in translating between these languages without any previous knowledge of them. They are now moving on to animal communication.
The process of gathering the raw data their algorithm requires is a little more complicated but they already have 50,000 hours of recordings of hump back whales communicating with each other. Later this year they will set up a network of microphones around one of the most popular meeting places for elephants, in a huge jungle clearing in the Congo. Monkey and apes are also on the list of upcoming oral communication recordings.
The hope is that the advanced AI algorithms they have developed for the human language translation project will find patterns that have some similarity to those identified in the architecture of human languages. If this is successful the result could be new AI Rosetta Stone-type software that can decode, and then translate, what animals are communicating to each other.
Raskin hopes that a scientifically verifiable way to translate animals communicating something like ‘stop’, or ‘you are hurting us’,
“would have profound implications for our judiciary system and the way we pursue our role as stewards of this tiny pale blue marble that we call home.”
Explaining the linguistic concepts that the project’s AI is based on, Raskin details:
“You can compare shapes of languages and it turns out every human language whether Esperanto, Portuguese, English or Japanese, all have the same universal shape.”
“We are taking a similar path with animal language to feed in 50,000 hours of hump back whale recordings that lets us build up this cloud which we can compare to the universal shape for human languages.”
Who knows, one day there might be an answer to the frustrating situation of your dog or cat clearly trying to communicate something to you that you finally understand when it is too late, after a significant delay or not at all.