Do you have a dog at home? Have you ever stared into your eyes as if you wanted to say something to you? Does he understand you when he looks at you or wags his tail or puts his ears in one way or another? Yes, technology could be close to solving that dialogue between the owner and his dog, and all this thanks to artificial intelligence (AI).
This is asserted by a study from the University of Northern Arizona unveiled earlier this year by Dr. Slobodchikoff. He and his team have been working for months on an artificial intelligence-based instrument that will learn from the dog’s vocalizations and translate it into English. Slobodchikoff has been studying and researching how prairie dogs communicate for more than 30 years, and has been able to help create the AI that will help dog lovers communicate with their pets.
And how has Slobodchikoff built the instrument that will allow us to know what our dogs want when they look at us or wag their tails or place their ears in one way or another? Thanks to an algorithm that will convert the vocalization of a dog to the language of Shakespeare. According to Slobodchikoff, the prairie dogs he has studied for three decades make high-pitched barks when in danger, and can even communicate the color of a person’s clothing with their particular form of language.
To discover this evidence Slobodchikoff and his team of scientists spent thousands of hours watching videos of the various types of barking made by prairie dogs. And they learned from images in the same way that artificial intelligence and machine learning learn.
From here, scientists from the University of Northern Arizona created the company Zoolingua, whose objective is to develop the technology necessary for people to communicate with their pets, which could allow talking with all kinds of animals and not only with the dogs.
Slobodchikoff says, given his discovery and the technology he is exploring so that people can communicate with their dogs, that “if we can do this with prairie dogs, surely we can do it with cats and dogs.”
Neural networks are also being trained at the University of Washington and the Allen Institute for Artificial Intelligence to predict and interpret dog behavior. And the results indicate that animals could be a source of data to train, in addition, artificial intelligence systems to think like a dog.
And to obtain these results, they mounted a GoPro camera on the head of a dog, to which they also placed sensors on its legs and on its body, to capture how it walked, played or looked for something in the park. Around 380 videos of all this were recorded, and with the information captured, the researchers analyzed the behavior of the dog with deep learning or deep learning technologies to extract patterns from the data and thus train the neural network.
Researchers from the University of Washington and the Allen Institute certified with this research that “dogs clearly demonstrate visual intelligence by recognizing food, obstacles, other humans and animals, so that a neural network trained to act like a dog shows the same intelligence ”.
