Animal friendly artificial intelligence applications

Our animal friends are everywhere in homes, streets, parks. Even though we couldn't talk to them or understand their language, we always found a way to communicate. But did you ever say that even if my cat has a tongue, he can tell me his troubles?

Maybe people who ask the same question to their pets have started to develop artificial intelligence algorithms that learn what the pets want to say with their voices and body language.

We always hear pet owners brag about their pets' obedience. The main thing lies in the person's ability to communicate with him, rather than the animal's obedience. We “teach” various words to our animals and expect them to understand each time we call them. In fact, animals do not learn words, but the stress, energy and intonation that the word carries. In other words, when you make another sound that sounds like "come" without saying "come", or even if you just radiate that energy, your dog may run towards you. Well, we consider it normal for him to understand the word "come", but what do we understand by the word "woof"?

Artificial intelligence algorithms that learn "what they want to say" with the voices and body languages of pets have begun to be developed. It is thought that the first pet translation application could be launched in 10 years. Artificial intelligence, as it's actually used here, is using "machine learning" as your observation technique to learn the basic communication methods of animals.

For example, thousands of barking videos that also show body language and tail movements are analyzed to understand what a dog is trying to convey when it barks or pinches its tail. In the first place, people's experiences are consulted for information on what these movements mean. An experienced dog groomer might comment, "He wants to go out," when he sees a certain sequence of movements. This comment is associated with the video and defined to the artificial intelligence algorithm. Artificial intelligence begins to decipher animal language by comparing and matching thousands of different expressions like this. When the preliminary work is completed, it will be possible to understand what the problem is by holding the mobile phone to the animal's face.

Of course, it is obvious how complex this communication is. Northern Arizona University's Dr. Con Slobodchikof is an academic who has been working on the communication of American prairie dogs for over 30 years. Dr. Slobodchikof found that prairie dogs use sophisticated sounds that can almost be considered a "language." These dogs can signal to each other that a carnivore is approaching by making high-pitched sounds. Their sounds vary depending on the type and size of the carnivore. Animals can communicate more complexly with different combinations of these sounds, they can even express the clothing color of a nearby human. Dr. Slobodchikof managed to develop an algorithm that translates sounds into English by processing the data he obtained with artificial intelligence. The doctor, who founded a company called Zoolingua, believes that in the near future, the expressions of domestic dogs and cats can be understood with this technology.

Of course, the limits of artificial intelligence do not end there. The upcoming project plans to make its artificial applications as intelligent as animals. Trying to develop this project with experiments, the team opens the "virtual playground", similar to the playgrounds prepared for the same animals, for the competition of artificial intelligence algorithms. Artificial intelligence developers plan to train and race artificial intelligence algorithms in this process. Of course, the main purpose of these artificial intelligences, just like animals, is to find food.

The video above belongs to the virtual playground olympics. “Deep reinforcement learning”, which is one of the important methods used in training artificial intelligence applications or algorithms, is a “reward-punishment” training method that we use in animal training. For this reason, artificial intelligence algorithms can be used to develop them by giving reward-punishment-based training just like animals.

There is an artificial intelligence-based development that will affect farm animals as well as our pets. A project that is currently operational in China and is used to monitor livestock. Chinese farmers are taking advantage of the heat, sound and vision capabilities of artificial intelligence to monitor the position, welfare and health of the animals they raise.

I feel like you've been thinking about questions like "I hope not all animals are fitted with GPS", and you have questions about how this practice can affect animals badly. However, this application, which is already made to have a good effect on the farmer and the animal, provides information about the health of the animals by using infrared rays in the tracking system. The voice recognition feature allows animals that are coughing or making distinctly different sounds to be distinguished and intervened before it is too late. Instead of RFID tags, that is, a GPS-like tracking system provider, it is possible to detect the location of the animal with tags or signs with an identification number. In this way, the safe development process of farm animals continues to progress all over the world with the support of artificial intelligence.

There is another project that makes me very happy and will also make animal lovers happy. This time, artificial intelligence plans to replace many animals such as rabbits, frogs, mice, monkeys, cats, dogs and so on, on which numerous and wrong experiments have been carried out in the laboratory. No matter how much the number of animals is tried to be reduced, the number of animals experimented on in laboratories does not decrease.

So what can artificial intelligence do to it? First of all, it has been realized that the experiments performed vary from animal to animal and that the results are not certain. There is software developed for this. The basis of this software is to analyze the relationships map between existing molecular structures and determined toxicology types and to produce correct combinations. And thanks to this map, possible reactions or compatibility of the eye, skin or DNA with possible chemicals can be analyzed without experimentation. Of course, although the debates against it continue, it is planned to develop the software and make it work in certain laboratories.

References:

References

Nurbuke TEKER

Increasing Quality in Software Projects: TMMI

Web Scraping

Blog

Animal friendly artificial intelligence applications

Stay Informed!