Gato is the name given to this system by its creators, which is owned by Deepmind, an artificial intelligence development company acquired in 2014 by Alphabet, of which Google is a major subsidiary.

This is Technology with artificial intelligence that can perform up to 604 different activities.

His activities include captioning images that describe their contents, playing various video games such as Atari, or sorting boxes and / or items into blocks using a robotic arm.

This artificial intelligence uses language modeling technology, which is the same system used in the development of voice assistants, such as Alexa or Google, to execute a user-specified command via spoken command.

The GATO operation consists of collecting data grouped into blocks or a series of tokens, which are then processed. Neural network. This type of network is a computer model that tries to work just like brain neurons biologically.

With this innovation, we hope to approach the so-called general artificial intelligence, which can learn in a similar way how people do to solve problems according to context or environment.

This is because the forms of artificial intelligence currently used in the operation of social networks, platforms and the Internet of Things are relevant to specific activities, ie they are not programmed to receive information from the environment.

For example, algorithms AI used for cloud social networking or entertainment services can only display publications and content suggestions based on previous searches by their users. What is known as weak artificial intelligence.

It got its name because the artificial intelligence systems already in place today do not process information through the same learning mechanisms as the human brain or other animal species, preventing them from performing various activities spontaneously.

However, the crucial aspect of Gato is that with the same program that was built into it, he performs hundreds of different tasks. According to its authors, “it can perceive and act on different behaviors in a variety of environments, using a single neural network with the same factors.”

According to the researchers, by developing these types of general artificial intelligence systems, it is possible to continue training technological agents who can perform a large number of tasks and, with a few additional data, optimally perform a greater number of equal activities.

Continue reading: