Facebook is trying to break an important threshold for artificial intelligence technology. Technologies we call artificial intelligence or AI can learn new things thanks to the algorithms of software developers. In short, the development of these systems depends on the team working behind it. Facebook has managed to develop self-learning artificial intelligence. The firm thus seeks to reduce its dependence on human beings.
Facebook took the first step for learning artificial intelligence
Facebook has managed to go beyond the learning done by humans through tags in its newly developed artificial intelligence algorithm. The social media giant is trying to give the artificial intelligence algorithm the ability to learn by itself.
Accordingly, more than a billion posts from Instagram were used for the Seer (Self-supERvised) algorithm. The artificial intelligence technology used decided which of these images were similar to each other.
The most interesting feature of the artificial intelligence algorithm is that it can solve the existence of another algorithm. Seer was shown a small number of pictures this time using cat tags. The algorithm noticed the bad tagging and was able to complete that there was another algorithm behind it.
Scientists from important universities in the USA also witnessed this artificial intelligence experiment conducted by Facebook. Scientist Yann LeCun, who serves as the Head of Facebook’s Science Department, described the ability to learn oneself with this artificial intelligence technology as a revolution.
LeCun explained that the selected 1 billion Instagram posts were randomly selected, and even simple visuals were not used to facilitate the work of artificial intelligence.
LeCun, Head of Facebook’s Science Department, is one of the prominent scientists in the USA working on artificial intelligence and machine learning. According to LeCun, learning artificial intelligence through labels is too cumbersome to move humanity forward.
Self-learning of artificial intelligence is already used to perceive and translate the language of texts. However, according to LeCun, the identification of images and their grouping by artificial intelligence is more complex than describing texts. Seer, on the other hand, led to an important development in this regard.
Seer artificial intelligence works by connecting objects and images, just like a human brain. The processing power requirement of this new artificial intelligence, which is likened to a large neural network, is also higher than its competitors.
Identifying and learning pictures, classifying them in conjunction with each other, is not the first time ever. Qbo, which was a Spanish startup before, gave the robots it developed the ability to learn. In fact, these robots had the ability to share the information they acquired with each other.