Currently, AI algorithms are able to learn and do functional work by recognizing objects in images with the unnecessary help of human’s help, shown by Facebook.

Seer (SElf-supERvised), a Facebook algorithm, can categorize and identify itself which objects look alike from more than a billion images scraped from Instagram.

Also, the algorithm will collect the images into one pile using thousands of labels of each object such as cat, fur, whiskers, and pointy ears.

“The results are impressive,” said Olga Russakovsky, an assistant professor at Princeton University who specializes in AI and
computer vision. “Getting self-supervised learning to work is very challenging, and breakthroughs in this space have
important downstream consequences for improved visual recognition.”

“The Facebook research is a landmark for “ self-supervised learning”, an AI approach,” said Facebook's chief scientist, Yann LeCun.

        LeCun pioneered the machine learning approach by feeding data to large  artificial neural networks known as deep learning as a better way to design and program

the machines to work in its own algorithm.  

‍

        Many useful applications can be applied to self-supervised learning without any difficulty, such as being able to read medical images without the need for labeling

so many scans and x-rays. LeCun also said that Seer could be used at Facebook to match advertisements  to post and to screen out unfriendly contents.

‍

        Prior to the fact that algorithms have been used to translate text which is difficult to recognize images, the Facebook research builds upon steady progress on

Seer to be more efficient and effective. The research team developed a new way for algorithms to recognize images even when one part of the image has been

changed.

‍

Source: https://www.wired.com/story/facebook-new-ai-teaches-itself-see-less-human-help/