Remember that popular sitcom from the 80s called the “Small Wonder”? It was based on the family of robotic engineers who modeled a robot like a girl and introduced her to the world as their daughter. The robot was named “Vicki” and needless to share that she made their life so much easier. This is due to the work of artificial intelligence (AI). We call it AI because here the human artificially made the robot intelligent by feeding it data to perform a set of actions effectively. This is how AI was being used in real-time worldwide so far. However, Facebook has stepped up to bring a change in it.
Mark Zuckerberg took the world by storm when he launched Facebook in 2004. And now he is looking forward to taking the world by storm again with its newest AI on Facebook. Facebook’s new AI has been designed to perform actions with lesser help from human beings. This is because the AI will be able to teach itself to see with less help. Yes, this sounds like the future of AI and perhaps it is. Hence, it’s important to know how this AI is intended to perform? How do the experts see this development from the uber successful but recently controversial Facebook?
Seer (for SElf-supERvised)
This strange looking subhead is the name of the latest Facebook algorithm that is intended to help its new AI perform actions on its own. This algorithm is designed to take a billion pictures from Instagram (now owned by Facebook) and use it as the base for the AI to teach itself to determine an outcome. The AI will use these pictures to choose for itself the items that resemble the image it has to look for. Let’s say it to identify an image of the cat. For this, it will gather around the pictures with bristles, sharp ears, and hide to perform a calculation and finally, name the image as “Cats.” This can help in the auto generation of hashtags.
As per reports, Facebook is still working on incorporating this completely. However, this has generated some strong views from the various experts in this field. And while we get to learn more about this development in the near future, let’s check what these experts have to say on this new AI experiment on Facebook.
Experts Opinion on Facebook’s New AI Experiment
Olga Russakovsky has stated that the results displayed by the “Seer” algorithm have been noteworthy so far. She is an associate teacher at Princeton University who has immense knowledge in the field of AI technology and PC vision. She further emphasizes on self supervised learning, as to how it will require more testing and how it can have outright downstream ramifications for improved visual acknowledgment. Russakovsky also pointed out that how it was evident that Instagram pictures were not handpicked to make free learning simpler.
Yann LeCun who is Facebook’s chief scientist states that Facebook research is a landmark in the field of AI developments. LeCun who states that self supervised learning can have many useful applications is a pioneer of machine learning that involves large artificial neural networks. He further described how the self supervised learning could be useful. He stated that the Seer algorithm can help in matching Ads in Facebook to relevant posts and will filter the undesirable contents.
Aude Olivia from MIT’s Computational Perception and Cognition lab emphasized that the research work at Facebook will help us in undertaking more ambitious projects related to visual recognition. She was even alarmed about the problems stating that such algorithms will require enormous computational power that could strain the available supply of chips.
Alexei Efros of UC Berkeley has stated that the Facebook research gives a good demonstration of how AI could be advanced in the future. Even though self supervised will require a gargantuan amount of data, he believes that this is how progress can be brought to AI.
We, at OpenGrowth, are committed to keeping you updated with the best content on the latest trendy topics from any major field. Also, both your feedback and suggestions are valuable to us. So, do share them in the comment section below.
