Facebook has not entered the robotics industry, but its researchers do use robots and, as they state , they are becoming increasingly intelligent in a short period of time.
The social media giant announced Monday that its engineers, working with scientists at New York University, have reduced the time it takes to teach a robotic arm how to grasp objects from hundreds or even thousands of times before to just a few dozen.
Given that Facebook doesn't sell bots, this announcement may seem like a no-brainer. However, developments like this could lead to improvements in other forms of machine learning, making the software Facebook uses to detect harmful or adverse user behavior on its platform better.
The company's need to use AI to detect extremist violence, hate speech, and misinformation on its platform is becoming increasingly urgent. Progress is certainly being made in this regard, but we are still several years away from creating systems that can automatically block such content without the need for human intervention.
Antoine Bordes, co-CEO of the company's artificial intelligence research labs, said that "the best thing about robotics is that it happens in real time, in the real world.".
Facebook began working on robots over the past year, forcing its researchers to think of creative ways to make machine learning more effective, Bordes said, but added that the company has no plans to commercialize its robots anytime soon.
In addition to the robotic arm, Facebook has been experimenting with ways to teach a six-legged robot how to walk. It said it eventually hopes to reduce the time it takes to train such a skill to hours instead of days or weeks.
Additionally, in collaboration with scientists at the University of California, Berkeley, Facebook has experimented with making robots learn through touch sensors.
Yann LeCun, chief AI scientist at Facebook, said the company needs to look ahead and be ready for future opportunities that may arise, even in the field of robotics.
