Some robots want your job
Not all robots with artificial intelligence throw their maker-mommies off of cliffs or want to enslave the human race. The scariest thing about robots isn't that they're evil (most of them just want to make our lives easier); it's that they will take our jobs, according to a new book by Martin Ford called Rise of the Robots: Technology and the Threat of a Jobless Future.
Forbes notes that our new friendly operatives could be able to do our laundry for us one day by learning from mistakes, just like the rest of us. These robots are taking cues from the way humans learn and incorporating them into their memory, making them our allies rather than our enemies, hopefully.
Robots that will really be useful as "employees" will be those that can learn on their own. Here are a couple of friendly robots that can learn new skills and might replace you one day.
Baxter and Sawyer are robots that learn from their human co-workers on the factory floor.
A report from UC Berkeley, last week, described how researchers there have been using deep learning to train robots to do everyday human tasks that require dexterity and spatial awareness. The report claims that the team has developed a trial and error learning process which mimics that of humans more closely.
BRETT (the Berkeley Robot for the Elimination of Tedious Tasks) learns to put puzzles together on its own.
By utilizing neural nets that recognize patterns in a changing environment, robots are beginning to react to sensory information in the form of pixels or sound waves. In a three dimensional landscape however, this system becomes more complex.
By giving BRETT (the Berkeley Robot for the Elimination of Tedious Tasks) a point system, the robot can perform a tedious task repeatedly until it is done accurately. If BRETT is given the beginning and end coordinates for a task, it can complete it in about ten minutes, but if only given the sensory data at hand, BRETT would take hours to finish.