UC Berkley researchers have hit a major milestone in the creation of usable AI. They’ve created a new set of algorithms that will allow robots to learn through trial and error — much like humans learn new tasks.
With this kind of educated automaton, there’s nothing they won’t be able to do. Think of mechanical beings assembling your next iPhone, building skyscrapers, or exploring Mars.
Is this where we see the first inklings of the robot apocalypse?
The algorithms create a technique based on reinforcement learning, where the test machine has to hang clothing, assemble a toy plane, put a cap back on a water bottle — all the kinds of things we learn through trial and error without thinking about it. You can see how this all works in a real ‘bot in the video here.
This ‘deep learning’ system is a big deal.
“What we’re reporting on here is a new approach to empowering a robot to learn,” said Professor Pieter Abbeel of UC Berkeley’s Department of Electrical Engineering and Computer Sciences. “The key is that when a robot is faced with something new, we won’t have to reprogram it. The exact same software, which encodes how the robot can learn, was used to allow the robot to learn all the different tasks we gave it.”
With this kind of programming, automated beings can adapt to their ever-changing environments without needing to wait for new instructions. Think of a Mars Rover that could continue its exploration without having to wait for the NASA nerds to figure out what’s wrong and send new code, which was possibly be outdated by the time it got there.
We’re headed to the future, alright. Is it just a matter of time before Rosie the Robot from the Jetsons has a place in your home?
Source: Berkeley News Center