Robots are learning to think for themselves, so should we be scared?

Posted by
Tom Victor

For all the fearmongering around the rise of robots, we’ve always been able to cling on to the fact that humans will remain better than them at one thing: independent thought.

Sure, there are elements of automation which we’ve already conceded will conceivably require minimal human input, but until now we’ve assumed intuition and the general premise of thinking for ourselves would remain a differentiator.

At the very least, we probably assumed it would take a long time before machines even came close to being taught stuff and actually learning new skills and techniques by themselves, even if they can already help each other through doors.

We were wrong.


We are all going to die

According to a new blog post from Deepmind, the AI-focused company bought by Google in 2014, robots have been taught complex tasks like stacking blocks. It might sound basic, but it’s more than you’d be able to achieve without an actual human brain.

Citing a new paper entitled ‘Learning by Playing - Solving Sparse Reward Tasks from Scratch’, the blog post looks at the role of basic understanding in clearing the way to accomplish more complex tasks.

“Just as a baby must develop coordination and balance before she crawls or walks, providing an agent with internal (auxiliary) goals corresponding to simple skills increases the chance it can understand and perform more complicated tasks,” it reads.

Essentially, the robots are encouraged to explore their own parameters and abilities to complete the block-stacking task, with rewards for successful completion.


Robots are learning to teach themselves things

As TheNextWeb explains: “The amazing part is that this particular machine isn’t following a program or doing something it was designed for. It’s just a robot trying to figure out how to make a human happy.”

It notes that more complex tasks like these, separate from pre-programmed and/or mundane tasks, could represent a sea-change when it comes to the future potential of robots to take on ‘real-world’ responsibilities.

Everyone’s got to start somewhere, even robots. It might be time to revise that 2060 prediction for their rise to dominance.

(Images: Deepmind/iStock/Rex)