Those who fear an inevitable robot apocalypse will probably think this is a very, very bad idea: Stanford engineers are currently working on creating a massive online brain that all the world's robots can access, reports Wired.
The project, called RoboBrain, is being backed by the National Science Foundation, the Office of Naval Research, Google, Microsoft, and Qualcomm. It makes use of cloud technology with the aim of developing a free-use Internet "brain" that will allow any robot anywhere to download a constantly expanding global database of knowledge. As individual robots learn new knowledge, this knowledge can also be uploaded to the global brain, meaning that all other robots hooked up to the system can learn it simultaneously.
It's an ambitious plan, one that researchers hope will quicken the artificial intelligence boom and allow robots to more effectively work alongside humans sooner rather than later.
"The purpose," said roboticist Ashutosh Saxena, who dreamed it all up, "is to build a very good knowledge graph — or a knowledge base — for robots to use."
While the system ought to be a powerful tool for roboticists and artificial intelligence researchers, it's also eerily reminiscent of Skynet (from the "Terminator" franchise), or at the very least some kind of rudimentary "Collective" brain, such as what drives the Borg in "Star Trek." The system could eventually allow robots to learn new knowledge at a pace exponentially faster than their human inventors. Isn't this the premise behind every fictional depiction of the robot apocalypse?
Researchers aren't worried about any of that at the moment. They're just working to overcome the many obstacles that artificial intelligence researchers have come to face. We're still a fair distance away from the robot maids we were all promised decades ago. In fact, seemingly "simple" problems like teaching a robot how to navigate a novel environment has proven to be a far more complicated task than roboticists originally anticipated.
So before intelligent machines can takeover as our servants, let alone takeover the world, they first have to learn to walk around and properly navigate the environment.
"Any intelligent agent in the real world needs to do three tasks: perception, planning, and language," explained Saxena.
These "rudimentary" tasks are what RoboBrain is really aimed at figuring out. Currently, robots pose more of a danger to humans by causing clumsy accidents, such as running into things that cross their paths or knocking things over, than they do in regards to global domination.
Even so, maybe one day, when our toasters refuse to toast and our smartphones live up to their names, we'll look back on this project and regret it. Until then, best to appreciate our clumsy machines while we still can.
Related on MNN:
- 10 robots and machines from the history books
- Robot genius solves Rubik's Cube in record 3 seconds
- Has a computer program passed the Turing Test? Don't believe the hype