So let’s say a group of robots needed “food” to survive. Whenever they find food, they not only go to “eat” it, but signal others that food is there.

Of course, this triggers more robots to show up. What if too many robots show up to feed and there isn’t enough food? Would the robots learn not to signal?

In this Wired article that is exactly what happens. The robots learn to conceal their signal so that they can feed and survive.

A true survival instinct can now be placed in the minds of robots.

This kind of breakthrough is huge.

Robots are expensive – they represent a huge investment to a company that buys them. If the company knows that the robots will actively work at ensuring their own safety, than corporations will be more willing to buy. Robots as they are now in industry and other applied areas are largely “dumb” – they do their job and do it as assigned. By giving them the ability to decide the best way to move and to act, they can do more complex tasks and do so without harm to themselves.

While this experiment might seem small, it really represents a large leap forward in robotics.