by Arielle Young

The artificial intelligence equipped for these miniature space-probe-ish robots on wheels is limited, but they have literally taken on an artificial life of their own, their designers claim. Their 'brains' are inspired by that of animals and have had the ability to evolve, in this case to deceive one another in a simple test. The creators of these foot-tall bots provisioned them with wheels, sensors and flashing light signals, all strung together by a digital neural network, and left them to their own devices in a testing arena. Small slits of paper labeled 'Food' and 'Poison' were placed at opposite ends of the arena with the assumption that a robot would identify and station itself by the sign labeled as 'Food', thus earning them points from their creators.

The initial reaction of the robots was essentially just rolling around and emitting a rather random light show. However, their ignorance did not last, which was noted after a few copies of the original robots' brains and some random reformatting:

"Soon the robots learned to follow the signals of others who’d gathered at the food. But there wasn’t enough space for all of them to feed, and the robots bumped and jostled for position. As before, only a few made it through the bottleneck of selection. And before long, they’d evolved to mute their signals, thus concealing their location," observed Brandon Keim of Wired Magazine.

However, the robots' signaling strategies are still very much in a prototypical stage. Their communication amongst one another apparently hit an equilibrium, but never ceased the use of light-flashing signaling completely, perhaps being one of their greatest downfalls in their pursuit to deceive. The little buggers' activity seems to mimic, in an underdeveloped fashion, what moths in nature do to signal one another; the moth eludes the biologist with its fantastic and sundry collection of complex signaling. The researchers wrote Monday in the Proceedings of the National Academy of Sciences:

“Evolutionary robotic systems implicitly encompass many behavioral components … thus allowing for an unbiased investigation of the factors driving signal evolution. The great degree of realism provided by evolutionary robotic systems thus provides a powerful tool for studies that cannot readily be performed with real organisms.”

Debatable, I say. With technology in a thick, vague cloud of different sciences constantly affecting its development, artificial intelligence (biology as an influence, for instance,) it may not be accurate to assume that a robot's 'brain' will not take on definite characteristics such as that of an organism. In addition, evolution is the key within the construct of this technology, creating values such as survival, perseverance, and a strong possibility of previous biological counterparts. All in all, robots may have what it takes to be organisms themselves... it may just be a matter of time, especially when biology is involved in artificial intelligence.

It is fascinating to see the development of artificial intelligence, and it is nearly frightening when said discovery pops out of a test run without being specifically theorized first. This is all quite reminiscent of Robot Wars... but without the human control. I wonder what the future has in store.

via WIRED