Harvard scientists have developed a self-assembling swarm of robots that can complete tasks on command.

According to the Washington Post, the Kilobots receive a 2-D shape and then arrange themselves to mimic whatever they are shown. There are 1,024 Kilobots and luckily they do not appear capable of organizing and plotting against their creators.

"I tell people that these robots are not very dangerous. The only way that they could hurt you is if you try to eat one. They can't even go over a piece of paper. So they're kind of stuck where they are," Study lead author Michael Rubenstein, a computer scientist at Harvard, told Reuters.

Leaders of the study, published in the journal Science, said the Kilobots are an important step forward in collective artificial intelligence. Rubenstein said the Kilobots display collective AI, but are not capable of comprehensive thought.

"This is a 'collective' of robots - a group of robots that work together to complete a common goal," he told Reuters. "If you call collective artificial intelligence the ability of a 'collective' to start to behave as a single entity, you could call this collective artificial intelligence."

Each cost $14 to build and individual Kilobots stand two inches tall, are 1.2 inches in diameter and scuttle around on three tiny point-like legs. The Kilobots need a flat, smooth surface, like a dry erase board, to optimize their operation. Anything even slightly coarser, like paper for example, would hold them up.

"The research on this phenomenon is now proceeding because we are convinced that such nanoclusters lend themselves as catalysts, whether in fuel cells, in photocatalytic water splitting, or for other important reactions in chemical engineering," Armin Hoell, also of Harvard Berlin, said in a press release. "But we could not have achieved this result using only electron microscopy, since it can only display details and sections of the specimen.

"Small-angle X-ray scattering is indispensable for measuring general trends and averages."