Digital Evolution

Evolution is no longer limited to just the natural world, as it has reached digital worlds as well. Artificial creations living in computational environments now can learn to adapt to them and can solve problems given to them by researchers.

Biologically, evolution is achieved through DNA replication, selection, and variation (eg. through mutations). Researchers have come up with unique ways to reproduce these three conditions in a computerized environment. For instance, replication can be performed by essentially copying and pasting a data structure. The result of implementing these three conditions is something called an evolutionary algorithm -- which is essentially a set of instructions the computer can follow to evolve and adapt.

Digital evolution draws much inspiration from biological evolution, and researchers in this field study evolutionary processes that are done by computational methods. Excitingly enough, researchers have found ways to implement natural principles that guide biological evolution in a digital environment.

The gradual process of the digital organisms evolving, however, often comes with hilarious surprises. Researchers in the field of digital evolution have countless stories of how their algorithms surprised them with creative and unexpected “solutions” or bugs that reveal just how creative these complex organisms can be. For instance, Cully et. al (2015) designed a simulated robot and tasked it with the problem of walking without using its “feet.”  

The robot flipped itself on its back and walked on its “elbows/knees.” Karl Sims, a computer graphics artist and researcher from MIT, tasked his AI with becoming the most “evolutionarily fit” by jumping the highest, only to find the robots cheated by simply making themselves as tall as possible, or somersaulting so their “feet” would be as high as possible and consequently appear as if they had jumped that high.

These algorithms’ abilities to constantly surprise us shows just how much we still have to learn in the field of digital evolution, but these robots’ amazing and out-of-the-box innovations promise a world of new discoveries and possibilities.

This article was inspired by a research paper which can be read here. Cover Art: "Innovation" by Chloe Barreau.