Artificial Life in Virtual Environments (ALIVE)

Intelligence from the ground up...

For close to 60 years, researchers in Artificial Intelligence have tirelessly worked toward developing systems that exhibit intelligent behvaior. As a community, we have made great strides toward achieving this goal. However, these systems require a tremendous amount of engineering and computational power and in the end still remain only as robust as their original creators could imagine they would need to be.

In stark contrast, we are surrounding with examples of systems that are tremendously adaptable yet have only the most rudimentary processing capabilities. One perfect example is the nematode Caenorhabditis elegans, which has only 302 neurons operating at less than 1 KHz. With these neurons, the organism is able to do a remarkable number of activities, which one might argue resemble intelligence. C. elegans is capable of adapting its body motions to move through dirt, water, and on flat surfaces. It is able to find food using a combination of exploration and chemotaxis, follow thermal gradients, and locate mates. They also demonstrate touch response, circadian rhythms, and show rudimentary learning through touch habituation.

Taking the perspective that the best way to understand something is to construct it, we are embarking on a new project to build a highly-detailed software simulation of an actual living organism, the nematode C. elegans The goals of this project are two-fold. First, we hope to develop more accurate models of neurons and nervous systems to use as the basis for next-generation control in AI. Second, we hope that by building this simulation, we will provide the biology community with a detailed training and research experimentation platform.