How Intelligence Emerges from Simple Agents and in Nature by Dr. Timothy Smith

Photo Source: Unsplash

In the age of artificial intelligence, a different kind of AI may open doors to robotics and therapeutics that depend not on computation but on emergent intelligence. Emergent intelligence refers to the phenomenon in which many simple agents, following basic rules, produce intelligent-seeming actions as a group. The group of simple agents becomes dramatically smarter than the sum of its parts. No central boss or plan drives the group's actions. Local rules governing how agents interact with their neighbors give rise to sophisticated global patterns. Three examples of emergent intelligence found in nature illustrate this phenomenon in action.

The first example involves the remarkable ability of ants to find the shortest route to a food source without a central planning system to map it. When scout ants go out to find food, they lay down a scent trail of chemicals called pheromones that other ants can follow. Once food has been discovered, other ants detect the pheromone and venture out to collect it, also leaving pheromones as they go. Pheromones degrade over time, and the ants that return to the hive faster than others leave a stronger, more concentrated trail that fades less than longer trails. The intelligence of route optimization occurs because the shorter trail gets selected by chemistry and time.

A second type of emergent intelligence, called flocking, schooling, or murmuration, occurs when birds or fish tightly group together and move in mesmerizing, undulating clouds. No command bird or fish dictates the group's movements. Rather, each agent follows simple rules. These rules include staying close to your neighbors, not hitting your neighbors, and flying or swimming in the same direction as your neighbors. Such simple rules help a flock or school avoid predators or obstacles. 

A third type of emergent intelligence, called stigmergy, comes from the Greek stigma (sign) and ergon (work) and involves indirect coordination in which individuals communicate and collaborate by modifying their shared environment. Termites build massive mounds with sophisticated heating, cooling, and ventilation systems without a central boss telling each termite what to do. They use a pheromone system in which each termite drops a mud pellet wherever it detects a chemical attractant, and pheromone-laced pellets attract more deposits nearby. This basic rule, along with the environment that affects how long the pheromones persist, results in chambers, vents, and passageways in a complex termite mound.

Emergent intelligence challenges the assumption that intelligence requires a brain. It suggests that sophisticated behavior derives from relationships, not from individual components. This profound idea has implications that range from robotics to understanding consciousness itself. Take, for example, a bag of marbles emptied at the top of a hill with some holes in it. The marbles will roll down the hill, filling the holes with other marbles, using the marbles in the holes as stepping stones on the way down to the bottom of the hill. To an observer, the behavior may look coordinated with the early marbles filling holes to help the later marbles get from the top of the hill to the bottom. Is this real intelligence, or is it simply the physics of gravity driving the process, not some emergent intelligence?

We cannot know for sure what the mind of an ant, termite, or bird truly perceives, but using simple agents that follow rules that determine their interactions has important implications for robot design. Using groups of simple robots with basic interaction rules generates emergent intelligence. Researchers at Cornell University have shown that tiny robots with limited mobility but that can assist each other in simple ways can move together through complex environments, but not alone. (news.cornell.edu) The future cancer therapy could use small, simple nanobots that signal to each other the presence of tumor cells using pheromone-like markers, which signal other nanobots to kill only the tumor cells. The emergent outcome of billions of rule-following nanobots could be the most targeted, precise cancer therapy ever developed. The nature of intelligence may not be resolved here, but the properties of emergent intelligence promise remarkable opportunities.

Dr. Smith’s career in scientific and information research spans the areas of bioinformatics, artificial intelligence, toxicology, and chemistry. He has published a number of peer-reviewed scientific papers. He has worked over the past seventeen years developing advanced analytics, machine learning, and knowledge management tools to enable research and support high-level decision making. Tim completed his Ph.D. in Toxicology at Cornell University and a Bachelor of Science in chemistry from the University of Washington.

You can buy his book on Amazon in paperback and in kindle format here.