I understand that when we try to implement swarm intelligence, it basically has multiple independent units communicating and with each other. What I am unsure of is if swarm intelligence an extension of artificial intelligence or if it is a discipline in and of itself.

4 Answers 4


I would call it a subset of AI. It's artificial intelligence but using a more decentralized model.


I've used Swarm Intelligence in my thesis and I'm studying AI in my master. I would say that AI is a huge field and Swarm Intelligence would intersect both AI - multi-agent system - and Operations Research - optimization - fields.


swam intelligence is based less on the learning and who currently has the needed information. For example an Full AI may learn and know things on its own. You give it some infomation and it stores it the best way it can. In SI when a node does not know something it asks the other nodes if they know. If the other nodes dont know then that node is forced to figure out what to do. If another node then comes around asked the same question then the node that learned it before just returns the answer. In some cases the swam has to work togetahter to learn "bigger" things but that is really based on how the swam is setup.


I propose that Swarm Intelligence is both Artificial Intelligence and a subfield of Artificial Intelligence.

Swarm Intelligence as a subfield implies (through action) that intelligent systems can cluster and work alone as clusters, solving problems, where each subsystem within each cluster depends on peer systems to solve an ultimate problem.

Artificial Intelligence, if you follow the concept Minsky described the human brain by in "Society of Mind," is a bunch of swarms and/or problem-solving systems coming together to solve 'big' problems. The subsystems in each swarm solve small problems (which may be analyzing obstructions and figuring out if help is needed); the swarm solves the problems each subsystem encountered via analyses.

I think the difference between Swarm Intelligence and Artificial Intelligence exists relative to the complexity of the system being gauged for intelligence. For example: Take one bee (the insect) and show it a meadow full of flowers. The bee can look at the set of flowers and realize that it doesn't have the capacity to carry all the nectar from the meadow back to its hive. That realization, in itself, can be looked at as intelligence ('artificial' if the bee was man-made). When the bee goes back to its hive and asks other bees to help it with the acquisition of the nectar, the other bees will agree. Once they all reach the meadow and disperse evenly among the flowers, if they all share the same problem and are working for the same cause--which may be to make a cup of honey--each bee is part of one large intelligence. So the swarm of bees was an intelligence on its own.

That's what I assume, but correct me if I'm wrong in my assumption.

  • Not all AI solutions that I've seen have a swarm of subsystems (in fact, "swarm" solutions are the minority of what I've seen). For example, I've seen pathfinding algorithms that are implemented as a single graph traversal. No swarms involved. Sep 12, 2011 at 18:36
  • @FrustratedWithFormsDesigner I know, but the collective intelligence is what can create one intelligence. That's where I'm getting at in my post.
    – Mr_Spock
    Sep 12, 2011 at 21:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.