6

Artificial intelligence (AI) refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. The term may also be applied to any machine that exhibits traits associated with a human mind such as learning and problem-solving.

According to the Wikipedia article on swarm intelligence

Swarm intelligence (SI) is the collective behavior of decentralized, self-organized systems, natural or artificial. The concept is employed in work on artificial intelligence.

The application of swarm principles to robots is called swarm robotics, while 'swarm intelligence' refers to the more general set of algorithms.

SI systems consist typically of a population of simple agents or boids interacting locally with one another and with their environment. The inspiration often comes from nature, especially biological systems.

These two terms seem to be related, especially in their application in computer science and software engineering. Is one a subset of another? Is one tool (SI) is used to build a system for the other(AI)? What are their differences and why are they significant?

Community
  • 1
Pluviophile
  • 1,223
  • 5
  • 17
  • 37

3 Answers3

3

Swarm intelligence (SI) is a sub-field of or an approach to artificial intelligence (AI), where you have multiple individuals (for example, artificial ants), which collectively can produce what we (or most of us) would intuitively call intelligent behaviour.

SI is sometimes categorized as a sub-field of evolutionary computation (which also includes evolutionary algorithms, such as genetic algorithms, genetic programming, evolution strategies, and so on), which is often considered a sub-field of AI or techniques to produce artificial intelligence, because all these techniques are often based on the use of multiple individuals/solutions (that either compete or collaborate with each other).

One of the most commonly used SI techniques is ant colony optimization algorithms (proposed by M. Dorigo and further developed by other people like Luca M. Gambardella), which have been successfully applied to solve the non-decision version of the NP-complete problem (in simple words, it's a combinatorial problem that may require exponential time to be solved in the usual case) known as the travelling salesman problem. There are other SI techniques, which are somehow similar to ACO algorithms, such as particle swarm optimization or the artificial bee colony algorithm.

Occasionally, SI may also be categorised as a sub-field of computational intelligence, which often refers to specific techniques to create artificially intelligent systems (i.e. programs that exhibit what we would call intelligence) that are more based on or inspired by the biology, such as neural networks, genetic algorithms, or, in fact, SI algorithms, such as ACO algorithms. However, CI can also be considered a sub-field of AI, given that it studies techniques to produce artificial intelligence, so, in the end, as I said above, SI is an approach to AI, which includes other approaches, such as evolutionary algorithms, rule-based systems, deep learning or other machine learning techniques.

nbro
  • 39,006
  • 12
  • 98
  • 176
  • What are `artificial ants`? Could you please elaborate on the example? – Pluviophile Jan 25 '21 at 14:37
  • 1
    Artificial ants are the "individuals" (i.e. the solutions to a problem or the programming abstraction that finds those solutions) in the ant colony optimization algorithms that I mention in my answer. If you learn a little bit about ACO algorithms, you will understand more about them. I suggest that you read, for example, http://www.scholarpedia.org/article/Ant_colony_optimization for more details. – nbro Jan 25 '21 at 14:39
2

Well, one of the simpler definitions for SI sounds like this:

The emergent collective intelligence of groups of simple agents.” (Bonabeau et al, 1999)

So, in order to get to the SI you have to use some kind of algorithms/AI to get simple intelligent agents. It's just cooperative intelligence, or cooperative AI if you wish. SI just uses today's AI/ML techniques to build the swarm, in same manners as reinforcement learning uses AI/ML techniques to make agents that can behave reasonably in large spaces by approximating value functions V(S) and policies pi(S). I hope this helps a little.

So AI/ML is kinda of a tool plugged in SI, as SI is field with it's own algorithm definitions and theory.

Igor
  • 77
  • 4
  • This answer is partially wrong. SI doesn't use today's AI/ML to build a swarm. It would be like saying that evolutionary algorithms use ML techniques (e.g. SVMs) to build a population. So, no, AI/ML is not kind of a tool plugged into SI. Actually, it's more the other way around. SI techniques are sometimes used in ML (for example, for hyper-parameter optization, see PSO). Not sure why people are upvoting this. I suggest that you edit this answer to address/fix these points/issues that I raised. – nbro Jan 25 '21 at 16:19
2

Artificial Intelligence, as its name suggests, is intelligence made by humans. It's usually thought of as having human-like behaviors and characteristics. However, it doesn't have to resemble humans to be AI. It just has to be made by humans. Many common AI algorithms aren't even made to resemble humans, they may just have similarities. Reinforcement learning is present in humans, but also in the many creatures with intelligence.

Swarm Intelligence is basically a lot of small stupid things working together to do something complex. Take for example ants. Each individual ant only follows a few very simple "instructions" like if has this chemical: follow. Like Evolutionary AI, it's taking mimicking features of nature. We humans just take a feature made by nature (Evolution/Swarming) and try to replicate some behaviors. Much like Evolutionary AI, Swarm Intelligence is a type of AI.

tl;dr:

  • AI: intelligence made by humans
  • SI: feature made by nature that humans are trying to copy
ChocolateOverflow
  • 129
  • 2
  • I think it's a bit restrictive or pointless to say that AI is intelligence made by humans. It would be better to say that AI is intelligence not found directly in nature or biological systems or organisms. Also, saying "Swarm Intelligence is basically a lot of **small stupid things** working together to do something _complex_." is not only quite provocative, but also not very clear at all. Also, ants do not follow deterministic if-then rules, but probabilistic ones, so this answer is quite misleading and actually partially wrong. – nbro Jan 25 '21 at 16:22
  • For example, "Like Evolutionary AI, it's part of nature.", no, not necessarily, in the sense that artificial ACO algorithms, for example, are not part of nature (unless you consider computer programs an extension of ourselves and part of nature in the sense that they are made of substances that are present in nature). So, I suggest that you edit this post to address my concerns, then your answer seems more or less reasonable to me. Not sure why people are upvoting this when there is some info that is not fully correct. – nbro Jan 25 '21 at 16:24