I am very interested in Artificial Life, or A-Life for short. This is (most times) considered a sub-branch from Artificial Intelligence (AI), but it's probably more a mix of biology and computer-science.
Whereas AI tries to create life-like intelligence by programming all the knowledge into the program (top-down), A-Life tries to simulate small and simple aspects of life and then tries to make those aspects cooperate and develop intelligent behaviour. It's the contrast between order (AI) and chaos (A-Life), on a certain level. Or more religious, AI thinks there is one universal rule for everything, God, if your christian- or islamic-religious, whereas A-life thinks everything just is because of chance (chaos). "How to over simplify theories". Btw, I am a christian myself, how's that for paradox.
How could one look upon these "small aspects of live cooperating"? Well, much used examples are an ant hill or birds flocking in their flight. The last example is easiest to explain. To have a computer program simulate the flocking-behaviour of birds it's enough to make bird objects, that will abeid by a simple rule:A bird wants to stay at a certain distance from the other birds closest to it.. Each bird will try to keep this distance by matching direction and velocity of the other birds. This rule and it's implementation causes the birds to move in nice swirves. This is also caused by the interactions of the birds. If one moves of a little, some other birds will follow and the group will move in a slightly different direction.
A demo of this : Flocking Boids And another one: Swarm animation You need to be using an Java-enabled browser to see these demos.
An other aspect of A-Life is trying to solve problems by numbers, large numbers. A problem, for instance finding a route through a maze (say New York or London), is given to some x1000 walkers. Their walk is random but each one walks with a given pattern. Then the best 1% of the batch (that got furtherest) are used to reproduce a fresh batch, making new ones with combinations of the 1%-ers. This is usually done by taking the patterns of 2 parents, cutting them in half and putting the parts together crossed. Put them in empty walker-shells and you have 2 new ones. Sometime a little mutation is entered, though that doesn't always seem neccesary. This method doesn't garanty the best solution or a solution at all, but tests and real implementations give very good results. Some theory suggest that the solution will mostly be suboptimal, but there are solutions to that too.
And another implementation is a learning, brain-like neural net. This consists of neurons that have a inputs and an output. The output gives a weighted value of the inputs. Neurons are connected outputs to inputs, but as a network wherein an output is not just connected to 1 other neuron, but to several. The inputs of some of the outer neurons are connected to sensors ("eyes") and some of the outputs at the other side of the netwerk are used to evaluate the result (what ever it is). Now it learns because someone enters input and evaluates the output. The person then gives an signal to the neurons to adjust weights of the neurons until the sensory input and the output "match". Well ok, this is a bit too abstract, even for my taste. Lets just say this is used for face-recognition from camera's.
If you're interested read "Artificial Life" By Steven Levy ISBN: 0-679-40774-X. Highly readable, without too much jargon.
Well, my thoughts on it, that is. Science is talking about making nano-bots, micro (nano actually) size robots that'll perfrom a simple task for a short time in your body. Because they work together thay could attack a cancer cell or repair a blood vessel. That's all nice and good and it will be helpfull, but what if a certain batch of nanites (another colorfull name) doesn't die after the periode? Well, than they'd probably keep repairing stuff. And if they wear after a while, they may not "repair" only vessels.
Or another horrorscript: In a large batch of bloodcell nanites, used for assembling hemoglobine, some nanites collide or fail and start adding stuff to them selfs. And in a large enough batch some nanites might eventually repair themselves. And in that batch of selfrepaires (that obviously never die) , eventually a mutation might make them reproducing. Et voila, virus-like walkers. And, as Jeff Goldblum so nicely stated in Jurrasic Park, "Nature will find a way".