The other day my neighbor and I were watching some television in what I will call an Altered State. So we are flipping back and forth between a nature program about ants and something vaguely related to the ever shrinking size of computer chips and memory. I mentioned that ants (individually) seem to have a pretty primitive existence, something that could be expressed in a computer code. Their “thinking” seems to be mostly predetermined responses to stimuli, interpreting chemical signals, and reacting to those inputs. I also brought up the fact that at the rate that the physical size of memory is shrinking that we would soon (if it wasn’t already possible) be able fit the amount of programming required to “program” an ant onto a chip approximately the size of an ant.
Therefore my question is twofold; are ants simple enough creatures or is current programming advanced enough to be able to replicate (in theory) the daily existence of an ant. Second, is there a chip approximately the size of an ant that has the capacity to hold the theoretical aforementioned code? If the answer to both questions is no, could this become possible in the near future?
I’m not sure that there is a definite answer to this, so I’m going to put this in IMHO for right now. Mods, by all means if a great debate breaks out feel free to move this.
I think we could probably model the typical behaviours of an ant, or several-to-many ants, but simulating the actual nervous system of an ant in sufficient detail would still be quite a serious task. There are something like 100,000 neurons in an ant’s nervous system.
I realize that we couldn’t realistically create a robot ant that could fit into a natural ant colony, but each ant has a finite number of reactions to a finite number of stimuli. Surely that could be turned into a program and saved into a storage device of some sort. Creating the physical means to receive and process those stimuli and the physical means to react to those stimuli is definately beyond our current capabilities. I’m thinking more along the lines of a “virtual ant in a box” and whether or not that could be saved on something about the size of an ant.
If you’re not worried about modelling the internal workings of an ant’s nervous system, then yes, I’d say it could technically be done; the trouble would be in actually doing it - capturing all the subtleties of response and interaction without knowing what was driving them; reverse-engineering an ant, as it were.
I’d estimate that it’s well within our technological capabilities to model on fairly modest hardware, but difficult to impossible to practically do.
A man has allready tried this. His name escapes me, but the cockroach episode of The X Files was based on his work.
He had the idea that human minds were too complex. He thought that AI should start with easier goals like bugs. His bug robots (genghis, attilla, others) are the size of a shoebox. They act like real bugs. He’s wanted to miniaturize for a long time. One of his goals is to send a load of ant sized robots to another planet as probes. I can’t remember his name.
Never look at our artificial computing devices and think that they can model any biological process easily and well, especially the nervous system of even a tiny insect like an ant.
I am not saying it can’t be done but this is a trap even the greatest researchers have fallen into time and time again. One almost insurmountable problem is that computers are almost completely unlike any biological nervous system, even the simplest ones. The fundamental principles and methods of operation are completely different. We know how computers work because we built them but nervous systems are still probably the greatest set of mysteries in science. Nervous systems can easily handle problems like any level of pattern matching easily while computers tend to be terrible at it while computers can add up a string of 1 million numbers with ease.
What all this means is that we probably can model some ant behavior fairly well but getting it just right is extremely difficult because the systems are so very different. Nervous systems don’t operate through binary logic and IF THEN trees. They do something very differently. That means that we would both need to understand how the ants behavior is generated in its entirety if we wanted to catch every subtlety and nuance. We would also have to know how to model that using an ON OFF structure and have the programming tools to do that. We could make something that looks like it is behaving like an ant but getting it right is deceptively difficult.
I don’t know how well they do at actually simulating the behavior of real ants, but there are path-finding algorithms that use multiple agents modeled after ants.