There’s a bit of hysterical UK media hype about a “grey goo” of self-replicating nanobots eventually taking over the planet - not helped by one of Prince Charles’s more wacky statements, and silly articles as referenced here.
However, it has made me think: presumably, eventually, the technology will exist to make real nanobots that could self-replicate, identifying and harvesting suitable materials, and building copies of themselves. I’m presuming, given enough of them and enough time, they might also mutate, natural selection allowing more efficient and (perhaps) more ‘intelligent’ nanobots.
I’m not suggesting that there is any real danger, but I wonder, purely speculatively, should a fourth law be added to Asimov’s conjecture: “Robots must not self-replicate”?
I say that all self-replicators should be designed so that flaws show up in their copies after a certain number of reproductions, and grow until they deteriorate completely. This will give them a lifespan, not unlike the ones that biological beings have. This way, if something goes wrong and they start growing out of control, the sudden burst of growth will be slowed as they become defective at a faster rate.
There’s a hypothesis that the reason why animal cells have a timer on them, that causes them to wear out after a certain number of splittings, is to slow the growth of cancer. The cancer cells age much faster than normal because they split more often, and it slows the growth. The side effect is that even without cancer the cells start deteriorating after a while, but this wouldn’t happen very often in the wild, and it’s delayed until after you have had all your chances to reproduce. Aging could be natures way of controlling cancer.
Hmmmm… flaws in the replication mechanism…
this would have to be approached very carefully - the flaws in the biological replication system are responsible for the existence of all the organisms on this planet. The system would have to be very carefully designed to avoid supplying the nanobots with an evolutionary mechanism.
The new law “Robots must not self replicate” is probably useful in Science Fiction, which is about the only place where Asimov’s other laws of robotics have any relevance.
There’s a difference between flaws in the reproductive system and flaws in the replication mechanism. The flaw in the reproduction system apears to be the telomere shortening system that garauntees a limited number of cell divisions. Something like this could be included with no possibility of inducing mutations.
As long as some environmental factor could be guaranteed not to subtly damage the device in just such a way as to disable the reproductive restriction, or damage the replication mechanisms in just such a way as to introduce a ‘mutation’ that bypasses the reproductive block.
Of course we’re talking purely speulative stuf here. But as with all these things we need to weigh up risk versus benefit. The idea of not using machines that could cure all disease because there is a one in 7 x10^342 chance of them ever malfunctioning would be just silly.
I suppose that would also depend on how many of the devices we’re planning to deploy - if it is anywhere near the same order of magnitude as the probability of them malfunctioning in a bad way (that could lead to runaway replication), then it is worth worrying about. It is a bit like the creationist probability arguments - the chances of self-replicating protein sequence arising spontaneously are minuscule, but given enough throws of the dice, it becomes almost inevitable.
The benefit of nanotech-based manufacturing would be lost or impossible without the ability of them to reproduce. One of the primary benefits is that you would design the product to be manufactured, then only crank out a small amout of bots programmed to build it, and they reproduce to the quantity to complete the job.
Would one envisage “worker” nanobots and “constructor” nanobots, or just a single self-replicator? I’m getting the image of a beehive here (and for some reason I’m also thinking about Fraggle Rock).
These things could establish themselves as a new ecology, assuming they do become adaptive…
it need not mean the end of normal biological life on Earth, but that is a possibility.
I tend to expect that nanomachines will be mostly made of proteins and other organic molecules, and will have specialised forms for different functions within a particular swarm (energy collection and storage, data processing, material gathering, self-replication and manufacturing)…
like an ant hill but more so.
There is plenty of opportunity for adaptation among such varied clades.
Having said that, parity checks in the design memory (in the DNA equivalent) should be able to make mutations practically impossible, if everything works correctly.
The short version is its not going to happen. Nanobots currently cannot survive in the open, cannot move around except for miniscule distances - which is why they are useful for medical treatments where they can be circulated by the body, and are easily killed by a simple electrical current. More directly, no nanobot can, as we understand it, be a universal builder or universal destroyer. In other words, any nanobot will be very limited in what it can and cannot use as a material. Plus, where will they get power from? Current technology could make some, but they have very limited power sources, only what can be carried with them.
And not one of those flaws can be easily corrected. Well, at least not without shattering the boundaries of physics.
WHat you are looking for is basically something that can make its own power, take down any material and then use it to make more nanobots (well, if it has the right materials), and is permanently in active mode. And its the size of a cell or smaller.
I can guarrantee you that such a thing cannot be built, according to our current understanding of physics.
Good question; I can’t remember how they were described in “Engine Of Creation” by Drexler, the bible of nanotech. Maybe someone who read it more recently can explain his view.
I can imagine either having two classes as you describe, or a universal nanobot that could do both the final work plus replicate itself.
No, microscopic (or macroscopic0 biological life cannot take in all materials and use them, nor can it destroy on the scale required by the “Grey Goo” scenario. They are basically devoted solely to taking in enough nutrients to survive, and have a very limited range of what they can use.
This is even worse - these things ould have to be very coordinated. How are you going to organize a bunch of mcro-germs?
Well, the main advantage that nanotechnology has over evolved life is that it contains data processing units which can be connected to larger control functions memory banks, the internet, even to human direct control;
it is wrong to imagine nanotechnology as being swarms of nano-sized robots creeping about, although that is part of the story… there also will be support microbots and mesobots, solar power arrays and power storage cells… like a living organism, but separated into discrete parts wherever necessary.
There is no thermodynamic difference between the phenomenon of life, and the phenomenon of nanotechnology;
similarly there is no difference between a human culture and a nanoswarm from an organisational point of view- one is just much smaller and more focused in intent than the other.
It is absurd to suggest that some attempt along these lines will not be made and eventually succeed or maybe kill us all.
Well, the main advantage that nanotechnology has over evolved life is that it contains data processing units which can be connected to larger control functions memory banks, the internet, even to human direct control;
it is wrong to imagine nanotechnology as being swarms of nano-sized robots creeping about, although that is part of the story… there also will be support microbots and mesobots, solar power arrays and power storage cells… like a living organism, but separated into discrete parts wherever necessary.
There is no thermodynamic difference between the phenomenon of life, and the phenomenon of nanotechnology;
similarly there is no difference between a human culture and a nanoswarm from an organisational point of view- one is just much smaller and more focused in intent than the other.
It is absurd to suggest that some attempt along these lines will not be made and eventually succeed or maybe kill us all.
I’m sorry, but this is dead wrong. nanomachines cannot be easily controlled. In fact, they can do nothing more than respond to extremely simple commands. They cannot be coordinated. What are you going to do, put little antenna on them? They cannot think. They only respond to immediate impulses. Having a “control system” would be pointless, because the bots would need minds to control!
Tne ony control system is what stimuli they respond to and what tools they have. If you give them a senstivity to a certain bacteria’ shell, and a tool that attaches a chemical marker to draw he attention of white blood cells, you’ve made a germ-killer. But that’s it. You can’t program it to move to the esophagus and hunt down bacteria, and then move to the stomach to watch out for a nasty stomach infection.
Plus, think about the scale difference. From your angle, its quite simple to tell a bunch of nanobots to organize themselves into a square one inch by one inch. From their end, that’s miles upon miles away. And they have no real sensory systems. You can’t put eyes or ears on them. They have no idea where they are.
Yes, Nanotech basically IS, the science of making mechanical germs.
Ha! A nanoswarm has no organization! Its a big pile of nanites.
Yeah, but if there is a) a large volume of the critters, b) self-replication, and c) the chance of mutation, then they don’t need control or organization (see “natural selection”).