The structure of the global catastrophe

Chapter 8. The risks connected with robots and nanotechnologies.

The word «nanotechnology» in the last years has been so worn out, that began to cause an allergy. On the one hand, this term have so stretched, that it began to name any colloid solution, and its image was spoiled as if nanotechnology is only a way of money-laundering. In Russia even the agencies of the real estate using in the name a prefix «nano». Therefore it is important to remind, that at the heart of concept about nanotechnology is the idea about molecular manufacture, that is about atomic assemblage of material objects by means of the microscopic manipulators named assemblers.

Actually, these molecular assemblers does not exist for the present, and many doubts in their practical realizability. Nanoassembler, on idea, represents the microscopic robot, in size with the live cell, capable under the program to construct material objects atom behind atom. Its basic feature that it theoretically can, in the presence of energy and materials to make its own copy, and quickly enough, during an order of 15 minutes by some estimations. It allows, having received at least one nanorobot to multiply them in unlimited quantity, and then to direct them on any tasks. Here opens grandiose prospects: for example, having thrown one nanorobot in a solution with nutrients, it is possible to grow up for some days in it the engine for a space rocket without any atomic defect, so, with the highest durability and indicators of reliability, draught and weights. Thus having spent for manufacture only cost of the most nutritious solution and energy. Which, in case of occurrence of such technology, also considerably will fall in price. Nanorobots, entered into a human body blood-groove, could correct all possible damages to it at cellular level. And so on.

The most important thing in about nanorobots is that all these fantastic possibilities became a reality, it is enough to make only one universal radio-guided nanorobot. That is in the development of nanotechnologies sooner or later most likely there will be a huge crisis or jump, original nanotechnological Singularity: before creation of nanorobot nanotechnology will be very much expansive branch with small return, and after - a horn of plenty.

To begin with I will result some the bright facts from first line science as sometimes we do not realise that, how much far we have moved. 9th number for 2007 of Russian edition of magazine «Scientific American» informs on creation of the molecular designer which allows to design everything from several standard blocks. In December 2007 was published 400-page “Roadmap of the development nanotechnology” on which tens scientists worked under the protection of well-known DARPA. Before creation of the first nanorobots (named in it «Atomically precise productive nanosystems» - nanosystems, allowing to carry out manufacture with atomic accuracy) according to this roadmap remains from 10 till 25 years.

The basic danger in the relationof nanotechnologies is considered distribution nanotechnologyical «grey goo», that is microscopic self-breeding robots. Its basic signs are that:

1. Diminutiveness.

2. Ability to self-reproduction.

3. Ability to independent distribution across all Earth.

4. Ability imperceptibly and effectively to destroy крупноразмерную technics and live organisms.

5. Anonymity.

6. Cheapness.

7. Autonomy from human (soldier).

Grey goo and based on it nanotechnological weapon is the higher expression of these principles uniting all of them together. However it is not so obligatory to unite all to uniform these principles to receive the dangerous and effective weapon - enough to realise some. Different combinations of these principles give different kinds robotic weapon. We will consider in the beginning dangerous robots.

The robot-spray

The basic problem with biological and chemical poisons are difficulties of their anonymous effective dispersion. This problem the tiny robot in size with a bird (for example, a model aircraft). The set of such robots could quickly and "pollinate" huge territory imperceptibly.

The self-reproducing robot.

Though it is considered, that for effective self-reproduction are necessary molecular nanotechnology probably, that it not so. Then quite macrodimensional robot could breed, using natural energy and materials. This process can be two-level and use the robot-uterus and robots-soldiers which it makes but which serve it. Creation of self-breeding robots is attractive because allows to create quickly and cheaply huge army or to develop large scale manufacture, especially in space where is expensive to send complete mechanisms. The risk consists in loss of control over such system capable to self-reproduction. It is important to underline that it is a question not of a strong universal artificial intellect, but about the system which has left from under the control with limited intelligence incapable to self-improvement. The big sizes and limited intelligence do it more vulnerable, and reduction of the sizes, increase of speed of reproduction and intellectuality increase its danger. A classical example of such threat in a biological kingdom - a locust. Probably, that such robot will contain biological elements as they will help to acquire faster substances from environment.

Cloud of microrobots

Such microrobots could be made as the weapon at huge factories, like modern factories for manufacture of chips, and even with application of the same technologies - lithography theoretically allows to make mobile parts, for example, small pendulums. At weight of several milligrammes such microrobots could fly freely in the atmosphere. Each such robot could contain enough poison to kill human or to create short circuit in the electrotechnical device. To attack all people on the Earth only a few tens tons of such robots were required. However if they are made on technologies and the prices of modern chips, such quantity will cost billions dollars.

The armies of large fighting robots leaving from under the control

Though the army of the USA is definitely aimed at full automation and replacement of people with robots, to this purpose more than ten years. As we already spoke, theoretically the certain robotised army can obtain the incorrect order, and start to attack all live beings, being inaccessible to cancellation of commands. To become real threat, it should be the world, huge army distributed on all around the Earth which do not have competitors.

The nanotehnological weapons

Nanotechnology allow to create very effective weapon which is capable to exterminate all people even without unlimited self-reproduction of this weapon. Roughly speaking, the cloud nanorobotоs can extend on a certain district - or across all Earth - to find out all people on it, to stick to them, to get into a blood-groove and then is synchronised to put a mortal blow. This flight is more dangerous than the blind biological weapon as against it quarantines do not operate and it cannot be stopped by not nanotechnological means prior to the beginning of attack. Also there is no simple dispersion and lose of copies. Therefore on 10 billion people with a stock 100 billion nanorobots, total weight in several grammes will suffice.

Further, if the robotics linearly develops, without grandiose jump - and such jump is possible only in case of occurrence of a superstrong artificial intellect - intermediate stages will include creation of independent universal robots of lesser sizes. Now we can see initial phases of this process. Even the largest systems now are not quite independent, though already exists androids, capable to perform simple work and the cars independently driving on a simple route. There are also more primitive mechanisms with the minimum weight in several grammes (for example, small helicopters) and experimental models of separate parts. Speed of progress in this area is very high. If in 2003 the majority of automatic cars could not start moving, in 2007 they have carried out the task on driving in a city with crossroads.

Therefore it is possible to tell, that before nanorobotоs there will be some more stages. These are independent cars-tanks, independent androids (in size with human or a dog), independent robots in size with a rat, about an insect, microrobots in size of millimetre and nanorobots. It is important to us to define, from what stage such robots can be dangerous to mankind. It is clear, what even several self-coping tanks are not dangerous. However level of danger increases more considerably, as more and more cheaply it is possible to make such robots, and also as easier to spread them all over the world. It is possible in process of reduction of their sizes and automation of their manafacuring especially, technologies of self-reproduction. If will be possible to stamp microrobots in size of mosquito for several cents for a piece they will already represent serious force. In the classical novel of Stanislav Lem "Invincible" «nanorobots» have size in several millimetres, but are capable to organize in complex structures. Further, recently, in connection with expansion of a cheap Chinese labour, on the second plan has departed the fact that even robots of the usual sizes can participate in manufacture themselves owing to the increasing automation of manufacture at factories. This process gradually goes, but it too can have a point of sharp exponential excess when the contribution of robots to their own manufacture will surpass the contribution of people. It will result in considerable reduction in price of such manufacture, and, hence, to increase in probability of creation armies of flying microrobots. One of possible technology of manufacuring of microrobots is to press them, as microcircuits, in lithographic process with etching of mobile parts.

Mutual strike by such armies of nanorobots can surpass on its catastrophic consequences an exchange of nuclear attacks. To believe in it is difficult, as it is difficult to think, that something very small can cause a huge damage. (Though technological evolution goes this way, that the smaller weapon has the bigger destroying force, and a nuclear bomb is in this row.) Strike by microrobots can not be such interesting entertainment as explosion of nuclear bomb, but can yield result as the ideal neutron bomb in the spirit of «empty town».

Microrobots can be applied and as tactical weapon and then they will fight with each other and with headquaters and as the weapon of intimidation and sweep, which function is carried out now by strategic nuclear forces. In this quality they can appear threat for all mankind, in case of casual or intended application. Thus microrobots surpass strategic nuclear forces - they allow to organise more imperceptible attack, more sudden, more anonymous, cheaper and putting bigger damage. However, it does not get staginess that can weaken their psychological influence - before the first real fighting application.

Unlimited spreading of self-breeding nanorobots

Possibility of this risk is specified for the first time by Drexler, and investigated in R.Frejtas's article «Problem of grey goo». In relation to nanorobots, no less than to AI, it is difficult to us to estimate probability of their occurrence and spreading because they still do not exist. At the same time creation of nanorobots has precedent in the field of biology, namely, the live cell is some kind of nanorobot. Fibers are self-gathering universal mechanisms, DNA - operating computer. In this sense and the artificial intellect has precedent in sense of human reason and a world science as image of superreason. Yudkowsky assumes, that from nanorobot and AI we are separated not by time or shortage of certain intermediate stages, but only by some absent knowledge. That is, if we possess sufficient knowledge, we could collect such sequence of DNA at which execution by a cell it would be formed nanoassembler - that is the robot, capable to collect other robots so, capable to self-reproduction. Often speak about nanofactories - that is factories which can create any designs from atoms and molecules. However nanofactory and nanoassembler are interchangeable because on universal nanofactory can create nanoassembler and on the contrary.

On the one hand, the idea that at each house will be nanofactory instead of microwave oven, making everything necessary, looks beautiful, but it demands realisation of measures of protection, bigger than if it was a question of a nuclear reactor in-home. Some systems of protection already was suggested, and they include the continuous ciphered connection of nanofactory to network, and complex self-checking of nanofactory. But, alas, all experiences on creation of absolutely protected electronics, optical disks, files have failed. It is thought, the reason of it that quantity of "brains" on the party of hackers much more, than on the party of the manufacturer, and the problem of the hacker is easier - not to defend all possible vulnerability, but to find at least one of them. Distribution of those or other systems of an artificial intellect too will make selection of keys of access to nanofactories easier.

Eric Dreksler estimates necessary quantity of atoms in nanorobot - a replicator which will represent something like minifactory with a conveyor tape and micro-machine tools, - in one billion. Each manipulator can carry out not less than one million operations a second that is typical for speed of work of enzymes. Then it can construct a device of billion atoms for 1000 seconds - that is to reproduce itself. This is approximately the same as for some bacteria which can split with speed one in 15 minutes, that is the same 1000 seconds. Such robot-replicator could breed 1 day prior to weight in 1 ton, and completely absorb weight of the Earth for 2 days. Catastrophe of this sort is called as "grey goo”. Because of small sizes of nanorobots within crucial first days this process will not have any external manifestations while billions nanorobots will be carried by a wind across all Earth. Only the direct hit of nuclear bomb in the centre of distribution in the very first hours could help. There are offers to make replicators incapable to breed in an environment in which is not present there is some crucial very rare chemical element. See article of R.Fraitas mentioned above «Problem of grey goo» where various scenarios of distribution dangerous nanorobots and protective counter-measures are considered more in detail. Fraitas notices, that nanorobots will show their presence by intensive allocation of heat in the course of reproduction, therefore it is important to adjust monitoring of environment about strange temperature anomalies. Besides, breeding nanorobots will require energy and material, and the source of both is only biomass.

R.Fraitas allocates some possible scenarios of grey goo:

«a grey plankton» - nanorobots, breeding at ocean and using resources of hydrates of methane at the bottom. They can destroy sea biosphere and lead to allocation of hotbed gases in atmosphere. The sea biosphere is extremely important as absorber СО, the generator of oxygen and food for people.

«a grey dust» - these nanorobots breed in air, creating an impenetrable barrier in atmosphere, the leader to «nuclear winter».

«the Grey lichen» - these nanorobots breed on rocks.

«the Grey goo eating a biomass» - as the most unpleasant variant. Thus the most favourable to the accelerated reproduction as the biomass contains also materials for construction, and an energy source.

The direct hit of a nuclear bomb in a flask with such replicator could destroy them, but even close hit - only to disseminate. The growth of bacterium is limited by the presence nutrient medium. If the universal replicator knows how to replace one atoms with others, it can consume almost any substance, except pure environments from one material. They can be very omnivorous also in an energy source choice, if possess the information on how to use different sources. Nevertheless provision of energy for grey goo will be more a challenge, than access to materials.

Probability of occurrence nanorobots and possible time for this event

Occurrence of microrobots in scales of grammes and miligramme looks practically inevitable, and all technologies for this purpose already exist. However they will not be replicators. (Progress in the field of miniaturization is described, according to some information, by factor in 8 % a year.)

But now nanorobots, which is less than bacterium in size, are in far prospect. If they are created by AI all possible harm from them can be brought in the column of risks of AI as it will operate their actions. (But nevertheless there is a variant, when AI would appeared clever enough to create nanorobots, and still too silly to supervise them). And even without AI more and more powerful computers will give chance to more precisely and faster calculate details of future micro-and nanorobots. Therefore we can expect, that progress in creation nanorobots will be accelerated.

However the state of affairs in the field is such that creation of nanorobots-replicators in the next years is improbable. Therefore it is possible to assume, that if nanorobots will be created without the aid of real AI, it will occur in an interval 2020-2040. If to compare nanotechnology to biotechnologies and AI, we will see, that these technologies is much less ready, and lag behind 20-30 years from the colleagues. Therefore chances that strong nanotechnology (that is nanoreplicators) will be created before AI, and before bioprinter are not so great.

Conclusions: we can face risks civilizational catastrophe created by microrobots, still before real nanorobots will be made. The smaller, cheaper and more capable to self-reproduction will be microrobots, the bigger damage they are capable to make. And the more subjects can possess them.