Can robots reproduce?

It is rumored that when Descartes left France to work as the tutor of young Queen Christina of Sweden he was asked by his royal student what could be said of the human body. Descartes answered that one could regard as a machine; whereby the Queen pointed to a clock and ordered him to “see to it that it produces offspring”.

Since this anecdotal conversation there have been many who imagined machines that reproduced. Stanislaw Lem in his novel “The Invincible” (1964) recounted the story of a spaceship landing on a distant planet to find a mechanical life form, the product of millions of years of mechanical evolution. Interestingly, Lem’s lifeform exhibited swarm intelligence: relatively “dumb” parts united into a hyper-organism with hyper-intelligence.

John von Neuman

Philosophy and literature pointed the way that science and technology followed. Self-replicating machines have been proposed since 1802 when William Paley formulated the first teleological argument of machines producing other machines. A detailed model for mechanical self-replication was suggested by John von Neumann: a universal constructor that was both an active component of the construction as well as the target of the copying process. This meant that the medium of replication was at the same time the storage of instructions for the replication. This notion allowed open-ended complexity and therefore errors in the replication – in other words, it opened up self-replicating non-biological systems to the laws of evolution. Neumann’s brilliant insight predated the discovery of the double DNA helix by Crick and Watson.

Although von Neumann’s model works in the mathematical space of cellular automata it was a clear demonstration that evolution may influence mechanical evolution.

RepRap project: self-replicating machines

We may imagine several other ways of orchestrating robot reproduction. For instance a robotic factory with three classes of robots: one for mining and transporting raw material, one for assembling raw materials into finished robots and one for designing processes and products. The latter class, the “brains” of the autonomous robotic factory, would have to be AI. Could this ever happen?

On planet Earth safety legislation impedes, although it does not preclude, the development of a fully autonomous robotic factory that reproduces itself.  Nevertheless, planting such a factory on a distant planet is a different story. Mars colonization could benefit from self-reproducing robots preparing the planet for human habitation. George Dyson has proposed using self-replicating robots in order to cut and ferry ice from Engeladus (a frozen Saturn satellite) to Mars and use it to terraform it.

Nasty self-replicators

Science fiction has worked various possible scenarios for robot reproduction, the commonest of all being robotic life running amok. But maybe we are missing an important point here. In robotic reproduction guided by Artificial Intelligence evolution will play a minor role, if any. Error correction will be automated in a teleologically-guided evolution designed by the supervisory programs.

Unlike natural evolution where high-level consciousness and intelligence evolved very late as by-products of cerebral development, in robotic evolution they will be the guiding forces. Brains will come before bodies.

Ironically, robotic evolution will be Intelligent Design par excellence. Creators of complex machines will be themselves highly complex machines . In this scenario it is highly probable that self-replication will involve recursive self-improvement, until the original supervisory programs are superseded by the next generation of superintelligent designers. At which point we will have arrived at the singularity point of human civilization.