3 Answers

  1. First, machines do not evolve, in the strict sense of the word, but simply evolve in the wake of human progress and needs. But, if you do not get bored, then:
    1. human evolution occurs stochastically, under the influence of many factors, while machines develop purposefully, under the influence of progress
    2. the evolution of biological systems is multi-factorial and puts pressure on a bunch of traits at the same time, creating a working system. the machine is created for a specific purpose, and there are far fewer selection application points.
    3. a living organism has no purpose and strives for maximum versatility; a machine, on the contrary, always performs a specific task and, it is unlikely that people will make universal performers – it is expensive and pointless.
    4. machines do not reproduce
    5. machines will always become more perfect, about the person it is less fair

  2. Biological evolution (including human evolution) is a natural (spontaneous) process of adaptation to the environment. Moreover, the measure of an individual's evolutionary success is ultimately its reproductive success (large and fertile offspring). Perhaps in the future, intervention in human evolution by genetic engineering methods will begin, but so far-so. The basis of biological evolution is a combination of heredity and variability, plus natural selection.

    The evolution (development) of machines is still artificial.

    • Machines do not have “heredity” passed on to “descendants”. Even if some machines are made by others, the design of “descendants” is determined not by the design of “parents”, but by the program of their work. Moreover, the program is determined by the person, not the machine itself. And even self-learning neural networks, the specific details of which arose spontaneously and are unknown to humans, are guided by the final “goals” initially set by people.
    • The carrier of the “heredity” of machines and the source of the “variability” are people (machine designers): not counting manufacturing defects, breakdowns, and any other uncontrolled deviations. Or computers (ordinary, or self-learning and simulating a neural network) that implement tasks set by a person.

    Can machine evolution become similar to biological evolution? Yes! Such a machine (a kind of “Adam”) you can create one (with or without “Eve”).

    • Reproduction and heredity. I think no one will doubt that it is possible to create a machine that can – within the framework of a certain program-reproduce itself (and more than once), and provide the “descendants” with the original program for reproduction. This will ensure “reproduction” and “heredity”. There will also be “mutations” in the form of errors when sending the program to “descendants”. But they will be mostly harmful. You can't build an evolution based on them.
    • Variability. Therefore, it is better to make” mutability “part of the” genome ” (inherited program). You can, for example, program a certain level (probability) of “random mutation” of a descendant design element within the framework of a mathematical evolutionary algorithm (Evolution_algorithms). It is critically important not only to implement a ” mutation “in the descendant construct (in hardware), but also to pass it a MODIFIED program, in which the real descendant construct will be specified as the original construct – “mutated”. You can also ” mutate “the” mutation mechanism ” itself (the evolutionary algorithm used) and write the “mutated” version to some descendants.
    • Gene exchange (“sexual reproduction” or horizontal transfer of “genes”). Contrary to Wikipedia's claim ,a “genetic” evolutionary algorithm (Genetic algorithm) can be implemented without “crossing” (the descendant will have only one “parent”) and without exchanging” hereditary information ” at all, providing variability with mutations alone. But you can – in addition to “mutations” – also “shuffle genes” (exchange program elements: both in terms of the design, and in terms of the mutation algorithm and information exchange). This can also be implemented in any way:
      • As “sexual reproduction” with a combination of elements of programs (“heredity”) of two different mechanisms (you can have three or more). Sexual reproduction can be made a single species, or the main one, or a rare additional one.
      • You can also simulate “horizontal gene transfer “(gene exchange without reproduction). But this is more difficult, since our ” heredity “(program) does not yet affect the design of the carrier itself (mechanism), unlike single-celled organisms that use horizontal exchange. In other words, the design will have to be supplemented with the ability to change itself when the program changes – in accordance with these changes. Well, and oprelelitsya with the type of horizontal exchange: displacement of old program elements with new ones, or addition without displacement. In the latter case, it is also necessary to determine how and which of the duplicate elements will manifest in the “phenotype” (in its own design), as well as with the use and transfer of “duplicates” to descendants.

    That is, even with the current state of the art, it is possible to create evolving mechanisms that have heredity and variability, and also-optionally- “shuffle genes” by “sexual reproduction” or”horizontal transfer”. It is clear that they will have to create acceptable “living conditions”, providing raw materials and energy. But living organisms are also not autonomous and need food, drink, etc.

    Verify that the variability and fertility created mechanisms (natural selection at this stage will manifest itself in the fact that some mutants will be unable to “reproduction”), people will have to limit the number of exponentially growing “population”, taking on the role of “predator” (deriving from the game and utilize non-breeding foraging areas individuals and part of the “prolific”; the latter can be pick randomly and equiprobably).

    After that, you can start changing the “environmental conditions” and follow the evolution. For example, gradually reduce the strength of raw materials. You can expect that increasing the thickness of parts will be a useful tool. Or reducing the speed of their operation (reducing the mechanical load). Evolution and divergence will begin. And in the descendants of “non-adapted” individuals, the details will break and they will “die out”, leaving no descendants.

    Finally, you can arrange an artificial selection based on some feature. And bring out some less adapted “domesticated” mechanisms that are optimized not for reproduction, but, for example, for their appearance (“exterior”).

    But if self-replicating evolving mechanisms are able to extract raw materials and energy themselves in nature (or in a human city), they can turn out to be dangerous “parasites” on human civilization, or its dangerous competitor for resources. They can also learn to extract resources from each other and become “predators”. Just give the selection a clue (a new trait), and it will develop it…

  3. The evolution of machines is artificial – machines do not improve with the help of genes, but with the help of people who write better code. The evolution of humans is natural, or at least it should be

Leave a Reply

Categories