Here’s my thing with this machines business, Phoneutria.
Now we know that organisms evolve (change) physically because of chemical errors during gene replication called mutations. These mutations are teleological accidents, meaning they aren’t random in a causal sense (as there is no such thing as randomess in a determinstic universe), but they aren’t planned or designed, either.
So each organism either possesses mutations which help it adapt and increase its fitness level (sexual selectivity included) in its environment, or possesses mutations that are ineffectual and make no difference, or it possesses mutations that decrease its fitness level.
When we say an organism is getting ‘better’, we mean it has undergone mutations which have increased its fitness level in an environment that is also changing.
So if a machine were invented, it would either have to have a logically replicating heritable unit like a gene that would be subject to mutation, or it would need to have the intelligence to be able to manipulate its own physical nature when the environment demanded such for the machines survival.
With the former, you’d basically have something like an organism, and it would mean little to call it a machine anymore. With the latter, you couldn’t just make a machine with a finite set of intelligence programs… it would have to somehow be able to learn, be able to process information which would identify dangerous changes in its environment, and execute the proper actions to modify itself to keep its fitness level.
Now this swarm intelligence you linked me too is interesting, but I fail to see how even if there is autopoiesis happening here, this holistic system of individual parts could attribute any of the kind of intelligence required to be able to sense dangerous changes in the machines environment.
Like the corrosive material example I mentioned… it is analogous to the human’s ability not only to correct itself in combating viruses and bacteria that threaten the body, but also develop ways to combat those things externally.
A machine might be able to sense damage to its own system by this material, but could it understand how the material itself develops, and devise a way to stop it?
How could a machine ever make that leap from ‘system senses corrosive material presence…repair, repair, repair’ to ‘where the fuck is that stuff coming from and how do we stop it.’
The advantage an organism has over this machine is that even if it doesn’t have the intelligence to solve such a problem, through several generations, mutations can occur genetically that make it immune to the problem anyway.
In the machines case, it couldn’t rely on this mutation as an adaptive strategy (because it has no genetic replicators), so it would have to figure out how to stop it if it were to survive. How do you program a machine to go from a turing machine with a set of finite programs to give it instruction, to a novel problem solver with an ability to learn?