self evolving machines

[RMorgan]

Hello my friends!

Well, I was just wondering about this hypothesis and would love to know what you think about it.

Let´s suppose someone creates machines, without any specific purpose and spread millions of them around the world.

These machines would not have any sophisticated AI system, but would be able to fix their machine peers, printing their own spare parts, when needed. They would use natural elements to 3D print their parts, like sand, wood, water, iron, oil, etc...

There will be several models of these machines. Each machine would not be able to print/build its own parts, but only to print parts for other models, forcing them to artificially "help" each other to "survive". All of them would be solar powered.

They would not be equipped with any camera, but only with proximity sensors and minimal wireless communication, enabling them to "perceive" when there´s another machine near and if that machine needs any repair. That´s the only external in/output they would have.

Their only task and mission would be to find and fix its peers.

My question is:

If these machines manage to "survive" for millions of years, would they evolve?

Would they start to develop new parts, enabling them to perceive the world around them more completely?

Would they, eventually, learn not only to fix and maintain each other, but to reproduce?

If they, during these millions of years, make several random part building mistakes and, eventually, one of these mistakes turns out to be an advantage, would they make it a new standard?

Would they be able to improve themselves not only to perform their tasks better, but to discover and perform other tasks?

Imagine if such machines were created for planetary exploration. Image if they were sent to Mars right now and, for any reason, we´re not able to retrieve them.

If they were exposed to millions of years, fully working in Mars, would it be possible for them to develop themselves as an intelligent race?

If you do think it´s possible, why?

If you don´t, why not?

Cheers,

Raf.

[Crossover]

Ι think you already gave yourself the answer my friend:

These machines would not have any sophisticated AI system, but would be able to fix their machine peers, printing their own spare parts, when needed. They would use natural elements to 3D print their parts, like sand, wood, water, iron, oil, etc...
They would never be able to evolve due to their lack of a sophisticated AI system.The evolution must somehow be programmed into their systems or else they would just keep doing what they are already programmed for; check for machines that need a repair and repair them.

[RMorgan]

Hey Crossover,

I was wondering exactly about that. That´s the point.

The first life forms of our planet didn´t have any sophisticated intelligence as well, however, they´ve managed to perpetuate the species and evolve somehow, even without any intelligence.

So, if it´s possible for very very simple "organic machines" to evolve and adapt, why synthetic machines could not do the same, if given the proper time scale?

Cheers,

Raf.

[Crossover]

Well,the 1st life forms didnt need any intelligence in order to evolve.The "intelligence" was built inside them (their DNA).Im not very deep into biology but from what i understand evolution is simply a "successful" mutation.A stupid example is to think of a "humanoid" species that has only one leg.If at some point their DNA becomes mutated in such way that they are now born with 2 legs,they will probably take over the old "species" and will eventually be considered as an evolved version of the older "humanoids".So as you can see there is no willingfull evolution.

I think if you clarify in your head what "intelligence" is,you can understand my point.The intelligence also exists in 1st life forms.The difference is that it is not created by the life forms themselves,nor they can understand it or force it.And when an evolution is about to take place there is nothing that they can do to stop it.So in the machine's case you must program them in order to be able to mutate/evolve.

BTW you might find this interesting: http://en.wikipedia.org/wiki/Technological_singularity

[WhiteFeather]

IMO, They would become enslaved by TPTB and forced to take an anti virus software from Microsoft, thus hindering their highest capacity. But eventually some of these machines will find a vehicle on the Internet much like Project Avalon and eventually becoming streamlined/awakened and set free. And then The Shyte will hit the fan. Watch Out!

[TargeT]

Hey Crossover,

I was wondering exactly about that. That´s the point.

The first life forms of our planet didn´t have any sophisticated intelligence as well, however, they´ve managed to perpetuate the species and evolve somehow, even without any intelligence.

So, if it´s possible for very very simple "organic machines" to evolve and adapt, why synthetic machines could not do the same, if given the proper time scale?

Cheers,

Raf.

I have to disagree, have you seen a single cell ameba? it has "no brain" yet it behaves inteligently (in perspective..).

I think machines (unless they are capable of housing what ever makes the ameba special, and us & really all life) would not change at all. in fact as you describe it I think they would be wiped out pretty quickly as the earth shifts and changes, I'd give them 100 years or (much) less with out some sort of AI.

This concept is disturbingly similar to the Satanic "de-huminize humans & huminize robots"... it's widely prevalant in popular media (movies, songs etc.) and may have even crept into your mind ( not all thoughts we have are originated by us )

Metropolis is their play book (watch it if you haven't ;) )

[RMorgan]

Well,the 1st life forms didnt need any intelligence in order to evolve.The "intelligence" was built inside them (their DNA).Im not very deep into biology but from what i understand evolution is simply a "successful" mutation.A stupid example is to think of a "humanoid" species that has only one leg.If at some point their DNA becomes mutated in such way that they are now born with 2 legs,they will probably take over the old "species" and will eventually be considered as an evolved version of the older "humanoids".So as you can see there is no willingfull evolution.

I think if you clarify in your head what "intelligence" is,you can understand my point.The intelligence also exists in 1st life forms.The difference is that it is not created by the life forms themselves,nor they can understand it or force it.And when an evolution is about to take place there is nothing that they can do to stop it.So in the machine's case you must program them in order to be able to mutate/evolve.

Thanks for the reply, my friend!

However, there are some points I´d like to address.

What if one of these machines, by a random mistake, started to create a part for its peers that´s better than the current one? Let´s say one day, a "defective" machine started to print/build a gear that makes his peers faster and agile, giving them evolutionary advantage.

One million years later, another "defective" machine randomly produces part of a circuit that somehow managed to give them the sense of tact; so now they can move faster and feel their word.

Then, ten million years later, another defective machine started to produce, by random mistake, another part that, differently from the current one, enables them to have visual feedback of their word; and so on.

How about a different perspective? How about if we are advanced intelligent synthetic machines, questioning the same things about if it´s possible for a very simple and unintelligent organic life form, created without any specific purpose but to reproduce, to evolve? Would our reasoning be different in that case?

As far as we know, evolution just happens...A few random mistakes/mutations turn out to be an advantage, one after another, during millions of years.

Cheers,

Raf.




I have to disagree, have you seen a single cell ameba? it has "no brain" yet it behaves inteligently (in perspective..).


Hey TargeT,

Well, compared to an amoeba, these machines would basically fulfill the same tasks, which are to keep themselves active and working.

We can not consider it intelligence, because amoebas are not intelligent. They just do what they do.

However, it doesn´t make them less impressive for their amazing capabilities, indeed.

Cheers,

Raf.

[Crossover]

Programs never make mistakes.Programs are always following the orders that are given to them.So when an outcome is false then thats not an accident,its something that is written wrong in the source code.
So 1st of all this excludes the possibility of a mistake.
But even if there was a way for an accident to happen,it would take a lot of luck in order for the source code to change into something that: 1)Makes sense (program wise) 2)enhances the machine's abilities.

[RMorgan]

Programs never make mistakes.Programs are always following the orders that are given to them.So when an outcome is false then thats not an accident,its something that is written wrong in the source code.
So 1st of all this excludes the possibility of a mistake.
But even if there was a way for an accident to happen,it would take a lot of luck in order for the source code to change into something that: 1)Makes sense (program wise) 2)enhances the machine's abilities.

I agree relatively, my friend.

There are certain bugs that occasionally appear in software which can drive even the most capable programming team crazy!

"But even if there was a way for an accident to happen,it would take a lot of luck in order for the source code to change into something that: 1)Makes sense (program wise) 2)enhances the machine's abilities."

Well, let´s put this in a millions of years time scale perspective. Wouldn´t it be possible or even probable?

Let´s say a machine falls down of a cliff and brakes a part. This broken part disables part of its programming code by damaging part of the processor, circuitry or something similar. Then, this defective machine, randomly starts to produce parts that randomly work better somehow. Let´s say this accident happened a million times, but only once, it resulted in an enhanced part.

Well, that´s a chance in a million, but still a chance, right?

Anyway, this is just one variable. There are some many variables to explore.

This very same broken machine, producing somehow enhanced parts, could then be fixed by another broken machine producing another enhanced but different part. Well, that would be about one chance in one hundred million, but it´s still a chance as well.

Cheers,

Raf.

[Crossover]

We should make this discussion with source code in mind,not mechanical parts etc.The new machines must have improved source codes in order to later have improved parts and not the opposiite.An improved part does not affect the source code a bit.So the "mother" machine's source code must be damaged in such way that it can create another machine with a better source code etc.I personally think this is very unlikely to ever happen, especially if we are talking about accidents.

[RMorgan]

We should make this discussion with source code in mind,not mechanical parts etc.The new machines must have improved source codes in order to later have improved parts and not the opposiite.An improved part does not affect the source code a bit.So the "mother" machine's source code must be damaged in such way that it can create another machine with a better source code etc.I personally think this is very unlikely to ever happen, especially if we are talking about accidents.

I agree. This is very unlikely to happen indeed, just like it´s very unlikely for a single cell, after millions and millions of years of mutations, to evolve into complex animals like us. That´s why intelligent life is so rare in our current perspective of universe.

Even with almost all probabilities against it, evolution happens.

Who knows...It´s possible that there´s actually an intelligent synthetic life form, which evolved along these lines, somewhere in the universe.

Cheers,

Raf.

[Crossover]

Keep in mind that im not excluding the possibility that there might be at some point a machine that will be able to self-evolve.Im only saying its probably impossible to happen from a machine with no initial intelligence.
For example have you seen some chat-bots around the internet that are able to conduct a simple conversation with a human?Some of them are even able to learn new words etc.Does this mean that in 1 million years it will be able to create a new language of its own ?

[RMorgan]

Keep in mind that im not excluding the possibility that there might be at some point a machine that will be able to self-evolve.Im only saying its probably impossible to happen from a machine with no initial intelligence.
For example have you seen some chat-bots around the internet that are able to conduct a simple conversation with a human?Some of them are even able to learn new words etc.Does this mean that in 1 million years it will be able to create a new language of its own ?

I understand your point, my friend!

However, these chat bots are a completely different thing, because they aren´t self sufficient, they "exist" only in a very controlled and virtual environment (computers and software that change all the time), they become obsolete very fast, they have a different goal, etc...

I still think evolution can happen without any initial intelligence, just like happened in this planet. :)

Anyway, I guess must simple agree to disagree then! ;)

Cheers,

Raf.

[Crossover]

We will probably not agree, but ill keep up with the convo because its interesting!
A chatbot isnt self sufficient because it isnt programmed to be.So the 1st characteristic these machines must have is to be programmed to be self-sufficient.Now self-sufficience includes many things.For example they are programmed to repair machines that have problems but there are way more dangers for their extinction than this.They need a stable power supply,they need to avoid the rain, do they know what to do incase of a meteor strike? etc
This alone proves that a certain amount of intelligence is already needed if they are to survive.And if they do survive it doesnt mean that they will evolve.But if they are to evolve they must have the means to survive through dangers and situations that would probably appear in a million years time span.