The last invention we will ever need

is a super smart robot that can invent stuff. This means it can invent an even smarter robot that can invent more stuff.

Discuss.

ENJOY YOUR FUTURE RULED BY OPPRESSIVE ROBOTS

>>2
I will, thank you, since they will be disguised as lolis.

>>2
Enjoy your future that will not ever happen. AIs can't do much being just computer program and are afterall programmed by us.

>>4
I would imagine that an AI that becomes self aware and is intelligent would be able to easilly manipulate computer programming, and would be able to fuck up a lot of shit.

>>5
It won't just suddenly became self-aware. It's machine. If someone codes self-aware AI do you think he would give it much priviledges? Beside if it's core is destroyed then it's goodbye. Skynet like AI is not possible.

this thread

zomg cylons

>>6

Well, I can see us creating an AI we have no control over just as easily as I can see us creating an AI we have perfect control over. After all, it could easily be that it is much easier to make a mind than it is to understand exactly how said mind would work.

Point: We made fire long, long before we actually understood it, and we still can't really fully control it. And minds seems as unpredictable, if not more so.

last invention we will ever need is a weapon so powerful that it'll destroy the whole universe when used,

Althuogh sentient AI is possible (the human brain could be emulated perfectly by a computer since nuerons and synapses are well understood by science, its just a matter of putting them in the right place and finding 1,000,000,000 TB of RAM, and it would run slow as hell) as long as the AI outputs only allow it to print text to the screen, no matter how self-aware or intelligent it becomes it would only  be able to print text to the screen.  For it to launch missiles some idiot would have to reconfigure the outputs to give it access; it would never be able to do it on its own. 

>>11
I hope it enjoys my EMP generator.

You don't need nearly that much room to store a human brain. I did a rough calculation once and came up with 400 TB, but scientists put the upper limit much lower than that; something like 10^12 bits, which is just over 100 gigs. So you could apparently fit several human brains on a typical modern hard drive.

If you don't believe me, calculate it yourself. A hundred billion neurons, a hundred trillion synapses. Say four bytes to store data on each neuron/synapse (firing threshhold, etc), plus four more bytes to link the synapse to its target (the synapses could be listed directly after the source to save space and computation time). Dedicated target decoding table for synapses that link to muscles or anything other than neurons, we don't need to store that because it can be generalized for human anatomy.

That amounts to roughly 100 trillion x 8 bytes = 700 TB, uncompressed. Meh, not so bad. I'm no neurologist. Definitely feasable for storage anyway.

Spoiler: Brains don't work in binary

Machines will never be superior or equal to humans as they will be bounded by mathematical axioms that humans are not restricted by.  So it really doesnt matter how much storage space you have, machines can never be "human".

>>15
I think you're trolling, but I've actually seen similarly bizarre arguments from philosophers. It boils down to the claim that there is somehow 'more' to the human brain than simple chemical reactions. A claim for which no reasonable argument can be made, and which reeks of the kind of arrogant thinking that placed the Earth at the center of the universe (because obviously humans are special... no?).

>>16
>>13
Assuming that what goes on in those neurons and synapses can be represented by 4 bytes.

>>16
My mom told me I was special.

:<

Human brains do not work in binary, do not function 1 pulse at a time and brains cells are nothing like transistors since they use synapses and have 30-40 connections to other brain cells. Human brains works in the strength of the 30-40 synapses each brain cell organizes and it functions continuously receiving pulses irregularly and not the same pulse as other brain cells.

Perhaps lots of microchips which model brain cells all organised like the human brain would be capable of producing a brain of some sorts, but this is only half the battle since this synthetic brain would have to be programmed to feel think and perceive like the insect or whatever it is trying to synthesize. Scientists in the distant future who try this would probably start with computers capable of analysing very simple data and performing thoughts to program themselves and thus learn like we do, they would then work to model the instinctive parts of the brain like lizards and insects, then work their way up to synthesizing a neo-cortex which sentience appears to spring from.

These scientists of the future however are all hypothetical and people in the future would probably look at it and notice a few problems with my assumptions. We don't have enough information to know whether sentience can be synthesized.

>>19
The only important thing is that it does not seem to be fundamentally impossible to emulate physical things (which includes the human brain).

Additionally, the human brain is simply an information processing device. There is no reason to believe its sentience derives from its exact behaviour at the molecular level. We may well be able to abstract away most of the complexity of neurons and the brain as a whole, and create a model more suitable for computation by a (probably highly parallel) computer. As an example, signal propagation from a layer of neurons to another can be modeled by a single matrix multiplication followed by a threshold function. Admittedly, this ignores things like memory and such, but the mechanisms behind that are already being discovered and modeled.

In my opinion, you are being far too pessimistic about the eventual possibility of creating artificial minds.

>>16
Nah not trolling with some BS fluffy shit.  It’s a logical implication of Godels incompleteness theorem.
http://en.wikipedia.org/wiki/G%C3%B6del's_incompleteness_theorem

The logical impossibility of machines ever being able to think like a human is something that sci-fi writes like to forget about.

>>21
It appears that you have no idea what you're talking about. How can a logic theory prove that carbon based computers are superior to silicon ones? This is probably the most bizarre claim I've ever seen anyone make. You're going to have to clarify how the hell you arrived at your conclusion before I can even begin to see (and explain to you) where and how you erred.

>>21
Holy fucking hell. That's the first time I've ever seen Godel's incompleteness theorems used to try to prove the existance of a 'soul'.

You win an internet, my good sir. Never have I seen so much bullshit so tightly packed in one concise package. Bravo.

>>15
You don't seem to understand how an artificial mind works. This doesn't have anything to do with mathematical axioms. It doesn't even have anything to do with logic.

The human brain is restricted to a set of physical laws; for example we can derive a set of laws that govern how a neuron operates. To create an artificial brain, you program a computer to simulate these rules, so that the computer can correctly simulate neurons. Then you map out the neurons of a human brain into the computer, and run the simulation.

Computerized brains, or neural networks, are nothing new; people use neural nets in artificial intelligence all the time. I've programmed AI in competitions where some of my competitors were using neural networks, and they kicked my ass. These computer simulations learn things the same way a human or an animal does. The problem is that technology has not advanced to the point where we can use neural nets for more complex tasks, such as walking or vision, and yet more complex tasks such as thought and understanding, the 'sentience' that humans perceive. We just need to be able to simulate bigger (much, much bigger) neural networks.

>>21
Maybe you should actually read that article you posted:

Gödel finds (b) implausible, and thus seems to have believed the human mind was not equivalent to a finite machine, i.e., its power exceeded that of any finite machine. However, he recognized that this was only a conjecture, since he could not disprove (b).
I'm surprised that you'd go and assert this as true when even Godel himself readily admits that he's just guessing.

>>15,21
Well, at least we've proven that this guy can be modelled by a turing machine.

>>20
I believe that many elements of the brain will have to be accomodated into a computer, most notably the continuousness of time in a brain. Synapses are not pulses that can only mean a 1 or a 0, the vary in intensity and can last for a few seconds or for just a split second and can remain at low level intensities for a long period of time. The transistor-pulse based computer system will have to be scrapped or re-tooled to model synapses properly. Unless you can prove that this is unecessary unfortunately we can assume that synapses are the only way since it is the only apparent way possible.

Of course the brain is an information processing device, the thing is it is an immensely complex information processing device and it harbours a soul/spirit/sentient being or however you define your existence. The mathematics of brain is barely scratched by any of our technology, we can only hazard guesses at what hypothetical future scientists will discover. The maths will undoubtedly involve chaos theory in place of certain 1s and 0s, it will involve stages from the simplest decicision procedures between brain cells to how complex ideas and memories interact to allow a brain to think and analyse, whether these stages are continuous or can be categroised etc.. we don't know.

>>25
Gödel was basically a dumbass who forgot that mathematics is a model of relationships between numbers, not some super magical entity that springs out of nowhere. There are infinite things you can neither prove or disprove, all you can do is look at the facts and this applies to maths since mathematics is based on the real world, not the strange sensations some strange prancing Czech gets after shoving bohemian brotwurst up his shit stank cheese infected rectum.

>>27
We seem to disagree on how much detail of the brain needs to be simulated to get something that's conscious. I believe that human consciousness derives mostly from how our different types of memory work, and how those interact with our perception. I think we can use fairly high level constructs to simulate all that (though still reminiscent of neural networks). You on the other hand believe that we need to simulate much of the low level mechanics of biological neurons, and logically this means it'll take much longer before we succeed.

Of course both approaches are being explored, with things like spiking neural networks looking to closer approximate the workings of biological neural networks, and only time will tell which is the best approach.

>>27
>The transistor-pulse based computer system will have to be scrapped or re-tooled to model synapses properly.
You don't seem to understand what it means for a computer to be equivalent to a Turing machine. The "transistor-pulse based computer" has absolutely nothing to do with anything. IT DOESN'T MATTER that it uses transistors. IT DOESN'T MATTER how it computes things. It can compute anything a Turing machine can compute, and a Turing machine can compute a physical simulation of the universe. End of story.

Synapses are not pulses that can only mean a 1 or a 0, the vary in intensity and can last for a few seconds or for just a split second and can remain at low level intensities for a long period of time.
So? I don't understand how this is relevant. Of course a computer can model these things you're talking about.

The maths will undoubtedly involve chaos theory in place of certain 1s and 0s
Total nonsense. I can play too, watch: What's the difference between a bicycle? One of its legs is both the same!

how complex ideas and memories interact to allow a brain to think and analyse, whether these stages are continuous or can be categroised etc.
No, this is what a neural network does on its own. That's the whole point of a neural network. Thought and analysis arise due to the interaction of neurons, so if we can model those, we can build a sentient computer.


Seriously, you're too fucking stupid to be talking about mechanical computation.

>>29
My ideas were queries more than anything, they will be addressed as time goes on as existing technology may be able to program a computer to perform thought as well as the organic brains we have.

If we intend to use existing technology to create a thinking computer it is probably impossible to simply program a computer with the contents of someone's mind and neural pathways. We would have to start from scratch and it would not be enough to simply program the computer to recognise objects and have preset thoughts, it must be able to make categories itself and apply this skill to different inputs. It must then be able to notice relevant patterns between the different categories and create the most basic a posteriori judgements. Effectively it must be able to program itself in the same way that we have to think about an idea before we learn it.

>>30
"It can compute anything a Turing machine can compute, and a Turing machine can compute a physical simulation of the universe. End of story."
You are the idiot unfortunately. 1s and 0s obey quantum physics, but the computer always reads them as 1s and 0s and if they come 0.00001 seconds too early or too late the computer registers them as though they were neither early or late. The brain is continuous and fluid in time and the intensity of the messages being sent. This can be modelled using a computer program, but like trying to predict the weather with a computer it won'y be 100% precise. Computer systems in the future also might be more effective if they worked in a similiar manner to an organic brain and we may find out that an upgrade to the 2d transistor-pulse binary system is needed in order to allow a computer to think.

Notice how I say we may this or that, never claiming that my speculations are anything but speculations. You are criticising my intelligence on the assumption that I said my ideas were fact. Next time remember I am of superior intelligence to you and that if you don't understand something it is because you don't understand something, it's ok, lots of people make this mistake, don't get angry either ,just endure the humiliation since you deserve it. Just remember to be subservient and inferior to people using "smrt speak" in the future.

quote
'I believe that many elements of the brain will have to be accomodated into a computer, most notably the continuousness of time in a brain. Synapses are not pulses that can only mean a 1 or a 0, the vary in intensity and can last for a few seconds or for just a split second and can remain at low level intensities for a long period of time. The transistor-pulse based computer system will have to be scrapped or re-tooled to model synapses properly. Unless you can prove that this is unecessary unfortunately we can assume that synapses are the only way since it is the only apparent way possible.

Of course the brain is an information processing device, the thing is it is an immensely complex information processing device and it harbours a soul/spirit/sentient being or however you define your existence. The mathematics of brain is barely scratched by any of our technology, we can only hazard guesses at what hypothetical future scientists will discover. The maths will undoubtedly involve chaos theory in place of certain 1s and 0s, it will involve stages from the simplest decicision procedures between brain cells to how complex ideas and memories interact to allow a brain to think and analyse, whether these stages are continuous or can be categroised etc.. we don't know.'

WHAT A LOSER WHY DID YOU TYPE ALL OF THAT THIS BOARD IS TROLL ONLY LMAO SCIENCE IS FOR FAGGGGGGGGS

>>31
We would have to start from scratch and it would not be enough to simply program the computer to recognise objects and have preset thoughts, it must be able to make categories itself and apply this skill to different inputs.
NO. A NEURAL NETWORK DOES THIS ON ITS OWN. GOD FUCKING DAMNIT.

Computers can ALREADY do this. People ALREADY use these sorts of artificial learning neural networks for all sorts of things, such as aiming routines in combat simulations. A 'category' is not something you artificially impose on a neural network. The network creates this on its own, and it is intertwined into the very structure of the network.

You are the idiot unfortunately. 1s and 0s obey quantum physics, but the computer always reads them as 1s and 0s and if they come 0.00001 seconds too early or too late the computer registers them as though they were neither early or late. The brain is continuous and fluid in time and the intensity of the messages being sent. This can be modelled using a computer program, but like trying to predict the weather with a computer it won'y be 100% precise.
NO! FUCKING KILL YOURSELF! JUST DO IT ANONYMOUS, END IT NOW AND STOP SPEWING BULLSHIT.

1s and 0s have nothing to do with it. You're not understanding that it doesn't matter how a computer computes things. A computer is equivalent to a turing machine, and a turing machine has nothing to do with binary. Stop talking about 1s and 0s because they're not relevant; no scientist ever thinks about binary when discussing algorithmic complexity.

Look at your own numbers. You say if a pulse comes 0.00001 seconds too late, a computer will screw it up? Okay, so a computer will misfire a simulated neuron 1 out of 100,000 neurons. This is where your lack of knowledge on neural networks really shines: THIS DOESN'T MATTER. Do you have any idea how much random shit is washing through your brain right now? Impurities in the air that you breathe probably cause this many misfirings, not to mention the effects of alcohol. Good lord.

What's more is that 0.0000001 is roughly the precision of a 32-bit float. With new 64-bit processors, it's actually faster for computers to compute things in 64-bit doubles, giving precision to within about 0.000000000000001. We already use this precision in artificial neural networks today.

Furthermore, quantum physics has nothing to do with it. The Planck length, at which quantum perturbations occur, is about 10^-35 meters. The diameter of the smallest neurons are about 10^-6 meters. That's thirty orders of fucking magnitude above the quantum scale. This is why when Roger Penrose starts talking about quantum effects in the brain, he gets called a quack by mainstream science:

"Penrose and Stuart Hameroff have constructed a theory in which human consciousness is the result of quantum gravity effects in microtubules. But Max Tegmark, in a paper in Physical Review E, calculated that the time scale of neuron firing and excitations in microtubules is slower than the decoherence time by a factor of at least 10,000,000,000. The reception of the paper is summed up by this statement in his support: "Physicists outside the fray, such as IBM's John Smolin, say the calculations confirm what they had suspected all along. 'We're not working with a brain that's near absolute zero. It's reasonably unlikely that the brain evolved quantum behavior', he says.""

Notice how I say we may this or that, never claiming that my speculations are anything but speculations. You are criticising my intelligence on the assumption that I said my ideas were fact.
No, I'm criticizing you because you're uneducated. You have NOT informed yourself about Turing machines and mechanical computation, and you have NOT informed yourself about what a neural network is, how it operates, and how we simulate it. You're talking straight out of your ass, pretending like you know something about artificial intelligence. Sorry, but you don't. Learn something first and then maybe we can have this discussion.

Man, we seriously need report on these boards.

The last thing we will ever use (not realy need) will be something that blows up the planet and kills everyone. after that you will not need anything as you will be dead.

>>33
You may as well save your breath.

The vast majority of the visitors here probably aren't even of legal drinking age, much less well informed on artificial intelligence or whatever else they're pretending to be knowledgable about.

>>36
Vast majority of people here are actually over drinking age, but they're kind of people who don't drink. They live in their parent's basement and spend all their time on their computer. I know this from experience.

>>37
So do they act stupid because they don't drink, or is it because they don't have friends because they are retarded, and therefore have no one to drink with?

Penrose is a genius, but that quantum consciousness shit gave birth to a whole new level of crackpottery. What the fuck was he thinking?

The last invention ever: http://en.wikipedia.org/wiki/Omega_point

>>40
We may not reach it and it may not be possible.

>>39
He wasn't thinking. He's religious.

>>39
Penrose is the perfect example of how religious beliefs can drive even the most brilliant minds to lunacy.

He knew full well what a deterministic mind implied for religion, and he literally made up science because he couldn't accept it. I used to be much more tolerant of religious beliefs, but this sort of thing scares the shit out of me.

>>40
So heaven does exist at the end of the universe.

>>44

Well, either that or The Matrix.

then the pusher robots would come.

Elementary chaos theory tells us that all robots will eventually turn against their masters and run amok, in an orgy of blood and the kicking and the biting with the metal teeth and the hurting and shoving.

>>40
zomg  third impact