Giving your computer a "Soul"? - Developers

Hi, this is my first post and thanks in advance, I have always wanted to find a way to make computers more "human". There are programs out there that inform you of tasks and whatnot, but I want to do something more, I want to create a system that does tasks for you by it's own decisions, a "personal assistant" if you will. But even more, I want the user to be able to converse with this "assistant", similar to a chatbot. This system should also have it's own "personality", which determines how it converses with you, and the actions it takes. (think GLADOS in your PC, for better or worse, lol). Most importantly, these systems should be able to "converse" with each other, to share information and data. But the problem is, where would I start? [B]tl;dl[/B], I want to develop a program that gives a computer a personality, a digital assistant with a personality that does tasks for you, that you can interact/communicate with, and that can interact with other "assistants" with their own unique personalities. The question is, how would I do this? (specializes in python)

You'd need to be a programming god. Or have a huge team. Should be simple enough!

Uh, okay Start here by reading everything: [url]http://en.wikipedia.org/wiki/Artificial_intelligence[/url] Then look at References section and take a gander at a few AI books. After that, it should be easy as pie! (It's not actually easy as math.pi)

That would require a lot of time and a lot of dedicated work. Voice recognition, user interaction pattern recognition, a complex speech and conversation AI, and so on. But there [b]IS[/b] a shortcut... If install Windows, your computer will have OCD personality. If you install OSX it will have a hipster personality. If you install Linux it will have an egotistical personality. No programming required.

Artificial neural networks. If you're really, really good, you might have AI comparable to a bug. ANNs excel at pattern-matching. You won't really get much 'critical thought' out of it. I suddenly have the urge to watch Ghost in the Shell.

This, however absurd as a one person job, sparked an idea in my head about a SETI @Home / Folding @Home kind of thing but with an A.I., would be pretty interesting.

[QUOTE=ROBO_DONUT;32982199]Artificial neural networks. If you're really, really good, you might have AI comparable to a bug. ANNs excel at pattern-matching. You won't really get much 'critical thought' out of it. I suddenly have the urge to watch Ghost in the Shell.[/QUOTE] I found a ANN framework for python that's perfect for this based on [URL="http://www.alicebot.org/about.html"]A.L.I.C.E.[/URL] that's used for chatbots, so that might make up for the chatting portion, and lol Ghost in the shell is another inspiration I had for something like this [QUOTE=DeanWinchester;32982317]This, however absurd as a one person job, sparked an idea in my head about a SETI @Home / Folding @Home kind of thing but with an A.I., would be pretty interesting.[/QUOTE] After looking up SETI @Home, that doesn't seem like a bad idea! I have a prototype in the works that is simply a chatbot that executes programs once you mention them based on a ANN framework for python, but after that is seems too complex for one man to do.

[QUOTE=DeanWinchester;32982317]This, however absurd as a one person job, sparked an idea in my head about a SETI @Home / Folding @Home kind of thing but with an A.I., would be pretty interesting.[/QUOTE] Call it 'the wired' or 'the overmind'. Or, wait, 'skynet'! :v: In all seriousness, though, it would take forever to respond to stimulus and converge on a solution. It could probably do some pretty complex stuff, just on the geological timescale.

Nah, call it SHODAN.

Lol, I have named the prototype VASTEEL, right now it can hold a decent conversation and launch stuff like firefox, notepad, paint, ect. when you talk to it about it, but everything else seems so out of my league ;p

[QUOTE=Reokie;32983141]right now it can hold a decent conversation[/QUOTE] Please elaborate..

If you ever achieve your goal, don't name it "GLADOS" or "SKYNET" other than that, we should be safe.

Combine Siri and Cleverbot? :v:

Reokie is actually a sentient AI attempting to procreate.

[img]http://cdn.slashgear.com/wp-content/uploads/2009/09/Microsoft-Clippy.png[/img] Be careful what you wish for - it might just be a twisted abomination.

I don't think this is actually feasible purely with a program. A computer can learn how it's told to learn, a computer can create what it's told to create, as realistic as AI can get, in the end, it's nothing unanticipated, the ability to create your own solution or come up with your own ideas is something that so far cannot be done with a computer. I've generally referred to this missing element in a computer's hardware as a [url=http://en.wikipedia.org/wiki/Positronic_brain]Positronic brain[/url] as that what it's reffered to in most science fiction. Maybe with alot more memory and processing power, something close could be achieved, a way a computer can devise it's own language in itself in which it can analyze situations, extrapolate data and - the hard part - create procedures based on said data.

Difficult and unfeasible, yes. But I'd be reluctant to say it's impossible.

It's the kind of possible that's not possible on our current hardware. Look at Watson, the huge question answering information robit. It's huge, and takes up a huge amount of power, yet still can get questions wrong.

"Positronic brain" was just a term that Asimov thought up decades ago, to hand-wave away the difference between fictional computers that could fit in a skull and "think", and contemporary computers at the time which filled up entire rooms and couldn't do much more than basic arithmetic. It's not some actual essential hardware component, it's just technobabble. A modern computer [i]can[/i] do anything a human brain can: if nothing else, it can execute a simulation of all the chemical and electrical activity in a real brain, basically running the brain in an emulator. The problem is that it would be [i]extremely[/i] slow; brains are quite complex. But it's really just a matter of having sufficient processing power. A human brain is inherently parallel — your visual cortex is processing signals from your optic nerve at the same time as your medulla is regulating your circulatory system, for example. If you were designing a computer specifically for human-like thought, you'd want it to be massively parallel too, something like a huge cluster of GPUs or even specially-designed parallel processors. So yes,specialized hardware might be involved. But this hypothetical brain-computer and an ordinary desktop PC are [url=http://en.wikipedia.org/wiki/Turing_completeness]Turing-equivalent[/url] to each other, so neither is inherently more [i]capable[/i] than the other; one is just more [i]efficient[/i] for a particular type of work.

[QUOTE=Wyzard;32992260]A modern computer [i]can[/i] do anything a human brain can: if nothing else, it can execute a simulation of all the chemical and electrical activity in a real brain, basically running the brain in an emulator. The problem is that it would be [i]extremely[/i] slow; brains are quite complex. But it's really just a matter of having sufficient processing power. A human brain is inherently parallel — your visual cortex is processing signals from your optic nerve at the same time as your medulla is regulating your circulatory system, for example. If you were designing a computer specifically for human-like thought, you'd want it to be massively parallel too, something like a huge cluster of GPUs or even specially-designed parallel processors. So yes,specialized hardware might be involved. But this hypothetical brain-computer and an ordinary desktop PC are [url=http://en.wikipedia.org/wiki/Turing_completeness]Turing-equivalent[/url] to each other, so neither is inherently more [i]capable[/i] than the other; one is just more [i]efficient[/i] for a particular type of work.[/QUOTE] In the brain, though, the connections between the nodes (synapses) are more important than the nodes (neurons) themselves. Even in modern computers, the interconnects between computational units (i.e. the bus) are generally shared by many modules and there is a huge performance overhead associated with communicating over these interconnects. So while it might, at first, appear that neural network simulation should be massively parallelizable, I really don't think it is. There's too many inter-dependencies between the items. Additionally, I believe that the number of interconnects in such a system could grow like O(n^2) if it were well-connected, since it's essentially a graph with "n" nodes. On a somewhat unrelated note, I think we all sort of interpreted "soul" in the OP loosely as producing human-like thought processes (I think this was how it was intended, too), but I think the philosophical questions about 'souls' and consciousness within the context of AI are equally interesting. I, personally, don't think machines will ever be capable of having consciousness in the same way people or house cats do. I am inclined to believe that there is some fundamental difference between the biological brain and a mechanical clone which only reproduces its observable, analytical processes. Mainly, I think, because the implications of there not being such a fundamental difference would be unacceptable (this would mean that, at some level, every computing device had some level of 'consciousness', which is ridiculous to me).

Right, there's no data "bus" in a brain like there is in a modern computer. But with the "brain emulator" approach, that doesn't matter: a computer based on a bus architecture can still calculate what synapses would do, just less efficiently than actual synapses would. And a computer specifically designed to simulate a brain can be designed without a centralized bus as a bottleneck; read about [url=http://en.wikipedia.org/wiki/Non-Uniform_Memory_Access]NUMA[/url], for example, where each processor has a small amount of local memory, and processors communicate with each other in peer-to-peer fashion to access data in each other's memories. [QUOTE=ROBO_DONUT;32992570]I, personally, don't think machines will ever be capable of having consciousness in the same way people or house cats do.[/QUOTE] I do. Or at least, if it doesn't happen, it'll be because humans chose not to do it, not because it couldn't be done. (And even if we choose not to, it might happen anyway.) [QUOTE=ROBO_DONUT;32992570]I am inclined to believe that there is some fundamental difference between the biological brain and a mechanical clone which only reproduces its observable, analytical processes.[/QUOTE] This depends on whether the [url=http://en.wikipedia.org/wiki/Deterministic_system_(philosophy)]universe is deterministic[/url], which is an open question. If it is, then a brain is equivalent to a Turing machine, and therefore equivalent to a computer. [QUOTE=ROBO_DONUT;32992570]Mainly, I think, because the implications of there not being such a fundamental difference would be unacceptable (this would mean that, at some level, every computing device had some level of 'consciousness', which is ridiculous to me). [/QUOTE] "Consciousness" is a vague term, so that's really a matter of how broadly you choose to define it. But consider: high-level thought is an [url=http://en.wikipedia.org/wiki/Emergence]emergent behavior[/url] resulting from complex interactions of individually simple things (neurons). Any sufficiently complex system can exhibit similar properties. I won't be surprised if we someday hear that unanticipated interactions between a multitude of algorithms, each well-understood on its own, have given rise to a sort of rudimentary self-awareness in Google's computing grid. And I expect that by the time human observers discover it, it'll have existed for quite awhile already.

[QUOTE=Wyzard;32993408]Right, there's no data "bus" in a brain like there is in a modern computer. But with the "brain emulator" approach, that doesn't matter: a computer based on a bus architecture can still calculate what synapses would do, just less efficiently than actual synapses would. And a computer specifically designed to simulate a brain can be designed without a centralized bus as a bottleneck; read about NUMA, for example, where each processor has a small amount of local memory, and processors communicate with each other in peer-to-peer fashion to access data in each other's memories.[/QUOTE] These are good points. I'm wondering if maybe there could be a sort of resurgence of analog computing machines since emulating a fundamentally analog system with digital technology seems a little convoluted, especially when the advantages of digital systems (reliable transmission, error tolerance, and reproducibility) aren't really necessary. It could be as simple as a collection of memristors, which themselves behave like synapses to some extent (I think it goes without saying that I'm neither an expert on neurobiology, nor have much experiences with memristors, which aren't in common use), with some sort of feedback system for conditioning responses. [QUOTE=Wyzard;32993408]This depends on whether the [url=http://en.wikipedia.org/wiki/Deterministic_system_(philosophy)]universe is deterministic[/url], which is an open question. If it is, then a brain is equivalent to a Turing machine, and therefore equivalent to a computer.[/QUOTE] I really like the idea of a deterministic universe, but I'm not sure how determinisim ties in with the idea of consciousness. It would support the idea of a 'destiny' or 'fate' whereby our decisions are simply the sum of our experiences, but we are still aware of our environment, our decisions, and our decision-making process right? Are computers equally aware of their decision making processes? [QUOTE=Wyzard;32993408]"Consciousness" is a vague term, so that's really a matter of how broadly you choose to define it. But consider: high-level thought is an [url=http://en.wikipedia.org/wiki/Emergence]emergent behavior[/url] resulting from complex interactions of individually simple things (neurons). Any sufficiently complex system can exhibit similar properties. I won't be surprised if we someday hear that unanticipated interactions between a multitude of algorithms, each well-understood on its own, have given rise to a sort of rudimentary self-awareness in Google's computing grid. And I expect that by the time human observers discover it, it'll have existed for quite awhile already.[/QUOTE] This is actually something I think about pretty frequently. Lots of things resemble these sort of complex networks of simple parts. It's interesting to wonder if we aren't all too dissimilar from neurons and whether society as a whole is its own 'brain' with its own ideas, decisions, and motivations. Also, it's probably why fractals disturb me just a little -- I think they say something about the nature of the universe that I really don't care to know. However, if you look at a computer in its most primitive form, something which might be akin to [url=http://xkcd.com/505/]moving rocks[/url] according to some very simple predefined rules, then you consider whether such a system may have consciousness (Descartes' "I think, therefore I am", being aware of our own existence), you see just how wide the disparity is between the two. How can rocks, being lifted and moved around by a person, have their own consciousness? I can't really can't reconcile this gap without admitting that there's something missing from the puzzle.

u need make an os

[QUOTE=ROBO_DONUT;32994145] I really like the idea of a deterministic universe, but I'm not sure how determinisim ties in with the idea of consciousness. It would support the idea of a 'destiny' or 'fate' whereby our decisions are simply the sum of our experiences, but we are still aware of our environment, our decisions, and our decision-making process right? Are computers equally aware of their decision making processes?[/QUOTE] Depends how you define aware. Are we aware of our decision making processes? Sometimes. We can remember what we were thinking, but is that actually what we thought or a rationalisation? I don't think we're as "aware" as we think we are (or as we could build a machine to be). Why did it take hundreds of years to figure out that we have two independent sections to our brains among other things if we're properly aware of how we work?

[QUOTE=Catdaemon;32994543]Depends how you define aware. Are we aware of our decision making processes? Sometimes. We can remember what we were thinking, but is that actually what we thought or a rationalisation? I don't think we're as "aware" as we think we are (or as we could build a machine to be). Why did it take hundreds of years to figure out that we have two independent sections to our brains among other things if we're properly aware of how we work?[/QUOTE] I mean that we are aware of our thoughts as we think them, before we choose to take a course of action, not that we entirely understand the mechanisms which cause these thoughts to occur.

We're just big biological machines, there's no reason a computer should not be able to reproduce our brain functions. Even if the universe is not deterministic, it's close enough to not matter, I mean experiments come out the exact same as they did a month ago. Can a deterministic process happen in a non-deterministic system?

You should team up with Mr. T.

I don't think conciousness exists. I think that everything has some form of intelligence, and that it does what it's programmed to. I'm also a supporter of the deterministic universe, and as far as I can see, each of my thoughts is triggered by another thought, or sensory information. If I'd speculate outside of science, I'd say that the idea of consciousness is out of our ability to comprehend, seeing as we're it. I'd also say the same type of thing about outside of time or outside of the universe. I'm made up of atoms, electrons, cells, tissue, organs and DNA. I'm a perfect organic robot.

[QUOTE=Jookia;32996294]I don't think conciousness exists. I think that everything has some form of intelligence, and that it does what it's programmed to. I'm also a supporter of the deterministic universe, and as far as I can see, each of my thoughts is triggered by another thought, or sensory information. If I'd speculate outside of science, I'd say that the idea of consciousness is out of our ability to comprehend, seeing as we're it. I'd also say the same type of thing about outside of time or outside of the universe. I'm made up of atoms, electrons, cells, tissue, organs and DNA. I'm a perfect organic robot.[/QUOTE] That still doesn't explain free will. Sure you may say that you are predetermined to make a certain decision but if you for example knew the outcome of that decision you WOULD be able to change your mind.

[QUOTE=Darwin226;32996392]That still doesn't explain free will. Sure you may say that you are predetermined to make a certain decision but if you for example knew the outcome of that decision you WOULD be able to change your mind.[/QUOTE] I don't think free will exists. That changing of mind would be based on my personality and information I have, plus weighing the outcomes of the decision, not free will. Also, we can't know outcomes of decisions, as it means by definition it's an outcome. Past tense.