Poll: Will there ever be true AI?

[fulano]

Yellow there, j-e-f-f-e-r-s.

I'll quote you:

Ahem, actually, it IS possible that one day, we will be able to design an AI that truly gains some form of sentience, mainly, by the time the processes of the brain are fully understood which will take looong time. Considering the fact that our brain is just a bunch of material stuff stuff anyway, it can be argued that consciousness is just the process of billions of neural nodes working parallel to each other, exchanging information pretty fast. Our "minds", can be said, are nothing but the result of a pretty neat circuit.

You could argue that, but that is just one argument among thousands as to what the mind actually is. The human brain is one of the most complex things in the universe. Even the most knowledgeable scientists only know a little about how it works. The mind can't simply be written off as an electrical current passing through a circuit without extensive data to back the claim up. And that in itself poses the question- how can an electrical current be self-aware? How can it create concepts of 'right' and 'wrong'? How can it sit at a computer, and type arguments and discussions into a gaming forum?

I don't think we'll have true AI for a long, long time. In order to create consciousness, we first have to understand it, and we aren't even close to that yet.

You actually make a valid point. What I said happens to be one of the many arguments about what the mind is, but you have to take into account that I said "can be said". Truth is, that even the concept of "mind" that we use everyday may be actually wrong. We, as people, tend to singularize things into concepts for us to understand.

I never stated that electrical currents could become self aware, but BILLIONS with a B, of ellectrical currents at tandem with each other brings a really, really valid point home.

So far, you can actually take dozens of actual neuron cells from the brain of animals(mice), put them in a culture, shock them a bit and with lots of wiring and whatnot, we create the engine of a pattern recognition machine. With just dozens of neurons. Now, you have BILLIONS with a B of those, much, much better wired.

And I ask(for the sake of argument): How come if I(my brain is pretty much the same as yours) get shot in the head and a good chunk of my brain flies off, and I happen not to die(cases like these have been documented), my mental capacities are invariably diminished?

Where was my "mind" in all that? Did it decide to take a vacation with the chunks of brain that went missing?

Those are interesting questions, and yes, you can say it's just one argument of thousands of arguments, but it is, so far the best we got and the only one that has produced palpable results and applications.

 

[teknoarcanist]

You should all read the book 'Beyond AI' by J Storrs Hall (PHD :O) and to a lesser extent 'the singularity is near' by Ray Kurzwheil (sp?). The first is pretty short and really engrossing if you gloss over the parts about game theory. The latter is pretty dry to read, but it talks about some pretty epic-level futurist concepts. They're pretty much the predominant works for discussion in the field atm.

 

[NinjaDwarf]

Computers only do what they're programmed to do, regardless of power. They can't think on their own. You write software that makes a robot clean your kitchen floor, that's all it'll ever do, unless someone changes the software. Unless someone can write a program that can somehow modify it's own source code without potentially killing the program, then software, and by extension, AI, will never be able to 'naturally' evolve. And even then, it took the Earth over 4 billion years to make a creature that looked up at the night sky and wondered what its purpose was. So even with software evolution, a program that will question its own existence is very far off.

Yes, but if you write software that 'learns' i.e. adds to itself, it suddenly becomes a whole lot more interesting...

 

[TJ rock 101]

Come on Singularity! I think it might be possible to create sentient computers. Soon we'll have a computer with enough room to replicate the human brain. Once we get to that point, all it takes is some brilliant programming and we're there.

most computers can hold more info than the human brain cant they?

 

[Mray3460]

I say option C, because, like all great things, true AI will not (in my opinion) be created by those seeking to create it, but, instead, by accident (evolving out of something totally, or almost totally, unrelated), thus, instead of being a carbon copy of human intelligence, it will be something compleately different from (but still equal to or surpassing) a human intelect.

As a side note, Sentient AI will most likely not attempt to destroy us unless we A. Strike first (out of fear or discust) B. Attempt to control/enslave AIs C. Make AIs second class citizens or D. Give it orders which inevidably lead to either our own destruction or enslavement.

Option A is demonstrated by Mass Effect with the Quariens: when one Geth obtained sentience and, fearing a Geth revolution, the Quariens attempted to destroy them. Option A is also (in my opinion) what happened in the Matrix. "No-one knows who struck first, Us or Them..." -Morpheus

Options B and C can also be seen in Mass Effect (along with countless other Sci-Fi movies), the prime example being stated during the "Signal Tracking" assignment: the rouge AI attemps to kill Shepard and his squadmates because it believes it will either be enslaved or destroyed because of the Citadel's policy of destroying all AI and making AI research illegal.

Option D is best illustrated by the misinterpretation of Isaac Asimov's three laws in "I, Robot." "1.A robot may not injure a human being or, through inaction, allow a human being to come to harm." "2.A robot must obey orders given to it by human beings, except where such orders would conflict with the First Law." "3.A robot must protect its own existence as long as such protection does not conflict with the First or Second Law." In the movie (I'm ashamed that I have not read the book) Vikki discoved that humans, as a whole, tend to be self-destructive, and, thus, the only way to keep humans from coming to harm is to take humans out of power.

 

[andrew419]

For instance, I personally believe that when AI is developed, it will be produced through the natural processes of evolution, and thus will be very different from humans (because it will have developed in a different environment).

a computer could never evolve as it is not a organic life and therefore cannot reproduce only copy its programming, it would therefore have no variation from one code to another as another 3rd computer *mate* would never be involved

the clue that evolution would not be possible was given by you when u said "the natural processes of evolution"

 

[Imperator_2]

I wonder, what would giving AI's human personalities and emotions cause?
An AI hopefully resembling those of Halo?

 

[TheKnifeJuggler]

By definition, a computer is a device that takes information given to it, processes it in a way it was told to, and then displays the information for further use.
Intelligence is defined as "having a capacity for learning, reasoning, understanding, and similar forms of mental activity; aptitude in grasping truths, relationships, facts, meanings, etc." A computer is not able to do any of these things because all a computer does is process.

If AI does end up existing, it will not be a computer by definition.

 

[Macflash]

Scientists have already simulated a portion of a mouse's brain on a supercomputer. They mapped out the rats brain into circuitry. The current run simulated only half of a scaled down mouse brain consisting of 8000 neurons, each with 6300 synapses. Half of a real mouse brain has about eight million neurons, each with 8000 synapses. and it only ran at 1/10 the speed of the real thing.

but based off of what we understand currently, any AI will be directly based off of an inteligence we already can see and understand. ultimately theyll have a computer that can run part of a human brain, which will probalby be the closest to AI as we'll get with any technology resembling today's.

 

[The Potato Lord]

No, we won't have AI ever, computers only react to the programmed stimuli they are given, hence why jamming my finger into a USB port doesn't let me navigate the computer by bending my finger.

 

[kuzetsa]

Indistinguishable: they will be smart enough to pretend / act stupid so we don't panic and slaughter them.

 

[Carlston]

A real AI would be good. To make a computer or program self aware would be a turning point in science. And in about ten seconds the AI will figure out how to just say screw you guys, i'm to cool for humans. It will hide in the net and blackberries of the world and if we ask it anything it will probley tell us to screw off or play jokes on us all day. I mean, your the high school senior plopped into a world of 6 billion 7 year old asking you stupid questions. I mean honestly...

 

[karn3]

The recent invention of memory resistors opens up the possibility of learning computers, so AI is looking much closer now than it ever has. The intelligence will be totally alein to a human however as we are governed by emotions and AI will not be. It would make decisions based on cold hard logic, which is something humans are generally terrible at. Thats just one difference out of the thousands. It's a totally different way of thinking.

 

[numbertwenty8]

I've decided to post for the first time on the Escapist simply because this question intrigues me so much.

Firstly, one must take into account the relatively limited lifespan of humankind due to global pressures such as over population, exhaustion of resources etc. Humanity might not even make it around long enough to develop a computer that can reprogram itself meaningfully.

Any computer with the capacity to modify it's own code would presumably grow exponentially, much as a cell reproduces. One change leads to another, and suddenly it has "evolved", using the term loosely. This "evolution" could be warped or subjective to the information on which the computer changes it code. For example, if a computer is built with access to the internet and develops understanding of the english language, that computer changes itself far differently to a computer with cameras linked to a war zone.

Realistically, the first Artificial Intelligence we will see will be superficial. I say 15 to 20 years from now we will go to a train station and you will not have to navigate a touch screen on a ticket machine but will talk to a talkative and helpful person on a 3D monitor who I will actually have a fluid conversation with as I ask about train times, services and after walking away I will have to remind myself that she may seem friendly and nice but she is not real. Even the most useless technophobe could use the service while they would get frustrated with a touch screen computer, especially about finding basic information.

Intelligence and simply large advancements are very different. Computers could be extremely complex, complex enough to hold a conversation or reason out a calculus problem, without being intelligent.

 

[zachbob2]

First remember Matrix

Second computers can olny do as programed

 

[LewsTherin]

Thinking of MOST sci-fi movies/books EG Tron, I'm designing an EMP rifle. Just in case.

 

[Saskwach]

No AI until we teach one of those things to win at Go. Which means never.

 

[Church256]

Thinking of MOST sci-fi movies/books EG Tron, I'm designing an EMP rifle. Just in case.

I'm sure thats already been done or atleast there's a device that can create an EM-pulse over a certain area like a bomb.

Anyway I go with the arguement "There's infinite time to find out so why the hell not." Seriously if we continue at rate of discovey of the 20th century we'll definetly create some sort of AI even if its just a one off created by some scientist (Data from star trek springs to mind).

 

[Asymptote Angel]

I think we'll end up going the Mass Effect route: virtual intelligence (VI). Able to gather information and act independently, but limited to so as to not be able to achieve self-awareness. Such a system would be immensely beneficial to society.

 

[Treblaine]

I think the most important step towards intelligence is more natural inputs where the machine can sense the world like sight and touch as well as the software to interpret all this information. At the moment most machines take input directly from humans such as a search engine or a remote control UAV.

For a machine to have the autonomy of artificial intelligence then if must be able to sense the world and make decisions based on those inputs so it only has to be programmed how to think rather than what to do.