Could a situation like "The Terminator" ever happen?

[Coco Pops]

First off some light reading.

Automated gun kills 9, wounds 14
http://www.wired.com/dangerroom/2007...bot-cannon-ki/

Did the software kill soldiers
http://www.itweb.co.za/index.php?opt...rticle&id=6157

South Korea has automated gun turrets for defense
http://www.gizmag.com/korea-dodamm-s...-turret/17198/

Well that's pretty terminator like if you want... But if we keep advancing AI's into the future could such a situation actually occur where the robots turn on their own platoons? They would become like the Terminator now wouldn't they?

But how long would it take for us to get the point where an AI says "no" to orders, or does it's own thing?

[escyos]

AI would be dictated by its programming, but if it has to perform something that partly contradicts its programming it would most likely be unable to process such a command.

If you had an AI that was just a blank slate and had to learn itself, then morality and ethics would be something it would probably not regard. Humans die, so why shouldn't the machine help them achieve this goal in a quick and efficient face-melty method.

[Snowman37]

In Terminator, it wasn't machines that turned on man. Rather, it was Skynet that turned on man. Skynet was an Artificial Intelligence plugged into everything U.S. Military. Skynet attacked Russia, Russia retaliated, Judgment Day. Skynet presumably used automated machine shops to build machines, and eventually, Terminators.

For a Terminator scenario to go down, there has to be a fully self aware, thinking, artificial intelligence, capable of total autonomy. We lack the ability to create such programming at this time, and I doubt we will ever be able to create such programming. Until then, any machines seemingly turning are just malfunctions.

[Quizziard]

In Terminator, it wasn't machines that turned on man. Rather, it was Skynet that turned on man. Skynet was an Artificial Intelligence plugged into everything U.S. Military. Skynet attacked Russia, Russia retaliated, Judgment Day. Skynet presumably used automated machine shops to build machines, and eventually, Terminators.

For a Terminator scenario to go down, there has to be a fully self aware, thinking, artificial intelligence, capable of total autonomy. We lack the ability to create such programming at this time, and I doubt we will ever be able to create such programming. Until then, any machines seemingly turning are just malfunctions.

Even more so, it was machines designing and building machines. Designing and building new versions, better than "themselves" (including presumably new building/designing machines). For me, that's even less likely than AI.

[knowles2]

AI would be dictated by its programming, but if it has to perform something that partly contradicts its programming it would most likely be unable to process such a command.

That not true

http://www.nytimes.com/2012/06/26/te...pagewanted=all

This computer taught itself to recognise cats, it was not programme to recognise cats, it learnt on it own to recognise cats. It also learnt to recognise human limbs, faces. It only around 15% accurate at the moment but then it only a fraction of the size of a true human vision cortex.

[Goose]

If we can't even design a computer to pass the Turing Test, then I doubt we'll develop machines capable of becoming self-aware and taking over the world.

[knowles2]

If we can't even design a computer to pass the Turing Test, then I doubt we'll develop machines capable of becoming self-aware and taking over the world.

We can already design machines to past the Turing tests, some of them are reaching 90+ percent at passing the Turing tests, An article last year, that Google has one that could achieve 96% pass rate, that 96 people out of 100 people would not know they were talking to a computer if they came across it in a chat room.

[Quizziard]

If we can't even design a computer to pass the Turing Test, then I doubt we'll develop machines capable of becoming self-aware and taking over the world.

Yet. We've been building motorcars for 100 years and we're still crashing thousands a day. We've only been building computers (as we know them) for fifty or so years. Allowing for Moore's Law, give them another couple of years and I'm sure we'll see Turing beaten. And five years later, by apps on handheld devices. That doesn't make it AI though, merely very well programmed.

[Coco Pops]

If we can't even design a computer to pass the Turing Test, then I doubt we'll develop machines capable of becoming self-aware and taking over the world.

I don't think that's the most definitive test for a computer being sentient, or conscious? Not that I know of any others.

[Goose]

I don't think that's the most definitive test for a computer being sentient, or conscious? Not that I know of any others.

No, but if we can't even figure out how make computers talk, then I don't see how it'll develop a consciousness.

[Coco Pops]

Or we could end up in an Avatar kind of future.

Check this out. Actual experiment.

http://stratrisks.com/geostrat/6807?...8StratRisks%29

[Morbo]

No. because it'll probably run on Windows and then an anonymous group of hackers will create a botnet that just destroys it.

[knowles2]

What do you mean by talk. Because both Google now and siri talk. And machines have held full length text to text conversation