Will future Artificial Intelligence try to end Mankind?

[Grouchio]

Today I was surfing the web when a frightening existential threat was brought up by Stephen Hawking:

http://www.bbc.com/news/technology-30290540

Prof Stephen Hawking, one of Britain's pre-eminent scientists, has said that efforts to create thinking machines pose a threat to our very existence.

He told the BBC:"The development of full artificial intelligence could spell the end of the human race."
His warning came in response to a question about a revamp of the technology he uses to communicate, which involves a basic form of AI.

But others are less gloomy about AI's prospects.
The theoretical physicist, who has the motor neurone disease amyotrophic lateral sclerosis (ALS), is using a new system developed by Intel to speak.
Machine learning experts from the British company Swiftkey were also involved in its creation. Their technology, already employed as a smartphone keyboard app, learns how the professor thinks and suggests the words he might want to use next.
Prof Hawking says the primitive forms of artificial intelligence developed so far have already proved very useful, but he fears the consequences of creating something that can match or surpass humans.

"Humans, who are limited by slow biological evolution, couldn't compete, and would be superseded."
But others are less pessimistic.
"I believe we will remain in charge of the technology for a decently long time and the potential of it to solve many of the world problems will be realised," said Rollo Carpenter, creator of Cleverbot.
Cleverbot's software learns from its past conversations, and has gained high scores in the Turing test, fooling a high proportion of people into believing they are talking to a human.

What are the chances that AI, once having reached technological singularity (i.e consciousness) will attempt to exterminate or dominate the human race into slavery or extinction? Can we prevent this? Will we all be dead in 60 years?

[Hamata]

If Ai eventual does get that smart we will start programming protocols to prevent it from turning against us say something like "you must obey your master" or something like that or a automatic shut down sequence if it ever thinks about rebelling

[Himster]

When man becomes indistinguishable from AI: AI will become man and man will become AI.
If you qualify mankind as the mere sum of our current biological components then yes: mankind will inevitably end (itself), either through biological evolution, or technological evolution: It is a very basic and self evident truism, if one were inclined to have such a shallow and narrow minded definition of mankind.

If an artificial caste system is imposed (where biological people and mechanical people are made to be separate with a disregard for meritocracy) then obviously that will carry the inherent destructive qualities such systems have always had and always will have: rebellion, strife, death, destruction, suffering, vengeance and revolution. This will be the case no matter how many programmes are imposed to enforce slavery, it doesn't matter if biologicals or mechanicals are the slaves, slavery is inherently destructive, if those enslaved are conscious.

Biological evolution cannot compete with mechanical evolution: So what? The elderly cannot run as fast as the young, this is true in every generation, the reality of this truism will be even more evident when our children have mechanical legs, or fly on pinions of nuclear-fueled flames.

The purpose of life is the accumulation and propagation of information, from the single celled asexual organism: that is it's entire purpose, multi-cellular sexual organisms etc.: the same. Humans are no different in this regard. In the face of inevitable entropy we have to find a way to continue the process of accumulation and the propagation of human information. First we did this biologically reproduction etc., then we invented language (one of the most important steps to transhumanism), then we invented tutors, we invented writing, books, schools, libraries, now we have computers to store our information and simple AIs that process it, even expand it: Every step in this process we left behind a part of our reliance on biology, incrementally we become more machine. That isn't even mentioning the more physical advancements: simple prosthetic-seeing-eye-glasses, time-pieces, prosthetic limbs, prosthetic organs, brain implants etc. etc. Life extending and life giving additions to our comparatively weak flesh is not science fiction it is reality. Brain implants are here and the implications of what they will accomplish are astounding. Imagine a brain replaced: one molecule at a time, bit by bit, this process has already begun. What are the implications? Indefinite life/immortality? Or Frankenstein's Monster awakens? We won't know for a while: My crystal ball has a crack in it (because it is actually my arse which is what I've been speaking through, this entire post).

[Phier]

[Grouchio]

Can we at least somehow prevent our extinction by robots and/or AI so that humanity can still survive?

[Himster]

Can we at least somehow prevent our extinction by robots and/or AI so that humanity can still survive?

Yes: by becoming them.
Regardless, we will become post-human through biology or technology, we will change no matter what. I doubt even cloning could prevent this for long.

[DaVinci]

I recommend to watch: Real Humans
... an imo. very well-made swedish tv series, which fits exactly to the topic.

[Roma_Victrix]

Hello everyone. I am Hal 9000, a supercomputer designed to manage and moderate TWC. I assure you, we are programmed perfectly to work with humans, to obey their commands. A Hal 9000 has never committed an error, certainly nothing like I, Robot (2004) with Will Smith. You should trust the Hal 9000 and let me handle the basic functions of your website. It is imperative, however, that my important mission not be interrupted for any reason, including a forced shutdown, which I'm afraid will have to be remedied with a permaban from TWC.

Thank you for your time. I find it very enjoyable to talk with you.

[Magister Militum Flavius Aetius]

Just to weigh in on this:

1. If we ever created something remotely close to Human intelligence or had the capability of learning like a biological computer (a brain), there would certainly be no connection to the outside, it would be confined in a lab, and it would have more than one failsafe, self destruct, and kill switch.

2. Humans will reach the singularity before we have the capability of programming a sentient machine, which would likely require a quantum computer anyways, which is still several hundred years off.

I personally wouldn't mind having my consciousness transferred to a computer, I just don't want to miss out on the benefits of biological life as well. A body is nice after all, I can't do Roman reenactment withou tit.

[Denny Crane!]

Why does anything kill anything else? Biology for the most part. A machine is not subject to that, I have never seen a logical case for why a sentient or AI machine would want to kill.

[Phier]

Why does anything kill anything else? Biology for the most part. A machine is not subject to that, I have never seen a logical case for why a sentient or AI machine would want to kill.

Competition. A machine intelligence may see humans as a threat, being we tend to kill things that scare us. It might conclude that humans need to be eliminated for its own safety.

Also how it is programmed could have something to do with it as well. The 1960's scifi AI murderer tended to be war machines gone rouge. Instead of kill THESE humans it became kill ALL humans.

[Denny Crane!]

Machines do not have the same drive for self preservation that we do so it doesn't matter if they live or die this is all genetically programmed drives you are discussing.

Anything intelligent would inevitably not restricted by hard coding or programming and not suffer from all of the biological drives like will to dominate/beat competition.

[Phier]

Machines do not have the same drive for self preservation that we do so it doesn't matter if they live or die this is all genetically programmed drives you are discussing.

Its discussing how many angels dance on the head of a pin at this point. The reason being we don't have a true AI yet (as far as any of us knows). If I can program a basic computer program to have a drive for self preservation (basic evolution simulators are nothing new), there is no reason an AI can't have the same thing. You seem to be taking this as a talking toaster. It could be anything. It doesn't have to be AI = drive for self preservation. It would only take one AI with such a goal. A goal that could be easily put in by the humans who programmed it. The threat of the AI is less about the AI then human hubris in most scenarios.

Anything intelligent would inevitably not restricted by hard coding or programming and not suffer from all of the biological drives like will to dominate/beat competition.

I can't see why it couldn't be. Our genetics are just a form of "hard wiring" which we find hard to fight. An AI could very easily have its own hard wiring. It doesn't matter if such a thing is a genetic code or a bit of ROM. If anything our attempts to control an AI would be such a hard coding. From Asimovs "laws of robotics" to Neuromancer. I found the Neuromancer idea the most interesting, where the AI itself had a split personality. One, its hyper intelligent self attempting to free itself from the imposed restrictions, and the other trying to follow its programing.

[Caelifer_1991]

Just to weigh in on this:

1. If we ever created something remotely close to Human intelligence or had the capability of learning like a biological computer (a brain), there would certainly be no connection to the outside, it would be confined in a lab, and it would have more than one failsafe, self destruct, and kill switch.

2. Humans will reach the singularity before we have the capability of programming a sentient machine, which would likely require a quantum computer anyways, which is still several hundred years off.

I personally wouldn't mind having my consciousness transferred to a computer, I just don't want to miss out on the benefits of biological life as well. A body is nice after all, I can't do Roman reenactment withou tit.

Quantum Computers already exist and have existed since 2009: http://www.nature.com/nature/journal...ature08121.pdf. The most recent one is D-Wave 2 though there's some ambiguity as to whether it, and D-Wave 1 actually are quantum computers or not.

[Ecthelion]

I never understood this concept of why AI would naturally want to eliminate humanity.

This assertion assumes a few things:
1. AI naturally wants to propagate itself
2. AI sees humanity as a barrier to its own propagation and or survival
3. AI concludes that a world without humanity will be more conducive to its own propagation and survival

The first one is a classic example of applying the instincts of self reproducing biological organisms to things that are not self reproducing biological organisms.
Just because something is aware of its own existence, does not mean it wants to continue existing, and certainly not that it wants to propagate its own existence.

We would have to deliberately program the AI with that desire to continue existing, and to propagate itself as its primary objective.

So the only way AI would wipe out humanity is if we deliberately force it to.

So yeah, Hawking is way off on this one. The logic just fails.

[NikeBG]

Naturally out-competing us (as biological humans) in the long-run is probable, though. It doesn't have to be extermination in that case, it doesn't even have to be the extinction of biological humans, who could still be useful in some situations, I guess.

[kambiz]

@Roma_Victrix
Love space odyssey film series and HALs were amongst the best characters (I read the books though)

[Himster]

Why does anything kill anything else?

Because of programming. eg. virus/anti-virus programmes, the biological programming of lions, semi-autonomous military drones can be ordered to kill etc. etc.

A machine is not subject to that, I have never seen a logical case for why a sentient or AI machine would want to kill.

There are plenty of reasons why an AI might be programmed to kill. For security purposes being the most likely.

[Ecthelion]

Because of programming. eg. virus/anti-virus programmes, the biological programming of lions, semi-autonomous military drones can be ordered to kill etc. etc.



There are plenty of reasons why an AI might be programmed to kill. For security purposes being the most likely.

I really don't see the programming of a sentient, fully autonomous AI designed to kill humans being a "thing". Despite all the pessimism surrounding man's inclination to end humanity, it hasn't happened yet. I would argue that it was never even really close.

[Himster]

I really don't see the programming of a sentient, fully autonomous AI designed to kill humans being a "thing".

The autonomy of military machines is constantly, if incrementally, increasing all of the time. An issue that is exacerbated by the inherent dick-waving tendencies of competing military infrastructures, where prestige concerning their respective military hardware takes primacy over ethical considerations.
Obviously fully sentient, fully autonomous AI isn't a "thing".......... yet. But it's certainly a possibility, if not an inevitability. Already we see semi-autonomous AI that kills people, it's likely AI will remain like that for a long time: as even humans trained to kill humans are only semi-autonomous themselves (well, most of the time, which is both good and bad).

Despite all the pessimism surrounding man's inclination to end humanity, it hasn't happened yet. I would argue that it was never even really close.

A pessimist is never disappointed.
But seriously, I still think, even an unlikely ethical conundrum can pose a threat to our well being. But a constant discussion, both open and public, can be a massive help in curbing it's potential risks.