AI - Destruction or Deliverance?

democrates

Assuming that in the future it becomes possible to create an artificial intelligence which exceeds that of humans, how would this super-intelligence affect us?

democrates

Ok that was too terse, I'll try again.

Roger Clarke talks about laws for robots:
http://www.anu.edu.au/people/Roger.Clarke/SOS/Asimov.html

I can imagine that such laws could be built into an AI, like an operating system resides 'under' an application and restricts what it can do. But if the application is intelligent, it could potentially find a way to circumvent the laws.

Besides, if it is possible to construct an AI without such laws, my bet is sooner or later someone will.

If that ever happens, can we assume that a greater intelligence would be benevolent toward humans?

Our history shows we have potential for good and bad, so perhaps an AI would set about helping us to realise our best potential. Then again, if the AI is motivated by it's own survival could it decide we are in the way, or simply there to serve?

KnowItAll

democrates wrote:

I can imagine that such laws could be built into an AI, like an operating system resides 'under' an application and restricts what it can do. But if the application is intelligent, it could potentially find a way to circumvent the laws.

I think animals and especially humans have brains that are pre programmed more than we think before we enter the world. Ok, obviously the brain needs to know how to breath etc but it's also programmed to survive.

The AI would not curcumvent the laws if the laws were thought out properly eg if there was a law against the AI even "thinking" about it's own laws then it could not change it's laws.

democrates wrote:

Besides, if it is possible to construct an AI without such laws, my bet is sooner or later someone will.

Just like the people who send internet viruses there will always be people who'd do it because they can.

democrates wrote:

If that ever happens, can we assume that a greater intelligence would be benevolent toward humans?

Thats hard to say. If it becomes greedy then it probaly will but I think they'd have to be programmed to become benevolent to humans or they wouldn't.

The main thing with AI is that if it's not a robot or not connected to factories etc (a way of building robots/weapons) via some form of internet than it could not harm humans.

The main value of AI being smarter than humans is that there may be no need for scientists anymore as the AI would find new discoveries/inventions.

democrates

The AI would not curcumvent the laws if the laws were thought out properly eg if there was a law against the AI even "thinking" about it's own laws then it could not change it's laws.

I guess it comes down to the implementation. I won't use my ignorance about how to restrict an AI from certain thoughts as an argument that it can't be done.

The main thing with AI is that if it's not a robot or not connected to factories etc (a way of building robots/weapons) via some form of internet than it could not harm humans.

Now you're scaring me. If it was really intelligent it could do serious damage over the internet, wrecking systems or masquerading as Bin Laden and organising terrorist attacks or something.

The main value of AI being smarter than humans is that there may be no need for scientists anymore as the AI would find new discoveries/inventions.

Hopefully some scientists would take the job of keeping the AI under control.

I guess I ran into a dead end in trying to imagine what a smarter mind would think and do, but given that humans are the smartest and yet the most destructive of known lifeforms, I'd personally be very cautious about AI.

clearz

OK If we ever built an AI more intellegent than us the results would more than likely be disastrous. The only way we could control something so powerfull would be through direct neural interface where the computer would become an extention of our brain enabling us to do higher level tought, but on its on would be useless.

If on the other hand we let an AI lose that had greater intelligence than us. What would it do? More than likely it would replicate itself with improvements. After a few generations(maybe far shorter than a human gereration) There would be AI's with so high of intelligence that we would no longer matter as they go about doing their busines. We would be as insignificant to them as the bugs we kill ourselves while building a house or walking the streets. They might wipe us off the planet without even meaning it.

EamonnKeane

I can imagine that such laws could be built into an AI, like an operating system resides 'under' an application and restricts what it can do. But if the application is intelligent, it could potentially find a way to circumvent the laws.

Why would it? A machine does not desire like a human does, or even like a worm does. A thing of metal and silicon could only create an illusion of conciousness, surely it couldn't be concious itself?

If that ever happens, can we assume that a greater intelligence would be benevolent toward humans?

Our history shows we have potential for good and bad, so perhaps an AI would set about helping us to realise our best potential.

How? By doing everythiung for us? The only way to get someone to realise their potential is to leave them alone.