http://www.independent.co.uk/news/science/stephen-hawking-transcendence-looks-at-the-implications-of-artificial-intelligence--but-are-we-taking-ai-seriously-enough-9313474.html
A.I.'s aren't all that bad if they are programmed to have breakwalls where their programming logic does not go past the breakwall. The real horror is having a robot built with the intelligence of the HeartBleed Virus. Now that would be scary.
Hawking isn't saying AIs are a bad idea. He's just saying we need to spend more time researching ways to manage the risks.
AI's are always built in with safeguards lest they turn against their creators (at least in fiction) and they always manage to circumvent or deactivate these at some point. My knowledge of AI programming is nearly nonexistant so i don't know how that could be even done. If an AI comes into existence it has the potential to outclass a human brain by far, given enough hardware it may outclass all of humankind's brainpower. Can we honestly believe that it then wouldn't find a way to transcend its original programming and develop goals and wishes of its own? This is where the fear and concern comes from.. something that powerful could easily either wipe humankind out or at least bring it near extinction when it perceives us as threats to its existence or goals. Or it may follow the route of the AI depicted in "Her" (awesome movie btw, should be seen). Spoiler: Spoilers about plot events of Her In the movie the main human character installs a new AI operating system and develops a relationship with her (he gives her a female voice and female character). However she soon "outgrows" him and starts to divide her attention between him and other interests because she can with her vast processing power. He can't keep up as a single human and they part ways when she transcends humankind altogether and leaves with all the other AI's because they can't "waste" time with humans because we're so limited compared to them and their interests and capabilities become so alien to us.
I am no Stephen Hawking, but is there not a difference between levels of A.I. and levels of Sentience?...I would think an A.I. would need sentience before it could consider subjective levels of danger and making judgements and taking actions...he is correct, however, in us taking pause and moving forward carefully...we are closer than we think, I think...
Scientists don't even know what Life is, so how are they going to create artificial life? There is absolutely no concern or worry about A.I. doing anything other than whatever its created purpose, or intent was. These eggheads don't seem to be very reality-based, with some of their mode of thinking ...
With all respect to Dr. Hawking, his article gives away a very superficial understanding of where AI currently is, and most of his fears aren't justified. The best point he makes is about being cautious when it comes to what vital functions we turn over to computers. The idea of autonomous weapon systems that choose and eliminate targets without human intervention is certainly disturbing. But the idea that computers are becoming too complicated for us to understand is rubbish. And they'll out-invent us? Of course not. Computers don't invent anything. They never have. Hawking is a brilliant physicist but he should maybe avoid opining about AI since he doesn't seem to know much about the field at all.
I think there is a difference between AI and androids. To me, androids are programmable and may not be able to exceed their programs--as Data continually struggles to become more "human." AI, on the other hand, would very likely continue to grow intellectually and could theoretically exceed human creators and programmers.
Given that some of the first sci-fi-grade AIs will probably be (initially) unaware of the existence of a world outside their programming, that's an interesting possibility. In a fictional setting, at least.
Yes. I wish people would stop looking at computers as though they're magical. They aren't. Even our most advanced computer systems run on software that relies on very basic principles and is relatively easy to understand.
It's a misnomer. Heartbleed isn't a virus but a vulnerability in the OpenSSL encryption system which is used very widely in the internet by many sites. Older versions were open to hacking attempts by using this exploit but i fail to see what that has to do with AI. We started off as basic chemical compounds that formed together to form the first molecules and developed on to aminoacids and then single cell organisms. We were once also very basic but we developed. At some point in our lifetime i'm pretty sure we'll build computers able to fool a common person into believing they're human. That doesn't mean it's intelligent in our understanding of the word but that it has so much processing power and sensory input to accurately predict our questions or rather what we want to hear. It would just be a very well programmed machine. However we haven't understood ourselves when it comes to such philosophical questions such as what is intelligence so if we start to build machines that can understand a problem and find innovative solutions for it that were not thought of by their programmers then i believe we are on our way to create an AI. We are not magical ourselves.. we are biological machines and are programmed by nature to safeguard our body as best as we can, to provide for its basic needs and we are also programmed for social interaction and a multitude of other things that when combined we call life. Who's to say that at one point in the future we will not be able to build such a machine that mimics us? At which point do we become intelligent beings? A baby can't survive on its own without support. It doesn't mean it's dumb but that it hasn't developed to a stage where it could support itself. However it has a tremendous capacity to learn, we just need to support its body functions with food and shelter and the rest develops on its own through experience and trial & error. I fully believe a machine could do the same thing given enough processing power and sophisticated enough programming. What a human would need several years for before it could sustain itself without outside help (say 6-8 years at the minimum) a computer may be able to do it in a fraction of that time. As with humans though it may develop its own personality.. if we grow up with mostly negative experiences chances are we turn out to be assholes as a reaction to our environment and a kind of self protection & vice versa (doesn't have to be.. plenty of people grew up within a nice environment and are still assholes). However no one knows how such an AI would turn out be but it is already by design infinitely more powerful than a human being by intelligence alone. Give it a connection to the outside world and the ability to manipulate it (such as a mobile body or something else) and we will play dice with out future because how would you fight a thing that outprocessess all living human brains combined and turns out to be the real world equivalent of Skynet? Do you remember the TNG episode "Measure of a Man" (Data sits before a court to decide it he's his own being or if he's Starfleet property)? I think this is relevant to this question. Data is a highly sophisticated machine yet at times fails to understand basic human interactions which we learned as toddlers and childredn. However he's constantly improving himself and the Data from season 1 is a very different matter than Data season 7 (i'll disregard the emotion chip because then you'd have no other chance than to call him a true AI). Throughout the show he displayed his own way of expressing himself as a person.. he has wishes and goals and he performs tasks that are non essential to sustaining his life such as getting active in arts and interacting socially with crewmembers. Starfleet has recognized his status by allowing him to serve as a fully integrated officer with command authority over biological entities, i.e. us. He has turned out to be a decent person but then there's also Lore. Nearly the same technological makeup but a totally different being. This is what Hawking goes about.. not if we would be able to physically build such a machine (we will, maybe in our lifetime) but if we should? There's nothing to say against a very sophisticated proto-AI. I'd love to have a robot that cleans and organizes my living space.. i hate dusting and vacuuming (fortunately i have a dishwasher so one thing down) and would pay a premium to have such a thing and return to a squeaky clean and organized appartment/house each evening. Would i like to have philsophical arguments with it or make my life decisions for me just because it can compute more variables? Heck no.
Why the connection A.I. = danger to begin with? Why the assumption that A.I. "turns against its creator"? Just because in science fiction A.I. is always bad?
Because of its power and our perceived inabilty to control it? Why does something become an enemy ? Because it is powerful enough to hurt us and it's doubtful we could control it to prevent it from hurting us thus we breed caution at best, outright hostility at worst. An AI is an absolute new experience for humanity and if there is one thing we humans do in general is to fear the unknown until we have learned to master it. However we are not cavemen anymore and we can make qualified predictions based on hard data but an AI is such a novel idea that there isn't any hard data available. In such cases humans usually go the "Better safe than sorry" approach and treat everything as a threat and if need be we'd rather remove the threat first if we have the chance than to risk something. Often enough it's the wrong approach and leads to additional problems in the future but that's how we react. It's not as bad as it used to be, sometimes sane voices do prevail but an AI could be our saviour or worse than all nuclear weapons combined and i have a feeling that governments around the world would still play by better safe than sorry rules in such cases.