Scientists Fear Killer Robot Reality

Terry Olaes

I Used to be the [H] News Guy
2FA
Joined
Nov 27, 2006
Messages
4,646
I know we can’t resist a Skynet reference whenever we hear about advanced robot AI and whatnot but this article in the Times Online from the UK takes all the fun out of it. Why? Because they agree with the jokes and are worried about robots and AI becoming self-aware.

Advances in artificial intelligence promise many benefits, but scientists are privately so worried they may be creating machines which end up outsmarting — and perhaps even endangering — humans that they held a secret meeting to discuss limiting their research.
 
I wonder if they, at the meeting, watched every movie where scientists inadvertently destroyed the earth with robots to note what went wrong and how to avoid it.

First thing I'd do is lock robots out of was anything to do with computerized weapons systems larger than an AA turret. No ICBMs or nukes for joo!
 
All they have to do is put in the big red button. A region of code inaccessible to the AI itself that causes the program to completely deactivate when receiving a special signal of some sort. Oh, and don't let the robots reproduce themselves, so they can't remove the big red button code from future generations.

Some say it is impossible to make a region of code completely inaccessible, but I disagree. Humans have these regions; I can't make my heart stop beating, for example.
 
Well, arguably, if the artificial life becomes advanced enough, they will be successor to humanity. In this sense, even if humanity may cease to exist, our creation will inherit this Earth and keep advancing.

Maybe not so great, but not totally horrible other than that there'd be no human left on Earth, just as the old saying of "no people, no problem" would apply here.
 
Someone entertain me please by answering this, what exactly are the benefits of having a true computer AI?
 
One more reason why we need to aim for transhumanism. Humans can't get wiped out by AI if they are the AI :D
 
Someone entertain me please by answering this, what exactly are the benefits of having a true computer AI?

Well part of the concern I would imgine is that this AI is very intellgient in some way and we get benefit from leverging this "Super" intelligence. Could be use to solve problems in medicine, computers, whatever.

Brain power is a valuable comodity.
 
Well part of the concern I would imgine is that this AI is very intellgient in some way and we get benefit from leverging this "Super" intelligence. Could be use to solve problems in medicine, computers, whatever.

Brain power is a valuable comodity.

In summary: too many humans are dumb :p
 
I don't see it as being a bad thing actually, since it seems Darwin has been on one helluva long vacation. Population control. It's either that or Soylent Green...
 
they also fear going outside, answering the door, and talking to women.
 
It's inevitable imo.. Just like cloning or stem cell research. Even if there was some "consensus" to avoid certain research it would still be done "secretly" by some one some where. I-Robot here we come
 
Couldn't true AI be used for rapid translation? Like teaching it multiple languages at the fastest speed it can learn?

Then again hearing "DEY TUK MAI JAWWWB!!!" in like 9 different languages probably wouldn't be all that funny.
 
Someone entertain me please by answering this, what exactly are the benefits of having a true computer AI?

I see what you're saying Sir, having robots do all work is the stupidest idea ever IMO. The unemployment rate would skyrocket to 75% if they took over our jobs.
 
I see what you're saying Sir, having robots do all work is the stupidest idea ever IMO. The unemployment rate would skyrocket to 75% if they took over our jobs.
No you're missing what I'm saying, I'm not saying robots are bad, hell even some sort of pseudo-AI (i.e. some crash-avoidance algorithm) isn't bad, but what good would it be for a robot (or computer) to actually think for itself.

The only feasibly one I saw was some super intelligence that can invent shit or what not... well if that's the only reason we might as well just kill ourselves right now if thinking has become too difficult. It's one thing to build some program that folds proteins or uses a brute force method to evolve some item to be "the best", it's another thing to have a computer invent something just because....

I still don't get why an AI would be needed/that useful.
 
No you're missing what I'm saying, I'm not saying robots are bad, hell even some sort of pseudo-AI (i.e. some crash-avoidance algorithm) isn't bad, but what good would it be for a robot (or computer) to actually think for itself.

The only feasibly one I saw was some super intelligence that can invent shit or what not... well if that's the only reason we might as well just kill ourselves right now if thinking has become too difficult. It's one thing to build some program that folds proteins or uses a brute force method to evolve some item to be "the best", it's another thing to have a computer invent something just because....

I still don't get why an AI would be needed/that useful.

I agree with you. Even if we have the ability to replace ourselves with something more intelligent, would we really want to do that? Right now, humans use robots to do the jobs we don't like (boring, repetitive, non-thinking tasks) with those we do like (jobs that stimulate our minds and allow us to use high-level thinking to solve abstract problems). If we turn over the high-level jobs to robots as well, what are we going to do?
 
HAHAHAHA!!! You have GOT to be kidding me? No offense, but I DOUBLE DOG DARE them to create a machine that can become "self aware"!!! Wait a second, let me give you a list of things that science and technology, in all of it's vast knowledge and advances, has not yet been able to accomplish.

1. CREATE a human limb, organ, much less anything resembling life
2. Accurately predict the weather, much less predict global warming (hahahahahaha) in the distant future
3. Remain consistent with itself, every year there is a new study that contradicts a previous or concurrent study.

And you honestly think we are capable of creating a computer SOOOO smart that it will become self aware??? Oh please... I think they have been watching too many movies, or become too full of themselves. I am a programmer myself, and I can tell you straight up that such a thing is utterly impossible. Even using advanced systems (that don't really exist) like "DNA Computers" or "Neuro Computers" it remains impossible. That is part of what makes life so unique and special in the universe. Go look up a man named Warner Gitt, wrote a very interesting book called "In the beginning was information".... I think you will find it quite interesting. It definitely touches on this subject.
 
I agree with you. Even if we have the ability to replace ourselves with something more intelligent, would we really want to do that? Right now, humans use robots to do the jobs we don't like (boring, repetitive, non-thinking tasks) with those we do like (jobs that stimulate our minds and allow us to use high-level thinking to solve abstract problems). If we turn over the high-level jobs to robots as well, what are we going to do?

Look up 'Resource-based society'. Think of Star Trek; with the invention of the replicator in that world (not unfeasible for us either) all material needs would be instantly satisfied as virtually anything one could want, from food to gadgets to jewelry, could be instantly replicated, for almost free. Kind of like the internet and digital media :)

If there'd be no need for humans to bother with something as mundane as 'jobs', we'd suddenly have all the time in the world to create a Star Trek-like society. Heck, we'd pretty much be forced to do so. We could focus on creative pursuits, on science, on just lazying around. Economical theory would have been thrown out of the window, so life would be pretty much free.

Sounds good, doesn't it? :)
 
After having read the article I clicked on the Nikola Tesla link and read the Times article about him from when he visited London. Having an entire room electrified through the air so that light bulbs can glow without touching anything sounds amazing considering that Tesla himself was unaffected by the the it. Why don't we have that stuff today? In some ways that over a 100 years old article seemed more sci-fi than the one about AI.
 
They could always spend more time teaching other humans GI.. Genuine Intelligence; it seems to be a rare commodity these days.
 
We at least need to invent robots that can govern the counties/world, even the simplest AI would do a pretty dang good job :)
 
Look up 'Resource-based society'. Think of Star Trek; with the invention of the replicator in that world (not unfeasible for us either) all material needs would be instantly satisfied as virtually anything one could want, from food to gadgets to jewelry, could be instantly replicated, for almost free. Kind of like the internet and digital media :)

If there'd be no need for humans to bother with something as mundane as 'jobs', we'd suddenly have all the time in the world to create a Star Trek-like society. Heck, we'd pretty much be forced to do so. We could focus on creative pursuits, on science, on just lazying around. Economical theory would have been thrown out of the window, so life would be pretty much free.

Sounds good, doesn't it? :)

Well, I guess that would be great for the people born and raised in such a society. But I honestly hope I never live to see it. You can't experience success without experiencing failure.
 
Well, I guess that would be great for the people born and raised in such a society. But I honestly hope I never live to see it. You can't experience success without experiencing failure.

You mean you'll experience more chances to learn from failure and learn in general if you're stuck in some kind of dead-end job?
 
Ah good times with fiction
Ghost in the Shell (series) by Masamune Shirow (those Japanese are just weird)
The Matrix by the Wachowski Brothers
iRobot by that crazy Russian Isaac Asimov (All hail the Asimov rules of law)
Who can forget Master Mold what a nasty robot supervillian!
Then there's always Arnie!
 
We will reach a “singularity” point with robotics and AI in the not too distant future, in our lifetime. Its just a fact and something we will have to deal with. You cant outlaw it, you cant stop it, it will happen.

In our lifetimes, we will share this world with another intelligent species that will be greater in strength and thought than our own. That is what truly horrifies us. We drop from the top of the mental chain to the 2nd place on planet Earth.

After that happens, there is absolutely no way to predict the outcome. It could be no more than the mundane to the extraordinary. We too may benefit in vast and unimaginable ways.. or we could become extinct at the hands of superiority.
 
Bah, Anyone that works with robots, programming and AI knows that any “singularity” is FAR off . . very far off (but quite possible)
And even then, it will not be by accident and hardly spawn a all powerful overlord at first go :eek:
Also in almost any scifi dooms-day scenario it is always the idiot humans that shoot themselves in foot with a bazooka! by giving the AI/robot unlimited access/resources/capability . . now we know that is unlikly . . we may be idiots but we are also always cheep. :p (moral do got give total control of all nukes to ANY single entity regardless of what/who they are.)

For example, why give a android/robot/AI any more capability then what is minimality needed? IE a android/robot maid with super strength, unlimited running time and a full and expandable AI? it just costs too much, and often technically not possible or wildly hard. who wants a maid the size of a truck?

Military units may need some of that extra bits and may have the $$ but then again with great power comes massive power use, cost and maintenance. An entire support system is needed. A good symbiotic man/robot would work great here.

So, the random whacked out robot/AI really can't be more then a nuisance unless we really go out of our way to insure it could. Also it is TOO easy to make mutable and simple safeguards requiring a human element. and everything including robots etc need resources to survive so there are all sorts of ways to "pull the pug"

Scifi loves working on peoples fears and making the most unlikely scenarios possible (as it should :) ) but as we find, reality keeps surprising us with stranger things.
IE if skynet was so smart why build terminators?! why not just build explosive robo rats/spiders/flys? or some bio defoliant? flesh eating nanites or all of the above?. . the humans would be screwed with no way to fight back. The fixation on a humanoid form was goof scifi but moronic for the AI.

Building a Golum/Frankenstein's monster/robot that would "go a muck" has been a fear since the concept was thought of. . . but that is also the basis of the fear of any and all tech/man made items and anything unknown and new . . .
We should always be wary but should get over the fear . . . since stagnation is the most assured way for extinction and adaptation and development MAY (nothing is assured) help us grow.
 
You mean you'll experience more chances to learn from failure and learn in general if you're stuck in some kind of dead-end job?

Well no, but you seem to be talking about a Brave New World society. Perhaps I misunderstand?

Regardless, this doesn't address my initial issue. We don't need AI to create a replicator.
 
All they have to do is put in the big red button. A region of code inaccessible to the AI itself that causes the program to completely deactivate when receiving a special signal of some sort. Oh, and don't let the robots reproduce themselves, so they can't remove the big red button code from future generations.

Some say it is impossible to make a region of code completely inaccessible, but I disagree. Humans have these regions; I can't make my heart stop beating, for example.

Actually you can slow it down a lot or rise it up, with meditation.
 
I hope people realize that as technology advances so will it's aid to humans. We don't just stay as we are as technology changes. We evolve with it. As true AI becomes a reality, so will electronic augmentation or genetic manipulation.
Informational learning (memorizing) in the future will be a thing of the past. Some chip in your head will probably just feed you the info. Advanced calculations? Pfft. Who needs a calculator, when your brain can calculate like the rainman.
Who knows, maybe in the future you will be come an AI in a tincan. Live to 50 and tranfer. There's one thing that I doubt even AI can make, and that's the shear variety of thought processes. We will be forced to evolve, be it through augmentation or genetic manipulation.
 
There is a HUGE difference between an intelligent system and a smart system. Technology today allows us to create smart systems: systems that can simulate millions to trillions of situations in a domain such as chess, speech recognition, or medical databases. Those systems draw on past data or controlled environments and use statistics to create probable situations. This is were computers are smarter than us because they can store more data, process calculations, and simulate more situations than any human can do.

We lack the hardware and the software to create any type of intelligent system. Can a system that beat the best chess players in the world learn to play checkers?? Can a system that can translate books in multiple languages draw conclusions about the character or themes of the story??? Can a medical system create new treatments for existing illnesses??? The answer is no, computers have always been limited to the domain they have programmed to do. The only way a system can learn or apply out of its scope is to be reprogrammed. For AI, That means that programmers would have to manually connect every domain with every other domain, which is infeasible on a storage and on a processing level.

The closest theory I know of cross connecting arbitrary situations is Advanced Neural Networks. This is where a system can weigh multiple inputs of different natures to a single deciding output. For example see Netflix's million dollar competition and check out the winning team, they combine hundreds of algorithms to weigh whether a user would like a movie or not. Those calculations take days, and again thats only in the domain of movie rental ratings. Orders of Magnitude isnt enough to express how much larger connecting everything seen in day to day situations, not to mention there is no basis evaluation for ANN of such scope. Evolutionary steps in hardware and software wont be enough to handle this. In my opinion it'd have to be something revolutionary in the computing world.

Finally, code can not create new code, only generators exist. When you start a new project in your IDE of choice it can create predefined files with your settings, and from there you have to program. It cannot develop new software that has never been created. Back to the chess example, it doesn't create strategies, it uses statistics to choose moves that give it the most likely path to winning. Having computers going haywire and not doing what its intended to be doing is one thing, having computers reprogram itself and become "self aware" is a long way away.
 
Back
Top