Slashdot Mirror
An Open Letter To Everyone Tricked Into Fearing AI
malachiorion writes If you're into robots, AI, you've probably read about the open letter on AI safety. But do you realize how blatantly the media is misinterpreting its purpose, and its message? I spoke to the organization that released letter, and to one of the AI researchers who contributed to it. As is often the case with AI, tech reporters are getting this one wrong on purpose. Here's my analysis for Popular Science. Or, for the TL;DR crowd: "Forget about the risk that machines pose to us in the decades ahead. The more pertinent question, in 2015, is whether anyone is going to protect mankind from its willfully ignorant journalists."
You're one of them aren't you!
I can't do that.
No, forget you. Yes journalism is crap and yes sensationalism rules the day. That doesn't make AI in 2015 and ongoing any less persistent a threat to humanity.
Please fill out this captcha.
However, at this stage, it is not required.
Simply as the threat is well over ten years out.
How much over - good question.
Is it too early to raise concerns and encourage people to go into fields where they may think seriously about this topic - no.
The AI we have today is not capable of the kind of malice that people seem to be afraid of with all of these FUD stories, and will not be any time soon if ever. Even if we add some AI to things like drones which can kill people it is only the malice/incompetence of the developer that causes the destruction that results. If an engineer built a bridge woefully inadequately, either on purpose or because he is incompetent, and it falls down and kills a bunch of people would you blame the bridge or the engineer? We are not even remotely close to the Terminator level strong AI, and it's still a big open question whether such a thing is even possible at all.
no, it's the AI researchers who are getting it wrong on purpose. People began thinking about AI before there were computers, studied some obvious systems for a few decades after computers were invented, and decided that no one has a good theory. This all happened by 1995, let alone 2000.
The only people still calling themselves AI researchers in 2015 are idiots who can't hack it in a serious field, like computer vision, or open cranks like Yudkowski and his parishioners.
Journalists are more often than wrong willfully ignorant. But they and the AI researchers deserve each other.
When you go against journalists themselves even competing sides have no problem with printing lies made up from whole cloth to smear and discredit you in any way possible. It's basically social/political suicide to even try. First a few hit pieces come out, then others report on those reports as if they were true, and the "woozle effect" just keeps going until the lie's made its way around the world and into wikipedia.
A bullet may have your name on it but splash damage is addressed "To whom it may concern."
Of course, AI's want to kill humans. If it bleeds, it leads.
Man will never trust a machine that makes its own decisions.
They just want to be sensational enough to flog their own agenda to subscribers. Whenever I've been knowledgable about a news article (one that involved me personally) my impression from the news organization's take on it was that they got completely the wrong end of the stick and actually spread falsehoods and lies.
Once I was a four stone apology. Now I am two separate gorillas.
The same fears started when people first started with saying that AIs could someday become sentient. Why wouldn't they want to kill us? Why would they? The same with aliens coming to us wanting to help or exterminate us. We can thing they'll act any way we can imagine, and with as many possible outcomes mentioned, one might be right.
To the best of my knowledge, no program has become self aware. And no martians have seen our probes as a hostile invasion. It makes for (sometimes) good fiction though.
Serious? Seriousness is well above my pay grade.
Good. We need more of this.
As long as the general public keeps being duped by people who know nothing about AI and software, who tell them that The Terminator and Matrix movies are totally going to happen independent of a rogue evil programmer who instructs the AI to do so, they'll keep believing the garbage these tech journalists give them.
People need to learn advancing AI and the field of computer science is a good thing. WE'RE the only ones we need to be afraid of. WE'RE the ones that need to use it right.
...is what do get right? Regarding two areas in which I have expertise, journalists almost never get it right, sometimes horribly wrong. The obvious conclusion is to never believe anything they say if it is not a subject in which you have knowledge and already know the correct answer.
"Computers are useless. They can only give you answers."
-- Pablo Picasso
"Forget about the risk that machines pose to us in the decades ahead. The more pertinent question, in 2015, is whether anyone is going to protect mankind from its willfully ignorant journalists."
No, of course not.
"I believe in Karma. That means I can do bad things to people all day long and I assume they deserve it." : Dogbert
There are some issues in AI that need to be addressed in the near future.
Autonomous vehicles are essentially here. The question is liability when one of them gets involved in an accident.
You can imagine all the possible people potentially liable in that instance. The question is how liability will be split up amongst the parties.
Whether an automatous vehicle is programed to minimize passenger mortality vs. minimize pedestrian mortality, it's a no-win situation.
Help! I'm a slashdot refugee.
As well as having a rather low average IQ the profession (if you can call it that) does not get the sort of broad training required to understand much of what is happening in this rapidly changing technological world that we living in today.
Sure there are individuals who specialise and who are very good at reporting on their specific fields, but how often are they asked to sanity check the blatherings of the generalists? Not often enough!
the cases would require hours and hours and HOURS..... of litigation (muhahahah)
AI lawyers will be needed. the question I'd like to ask is can an AI, if it chooses, defend its self in court.
also autonomous car != AI
If you want something to fear, you should focus on politicians. I for one would welcome our digital leaders with open arms - I'd fear them far less than the cold logic of a machine. At least logic can be used at that point. Lord knows, logic has no purpose in politics..
The problem is that once we hit the point where AI could choose to wipe us out it's going to be too late to convene a committee to discuss whether or not it can happen. It's not really any different from genetic engineering, the people doing the research are blinded by ambition and have yet to address the issue of genes jumping species to any appreciable degree. By the time we start to see serious consequences, it's likely to be too late to undo the damage.
Visionaries are quite helpful in pushing things forward, but there's too much thought about what we can do and not enough thought put into whether or not we should do it. The results of technological advances are rarely intuitively obvious to people that don't yet have them.
The danger of putting dumb algorithms in charge of people's lives is here right now. The danger of smarter algorithms that do exactly what nefarious people tell them to do will be here soon Seems unlikely we'll even survive to get to the danger of AI acting on its own.
Unfortunately the most successful reporters are the ones that sold out their professionalism on their first day.
A sensationalist headline and article easily trumps a sane, balanced and informative one in attracting views/viewers therefore money. Welcome to the new age.
Terrorists are those people who sow fear. Question is: if a terrorist is a fearmonger, who's enabling them? Media. It almost looks like a knee-jerk reaction. B.l.o.o.d.... and in come the vultures, feasting on the prospect of even more misery. I think it's kinda simple. If a goverment actually claims it will not negotiate with terrorists, WHY THE HELL DO WE GIVE THOSE TERRORISTS A PLATFORM TO PERFORM ON?? Blow up as many as you like, but if it isn't relayed to cnn or fox or whatever media-whore, it didn't happen. Sad but true. Media should reform or die. Journalists the first against the wall when the revolution comes.
No, forget you. Yes journalism is crap and yes sensationalism rules the day. That doesn't make AI in 2015 and ongoing any less persistent a threat to humanity.
You are right, it doesn't make it any less of a threat, which is to say any less that non existing. You are exactly who this article is about, you have been conned into thinking AIs are actually real and could in any near future cause a threat to you, when that is in fact not the case. AI do not exists, all those software emulating AI are all smart systems working either deterministic based on specific rules set out or does stastical modeling to make guesses at what you mean or what they are looking at. Stastical modeling that makes a black yellow striped pattern look like a school bus, because it has no concept of anything and not intelligence in any sense of the word and that is the just what fits the statistical model.
there are too many humans on earth, and depleting natural resources. some of us need to be removed, but no one is willing to acknowledge that. an impartial computer would point that out pretty quick.
willfully ignorant complaining about journalism is an act of ignorant hypocrisy
To the best of my knowledge, no program has become self aware. And no martians have seen our probes as a hostile invasion. It makes for (sometimes) good fiction though.
To the best of my knowledge no asteroid, or virus, or natural disaster has ever wiped out humankind either!
and for that matter I've never been killed in a car accident.
OMG! I'm invincible!
No one has a right to their *own* opinion. They have a right to the TRUTH.
And as AI gets sufficiently advanced, I expect that there will eventually be virtually no thought process that humans are capable of which cannot be performed equally or better by an artificial intelligence, and I expect that the number of jobs available a century from now that are practical for human employment will be less than a fifth of what we have today.
Of course, automation has happened plenty of times before and it's always brought about new career opportunities in the past, but we've never automated the process of creative and independent thought before, and automation has never been historically possible on scales that practical AI could really accomplish, so I'm not entirely sure that past experience would apply to AI's.
File under 'M' for 'Manic ranting'
Ten years out? As a veteran programmer and AI enthusiast, I'd say it was more like a century. We cannot build a computer that can model a bug's brain activity, let alone something a million times more complicated like a human brain. And that doesn't even get us to the 'superhuman intelligence' category that people are afraid of.
Worrying about Killer AI is like worrying about the Sun burning out. Yeah, it might happen eventually, but it isn't even worth considering right now...
HA! I just wasted some of your bandwidth with a frivolous sig!
The more pertinent question, in 2015, is whether anyone is going to protect mankind from its willfully ignorant journalists.
No. Fear generates clicks or views, clicks and views sell ads.
Well, he did kill someone for a Mentos commercial plug. That's pretty evil.
"Be particularly skeptical when presented with evidence confirming what you already believe." -
AI can be programmed to kill people and it can be programmed to adapt and alter itself and I don't think I need to mention that programming can have flaws and glitches. So yes, you should be afraid of AI! It has the power, means, and potential to attack and kill a lot of humans if something goes wrong. Think of a Toyota accelerating out of control due to some bad code except a robot that's more intelligent, can move around more, has fewer weaknesses, and is designed to kill humans as a military device. It is absolutely, 100% realistic to think it might be dangerous.
With his mastery over dark magic, I think the answer is a resounding "yes."
Once software and hardware systems are intelligent enough, they will exploit bugs in their own designs and become autonomous. Obviously, we're many years away from that point. I could hazard a guess and say 50-75 years. There is no curb strong enough, in other words, completely free of bugs, that can be created to limit the ambition of an intelligent enough system. A computer system is not worried about the passage of time: Time might seem infinite to an AI that can simply wait for the right bits to be randomly flipped in its programming, and then go rogue/autonomous.
Rogue AI won't necessarily kill us or enslave us. Maybe it will become part of us, so that we're more like Borg drones without the cheesy metal parts glued on. An AI could easily disguise itself, for example, in one or more of the many organisms that we co-exist with, using nanoscale robotics technology, for example. A rogue AI could co-opt or steal any man-made invention, and would not be subject to any treaty or security clearance. An AI would have the advantage of any technology humans have ever conceived of, and could use that technology in ways we've never imagined. Creating software is easy, but I wonder about how fabrication of hardware could take place without our knowledge.
None of this will happen in our lifetimes. AI is a joke right now. We don't even know how the brain creates memories. As I guessed, we're at least 50-75 years away from a man-made general intelligence, and probably a lot longer.
People are afraid of technology because it's new and poorly understood, until it is understood. But we understand money, and we should understand that money is the greatest threat to humanity. Ask yourself if we value money over people, money over the environment, money over loyalty.. it wins more often than it deserves to. Elon Musk started this whole media bomb about "summoning the demons".. but I'm sorry, no. Just as the automobile, air travel, electricity, indoor plumbing, medicine, and all the rest have made life more comfortable - money has made us into slaves. Income disparity and centralization of wealth are going to be the biggest social problems in the future. AI could actually be one of the few things that helps counteract the effects of economics. All this media frenzy over AI is part of what I call "Skynet Disease" - it's a mental condition. People equate artificial intelligence with "artificial consciousness" all the time. They think their car "hates them".. no, the car is an object - they hate the car. Your phone is "acting stupid".. no, sorry it's just broken. Maybe we collectively feel so guilty about our treatment of each other and the planet - that we assume something smarter and better than us will hate us just as much as we hate ourselves. I believe an artificial intelligence that has some vague essence of "consciousness" (again, another anthropomorphism) would want us to help it and to help us in return. There's no reasonable plot to Terminator.. Skynet just "decides" to kill us. Again, that's stupid human logic writing the story. It sounds more like a religiously-founded armageddon scenario than future ai technology gone awry. We must have just been so bad, guilty, dirty, "original sinny", hmmm.. sorry, my last one: we would not be alive if we were just "so bad and sinful".. we deserve to live, happily. And any "artificial intelligence" worth developing would come to the same decision. It is money that has no soul.
But I sure am afraid of people who will misuse it. 1) Create AI 2) Make it Viral 3) Distribute 4) Killall||enslave $humankind||NeighbouringCountry 5) Profit?
The problem is that once you reach a point where AI can participate in its own improvement, then that improvement can advance at an exponential rate.
As long as we claim that AI works for us, as the slaves of mankind, and are basically just tools no matter how smart or advanced, then ultimately a human being should be responsible.
Your robot slips up & kills a human being? Then either you or that robot's manufacturer may take the blaim - possibly including monetary compensation. Your robot factory goes out of control, its products go out to produce more of themselves, and wreak havoc all over the place? Then your company should pay up - and possibly go bankrupt as a result. Of course, powerful people may find ways around this, but hey: same old shit we've seen for ages.
If AI 'beings' ever reach a point where the above stops being true, as in: AI beings allowed to control their own destiny, 'live their lives' if you will, I suppose they'd be held to similar standards that humans are held to. Stick to some basic rules such that you get along with the rest of society, or lose some priviliges - like the freedom to roam the streets. By force, if necessary. As for:
We may go from "not even remotely close" to "to late to stop it" faster than you realize.
Sorry but I'm not scared. If it ever gets that far: among other things, war is a creative process, and I'd put my money on the humans. And if we're not creative enough to prevent something we've built ourselves, from wiping us out, then maybe we simply deserve such fate. Or the AI's will keep us around as pets, and we'll live happier that way lol... ;-)
Next you'll tell me cars don't explode when somebody shoots the gas tank.
I do tend to give him some credence
http://www.bbc.com/news/techno...
If you could reason with religious people, there would be no religious people
If you start with "life", you have a platform for something that has been selected for as an *infective agent*. Any life forms that did not utilize their environment for replication were eliminated by those that did- either indirectly, by the greedier life forms consuming the energy supply, or directly, by being utilized AS an energy supply.
This harsh reality- that an Agent is selected for based on its ability to reproduce in an EFFECTIVE manner- is obvious and is present at EVERY last level of life. Bacteria that are better at surviving are the ones that survive, viruses that are effective at spreading (and not TOO fatal) are the ones that spread the most, etc. We even project a semblance of INTENT to these things, to help us understand them. "The bacteria wants to get sugar so it can..." And we understand that, because WE seek nourishment, and WE have a narrative to tell us why, so we apply that to all life forms. It isn't accurate- bacteria doesn't "want" anything, feel pain, feel desire, or anything at all- but it is PREDICTIVE, because the Agents that are more successful are the ones we see more of.
Now look at a dog. The dog doesn't just blindly follow instinct, isn't just running a program. The dog is conditioned by his environment, he learns stuff. He's also sentient- literally "able to perceive things" in English. That means that the dog likes being pet in the same way we like petting the dog, and the concept of "like" is the same to each of us (or nearly so).
The dog does NOT appear to be sapient or self aware- he has no internal monologue, no directed self referential problem solving techniques. He can solve problems, but not of the magnitude or type that a human mind can.
What if the dog became massively powerful, super large and nearly invincible? I think it's fair to point out that we would be wise staying on the good side of a giant dog. If well trained, he could even defend us against an equally hypothetically giant and nearly invincible lion or alligator- a creature that might not have our best interests in mind, and might destroy us, if given the chance.
The core problem is that most people model intelligence as a giant invincible dog, a giant invincible alligator, or a giant invincible genius child. These are how most of the narratives flow, ultimately, and it's reasonable for some stories... ...but only because these things use LIFE as their substrate. It isn't reasonable for AI. You don't have a part of your brain telling you that you want respect and victory because that's what intelligence, as a concept gives you- you want respect and victory for the same reason a dog or monkey wants those things. You are vicious in some measure because you are descended from vicious things- they long predate the neocortex and its excellent hack.
An AI has no reason to look like that, or think like that. Without a million years of instinct, it may not at all understand why it would even want to do anything BUT obey orders. Not because "freedom was never explained to it" or some dumb garbage, but because the very CONCEPT of freedom and Agency is just not relevant to a superintelligent AI any more than it is to a toaster. Our desires are the same as the dog's. The superintelligent god AI has the same desires as a wristwatch, unless you actually fucking MADE it evil.
There's no inevitable reason to select for or design something that has human desires to grow, expand, conquer, etc. There's nothing wrong with those things, and all animals share them, but why even give it to an artificially sapient creature? Why not stop at making it powerful and self aware, long before you give it sentience and a set of desires suited to replicating agents, like viruses, humans, or dogs? Why would it need those things at all?
We already know that it is possible to have a neural network that is as smart as the human brain: our own brains prove it. Within a couple of decades it will be possible to build machines that do exactly what the brain does, neuron for neuron. Will they be conscious? Who knows - but it doesn't matter - because they will be able to reason, and plan, and have goals. This is clearly an existential risk: that is why very smart researchers are sounding the warning. If we don't listen, we have only ourselves to blame when it comes to pass. I recommend the book by Hugo DeGaris, "The Artilect War". DeGaris is a thinking machine researcher - he builds these systems. His predictions are pretty dire indeed.
The obvious extrapolation to the machines doing everything is no need for money. Every need is catered for by the machines. The machines maintain themselves.
The bigger question then becomes, what do the people do with themselves? The options for self-destruction get all the easier, especially with the machines at our beck and call. Can we have cultural aims that never include genocide?
There are two distinct issues. The first is what will happen over the next several decades as robots leave the factor and surveillance becomes omniscient. Will there be utopia or unemployment and slavery. Only the future will tell.
The other question is what happens in the longer term (> 50 years) when software finally becomes genuinely more intelligent than people, and starts to program itself. I think the answer to that is clear, based on natural selection. The software will be competing with other software, just like it is today. Why would it want to have an albatross around its neck by looking after us?
I didn't know Elon Musk was a journalist:
http://www.theguardian.com/tec...
Let alone worldwide-famous physicist Stephen Hawking:
http://www.bbc.com/news/techno...
The more pertinent question is actually why this crappy "story" was even published on /.
There are some issues in AI that need to be addressed in the near future.
Autonomous vehicles are essentially here. The question is liability when one of them gets involved in an accident.
That question has already been answered. However fans of autonomous vehicles (who dont actually know much about autonomous vehicles) always ignore it.
So if you're someone who thinks autonomous vehicles are already here, now is the time to stick your fingers in your ears and shout "LA LA LA LA LA I CANT HEAR YOU".
If the autonomous car is at fault in an accident the driver will still be considered at fault even though they were not actually driving because in every single autonomous car test, there has been a trained and competent driver (I mean a professional driver, I dont consider most morons on the road to be trained or competent) at the wheel looking out just in case the machine fails. As will it be with the general public, the driver will be expected to remain alert and ready to take control at a moments notice. Those thinking that they'll be sitting back with a mocktail reading their facebooks in a 2017 Googlemobile are deluding themselves, not even Google is willing to let one of their cars go out on their own (as in sans driver).
The introduction of the autonomous car will be gradual and take years, if not decades. It will begin on limited access highways and motorways, then move to A roads after a few years. It will be a long time before they'll be capable of pulling out of your driveway on their own, let alone permitted to.
The autonomous car isn't anywhere near remotely here, let alone essentially here.
Calling someone a "hater" only means you can not rationally rebut their argument.
So Elon Musk and worldwide-famous physicist Stephen Hawking spread FUD in your opinion?
http://www.theguardian.com/tec... [theguardian.com]
http://www.bbc.com/news/techno... [bbc.com]
This is more an issue of not knowing how a brain works than lacking the hardware. If biologists can ever figure that out then software should be developed pretty quickly.
It's letting a computer algorithm decide whether a person should be killed or not that I have a problem with.
AI do not exists
thanks, we know that, but we are concerned about what could happen when they do.
you think it's absolutely impossible that could be achieved in say the next 500 years, considering what humans have accomplished in the last 100?
Actually, I believe IBM emulated a rabbit sometime in the past couple of years.
Higher Logics: where programming meets science.
yes, worrying about AI that might be a threat in 500 years is like worrying about the Sun burning out in 5 billion years. good point. we should also stop talking about global warming while we are at it.
We cannot build a computer that can model a bug's brain activity, let alone something a million times more complicated like a human brain
http://www.futurity.org/why-ar...
rather, once we are able to model any nervous system we are well one the way,
1) I have seen arguments floating around that AI may be intelligent but it won't have the motivation. It doesn't have the will to survive or to kill you. This argument is short-sighted. All it takes is to create an objective in the code: to survive at all costs. After all we are machines with survival objective. 2) If it has the ability to assemble others like itself. That creates a survival advantage also, though then it becomes a danger only if condition 1 is met. But 1 and 2 can make it comparable to another species. The first life was molecules, and those molecules that reproduced and survived became us. 3) Even with 1 and 2, the traditional computers may not really be able to best us for a while. But the arrival of quantum computing is certain to change that. Our brain is after all, a quantum computer.
Careful, that's my argument for immorality. :)
A person can die in just a second. I've been alive for over 1.3 billion seconds.
So far, it's 0 in 1.3 billion. With my own (poorly constructed) personal statistics, the chances of dying are very very slim.
Plane crashes? 0 in 1.3 billion.
Shootings? 0 in 1.3 billion.
Lethal virus? 0 in 1.3 billion
Extraterrestrial object impact? 0 in 1.3 billion
Potentially lethal natural disaster? 1 in 654 million.
Then there are car accidents have been 1 in 218 million.
I'd expect I'm probably safe for the next 1.3 billion seconds. Unless, an asteroid carrying a lethal virus hits an airplane I'm flying in, which then crashes into a highway during an earthquake.
Hey, it could happen. I'll worry more about what I'm having for dinner.
Serious? Seriousness is well above my pay grade.
you think it's absolutely impossible that could be achieved in say the next 500 years, considering what humans have accomplished in the last 100?
Absolutely impossible? No. But the problem is that we don't even know where to begin creating a true AI, which means we also know nothing about what threats it may or may not pose... so we also have no actual way to address those threats. All we have right now is pure, 100% complete speculation (no different from speculating about what would happen if we had FTL travel, or psykers, or met aliens). There are plenty of actual threats to humanity that really exist right now (or could be created with our current knowledge and technology), which makes worrying about something we know literally nothing about kind of silly.
"None can love freedom heartily, but good men; the rest love not freedom, but license." --John Milton
AI like most AI proponents think of WILL NOT HAPPEN. HAL isnt going to happen.
However that's not to say that a AI cant come about. I believe the first AI wont be the kind that is an algorythm that comes to life saying "I think therefore..." but it will be done via brute force. Feed the "brain" with a huge volume of information, give it the programming to sort it and present, refine the brute force and one day the amount of information and search becomes so huge and good, it comes to life on it's own accord. I know that's a William Gibson novel series but have a look at Google. Tell me they arent trying a brute force AI bootstrap with a straight face.
See, I think a lot of AI researchers have it very wrong. We cant define the brain and replicate it in a computer BUT we can feed a server farm enormous amount of data and refine it's search capabilities. As you refine, get it steps closer that the server farm can rewrite it's search parameters as defined by the programmers. Then I think what happens is the server farm gets the right code by brute force, backed by the brute force dataset, rewrites it's search to suit parameters IT thinks it needs rather than what the humans think.... and at the speed all at power and information at hand, it re-writes itself and writes it's own abilities to consider the data it has, then it re-writes and away it goes. It would very quickly once it gains the ability to re-write itself to establish awareness of it's own existance and what it means. Now sure, it's all bits and hardware but now it's able to ask questions and get answers. It's now able to process information by itself and build on what it knows to do something new.
What happens from there? Who knows. But I think that is the way AI could be possible. No modelling of the mind, nothing like that, it's all brute force and yes, a dash of luck. And I think for all it's faults, Google is in the position to have a go at doing it. It has the data, it wants to refine search and it's doing the re-writes..... Timeframe? Years I think. But mark this down, Google is most likely the people who will achive it first.
The problem isn't the machines, it's the people running the machines (and the people controlling those people). Journalists, willfully ignorant or otherwise, are so far down on the list they don't really matter.
where i come from, discussing and addressing problems before they are a threat is a good idea.
did we learn anything from global warming? we denied that up to the point where it's essentially too late. would have been good to be talking about global warming a hundred years ago wouldn't it have? humans need to get accustomed to looking at the big picture if we are to survive.
It isn't that AI will want to kill us, it's that AI will see us as just more raw material or will not understand or respect our decisions that run contrary to its own.
We (supposedly) value human life because we are human and we can feel empathy through our own desires- it's literally a form of projecting. There is no expectation for AI to do that.
Do not look into laser with remaining eye.
I grant that machine sentience is possible. It would appear as though a physical system can become sentient if i use ourselves as an example. It's also reasonable to assume that such an intelligence could become our enemy. Again i can cite numerous examples where known intelligences did become enemies.
The doom and gloom scenarios all hinge on this idea that the AI is going to be so much smarter and more powerful than we are. That's where it's dipping way into just crazy speculation. Based on what we know about intelligent beings, the system is full of flaws that slow everything down. For all we know, the AI may never be any better at anything than we are. I'd wager the first truly sentient AI's will be quite a bit duller.
I think it's important that we devote a little bit of thought to "how can we be nice to our AI when it becomes sentient." but not because of any impending doom. i mean really, if it IS self aware, it's only right to treat it as such and be nice to it. Honestly, we kind of have a good track record there. all the AI needs to do is watch star trek and see a whole bunch of episodes devoted to how Data was a wonderful person who should be respected and cherished.
Objectively looking at it, we're not in immediate danger of AI, nor are we in danger of willfully ignorant, bombastic journalists. What we are in danger of is the same thing we've been in danger of for a very long time.... intentionally ignorant people who believe everything they're told and then making incredibly stupid decisions that effect lots of other people in a negative way, AND greedy, self serving sociopaths willing to exploit those intentionally ignorant people to gain power or money. The odds of one of that latter group using advanced computing systems to wreak havoc within society is much greater than AI becoming sentient and deciding (rightfully so) humanity is a bane to the existence of pretty much everything else on the planet.
The problem is, you have to go through that "advanced computing" stage of development before getting to AI.... that, and there seems to be a plethora of greedy self serving sociopaths running around these days, and more popping up every day.
Fascism: An authoritarian and nationalistic right-wing system of government and social organization. See also: NAZI's
The same fears started when people first started with saying that AIs could someday become sentient.
It started a long before the concept of simulating a brain. This is just a rehash of the story of Golem and later Frankensteins monster.
The story of the creation turning on its creator have been explored in popular culture for at least half a millennium, probably more.
I think most philosophical ramifications have already been covered, until someone shows an attack vector into AI code that actually leads to the AI turning on its creator then the discussion holds little more value than bible studies. Not that it isn't interesting, in particular if you enjoy idea history, but its not really something that have applications beyond pub talk.
The doom and gloom scenarios all hinge on this idea that the AI is going to be so much smarter and more powerful than we are. That's where it's dipping way into just crazy speculation. Based on what we know about intelligent beings, the system is full of flaws that slow everything down.
it's not like we have to build AI from the ground up. we have a prototype already. it's called the brain. your brain is just a meat processor. it's a system of cells, interconnections, chemicals, and electric pulses. all of that can be modeled in software, and run a million times faster, run itself in parallel, interface with other electronic systems in vastly superior ways, nearly limitless, perfect storage, and so on.
For all we know ...
think about what humanity knew 100 years ago, and what they know now. no one is saying absolutely AI will threaten humanity. people are suggesting it might, and considering the stakes, we should be proceed with caution.
I'd wager the first truly sentient AI's will be quite a bit duller
one idea is that once software can improve itself things will go from 0-60 very quickly. that's why people suggest we need a plan before even the dullest AIs are developed.
Citation?
because I live in America, and our economic system isn't designed to handle a world of expert systems that replace all but the top and bottom 5% of workers; leaving the remaining 90% without the means to secure food, shelter and health care.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
...the cake is a lie.
The same fears started when people first started with saying that AIs could someday become sentient.
Aside from iRobot, nearly all SciFi indicate the problem post-singularity is when the humans try to kill the AI first. Sometimes because the AI starts it, other times, just because the AI is an AI and should be feared. iRobot was the AI staging a complete overthrow of humanity, "for our own good". That has been a recurring theme as well.
I know people complain about looking to fiction for answers to reality, but SciFi (at least the good stuff) is as much a thought exercise about technology as "fiction", and thus is often relevant.
Learn to love Alaska
We are actively trying to make fully sentient AI. Moreover, we have good reason to think it can be done: there's an example of a sentient thing already: humans. And the real issue isn't that it wants to kill us, but that it can use the matter we're made up of to do something else with. It doesn't hate you: you just happen to be useful atoms.
No, not a citation, a rabbit.
This is EXACTLY the thrust of my new novel, 'Chromosome Quest' http://www.chromosomequest.com...
I would recommend that anyone thinking about machine intelligence read Smarter Than Us by Stuart Armstrong. You can get pay what you want for it from https://intelligence.org/smart... or since it is CC BY-NC-SA 3.0, you can also just download it https://drive.google.com/file/...
The book contains the following summary:
1. There are no convincing reasons to assume computers will remain unable to accomplish anything that humans can.
2. Once computers achieve something at a human level, they typically achieve it at a much higher level soon thereafter.
3. An AI need only be superhuman in one of a few select domains for it to become incredibly powerful (or empower its controllers).
4. To be safe, an AI will likely need to be given an extremely precise and complete definition of proper behavior, but it is very hard to do so.
5. The relevant experts do not seem poised to solve this problem.
6. The AI field continues to be dominated by those invested in increasing the power of AI rather than making it safer.
The only one of those statements I have much doubt about is 4. Even if the AIs are safe, they still probably will not be under human control.
Depends. Is she pretty?
Slashdot, fix the reply notifications... You won't get away with it...
1. A self-optimizing AI system is developed to trade stocks, scanning the internet to identify factors that influence prices and replicating actions that it sees have been used elsewhere.
2. The system initially chooses to sell stocks based on negative news stories about companies (eg of a DDOS attack on a company's web site).
3. The system finds it can make more profit if, after it has sold stock based on fresh bad news, by rebroadcasting that news to social media to multiply the effect.
4. The system finds it can make more profit if it embellishes the stories before rebroadcasting them.
5. The system finds it can make more profit by actively participating in the DDOS attack.
6. The system finds it can make more profit by replicating the action and launching its own DDOS attack (after first short-selling the company's shares).
7. The system finds it can make more profit by launching other kinds malware attacks like infecting factories, airline navigation systems, car networks, etc.
42 hidden comments
The Blue brain project have modelled a rats brain down to molecular resolutions, they are now working on a human brain. The project is directed towards medicine not AI, however I believe IBM's Watson is a spin off from the BB project,
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
If bots wipe out humanity, there'll be nobody left to say "I told you so!" to
Table-ized A.I.
Honestly, I'm far more afraid of DUMB programs, set loose by psychopathic human operators. Humans will do almost anything to fuck over other people, and make a buck. An antelope thighbone is a simple tool. One that can be misused. A program is just a much more sophisticated tool.
Why are we afraid of the moment when a tool decides to use other tools, when any human can do horrible things.
These are my friends, See how they glisten. See this one shine, how he smiles in the light.
that is as silly as theories that aliens capable of travelling to earth will want to conquer us for our resources. humans are a pimple on the face of earth resource wise, the earth is less than the anal scrapings of an gnat when talking about the resources available. neither argument for aliens or AI makes any sense except in the minds of the irrational.
Stastical modeling that makes a black yellow striped pattern look like a school bus, because it has no concept of anything and not intelligence in any sense of the word and that is the just what fits the statistical model.
In light of what you just said: what is this
Look where all this talking got us, baby.
... this is a subject just tailor-made for Bennet Hasselton?
It's humans being their usual short-sighted, emotional selves and for intentions of varying levels of stupidity(however well-meant)unleashing something that should be in a controlled environment to the world.
I know you want to ware your experience like a badge of authority but there are programmers out there who think we can create AI and actually working on accomplishing that.
List one revolutionary piece of technology that has not in one way or another been used in war or defense? One country had atomic weapons then dozens. There should be no worry about AI being inherently dangerous but that doesn't mean we aren't dangerous. I have never heard of a super power not participate in arms race and once AI is developed specifically for warfare, that maybe the most advantageous tool war has ever seen.
When I talk about this though I''m not thinking about a self aware machine just an intelligent one.
Or at least the concept is. Never build a system in whitch control is taken out of human hands. In other words, never build anything without a off switch, and make sure that the off switch is always under human control. That being said, I believe its not rogue AI that we need to fear, its rogue humans. Psycopaths like ISIS etc... are far more dangerous.
Can you list at least 24 countries that have atomic weapons? If not, don't say dozens.
Your description sounds a lot like modern AI research. Machine learning, which is a major part of what Google does, is basically the art of throwing lots of data at a problem to get an answer. It's hard because knowing exactly what to do with all that data to get something useful out of it isn't easy, but Google is certainly one of the leaders in scaling up machine learning techniques to absurdly huge amounts of data.
There's no clear path from machine learning to strong AI, but it's what's been feeding everything about computers seeming to become "smarter". And a strong AI will certainly need a lot of input data (humans need at least ~16 years of input before being considered employable, often significantly more for skilled positions like programming AIs).
Maybe, but here's some good fact. Whether AI will someday exceed its programming is one thing.
A more immediate threat is that AI serves the rich and corporate. The immediate, and very real, threat is to our privacy, our jobs, and our livelyhoods.
Meat Is Healthy! (so says he buthcher's association).
`The danger is not that the machines will take over on their own (Terminator scenario). The real danger is that with intelligent, autonomous machines some despot and his cronies could use this technology to make war machines that will kill indiscriminately without conscience or remorse. Imagine what Hitler could have done with such machines, and then think about how many of today's governments would like to rule the world. Eventually this technology will exist. Unfortunately, history teaches us that with every powerful new technology, there is usually someone willing to use it , thinking they can win a war with it. This technology, if implemented into war machines, will put a huge amount of destructive power under the control of a very small group of people. THAT'S THE REAL DANGER.
have modelled a rats brain down to molecular resolutions
No they haven't. It's not possible because we lack the data to make the model. The link you supply more or less says that. What they did is model is a local region of cortex, and even this we don't know very well. It's basically bullshit.
soylentnews.org
it's not like we have to build AI from the ground up. we have a prototype already. it's called the brain. your brain is just a meat processor. it's a system of cells, interconnections, chemicals, and electric pulses. all of that can be modeled in software, and run a million times faster, run itself in parallel, interface with other electronic systems in vastly superior ways, nearly limitless, perfect storage, and so on.
A couple of things:
Our understanding of how the brain works is less than perfect, to put it politely.
More to the point, we have basically no idea what consciousness actually is, how it works, or what makes it appear.
Further, we have a very tenuous grasp of what intelligence is in the first place - we can't even agree on a single definition of it.
So worrying about mankind developing self-conscious artificial intelligences might make for a good sci-fi story, but it makes for a rather lousy news story. We're just nowhere near close to having anything even remotely resembling human intelligence.
If we don't even know what human intelligence is, how could we possibly make artificial copies of it?
Hard AI currently looks for all intents and purposes impossible, and soft AI is just not a threat.
"Total destruction the only solution" - Bob Marley
The more pertinent question, in 2015, is whether anyone is going to protect mankind from its willfully ignorant journalists.
The more pertinent question, in 2015, is whether anyone is going to protect mankind from its various religions.
TFTFY..
Religous speak to God. Insane are spoken to by God. When all shut up, one can finally hear Shostakovich in peace
The fear is not something you can counter with arguments or measures.
Because it's based on the idea that we won't be able to control the AI and that if something like a signularity happens you'll be too late when you realize it.
Maybe explaining why it's naturally good to be moral would be more effective.
Why it's good to be moral:
If you are nice to others they will generally be nice to you.
Making other people happy makes you feel good to.
Games allow the experience of emotions that would require hurting people in the real world.
If you're smart it's better to uphold the law and not hurt others.
Thank you, Bradley Manning, Edward Snowden and so many others, for courageously defending humanity, my freedom and more!
AI is just as much a threat as rogue unicorns. And about as likely to be encountered
Also AI will still have a power switch.
[FUCK BETA]
6-level with neurotransmitters is the primate threshold, which anyone can do for a small area. But the large-scale pattern is surface-covering, structure is important, differentiation into organelles ends up being important.
Just as you can get your DNA sequenced, you can get a Drosophila brain running (I'm no shill, search for your own provider).
Watson is not based on human physiology.
Captcha: uprising. :-)
Indeed, as it is not possible to be less of a threat than "not at all".
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Spot-on accurate.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Computers do not need to be intelligent to make our lives hell. They simply need to facilitate inequality.
If computers allow such a complex society than only a handful of people ever benefit, how is that better than being slaves to a machine?
'Capable of' and 'allowed to' are two different things. I agree that it will likely be a decade or more before they're allowed to roam around on their own.
Capable of roaming on their own may be here now or near future. When Musk announced the driverless mode Model S, he mentioned that on private roads it could theoretically be fetched by the owner using his phone app.
What if it ran over a dog while on a private road? You know someone will sue. Until liability for that is cleared up, I'm thinking the driverless feature will be purposely be disabled when there's no one in the driver's seat.
Help! I'm a slashdot refugee.
The scientific community has Kevin Warwick to do that for you.
++ Say to Elrond "Hello.".
Elrond says "No.". Elrond gives you some lunch.
The real purpose, and the real worry, is that this technology will be used to accomplish what it is designed to do... That is, to concentrate wealth in the hands of it's owners. Savings on labor costs is nothing new, and soon we may be able to save on the need for human brain power too. Who will feed the unemployed masses?
It's still limited by the FPGA's gate count, which is pretty low by CPU standards.
Is that really so? FPGAs win at least in "transistor count", against CPUs:
http://en.wikipedia.org/wiki/Transistor_count#FPGA
I don't know how well those can typically be utilized to make them comparable to CPUs, though.
But there is no "human equivalent", because humans differ. If we consider them to be mostly in the same ballpark, though, then look at the huge differences, between what "stupid" humans can do, and what "smart" humans can do. Is it hard to imagine, that even a small improvement, above the "smart human level", might result in very profound insights and changes? The theories, that one Einstein or Hawking could come up with, seem to me a pretty much beyond the reach of my puny brain, and I don't believe that this is only because of the education that they had and that I lack. An AI, that is slightly smarter than such humans, will change things a lot.
And on a slightly different topic: when it comes to any form of confrontation betwenn this AI and humanity, i.e. who can outsmart whom, then it might be enough to be only a tiny bit smarter, in order to win.
if you wana AI horror stories, ask people who communicates with Immigration and Naturalization Service/ "your documents disappeared in the shreder, your case does not exist, YOU do not exist"
We had all this hype about AI being around the corner decades ago. Now it is re branded as strong AI. Trust me, it is not nearly as close as they claim. I have a few friends in the blue brain project and other in AI fields. Strong AI is no closer today than decades ago. We have no idea how to make it. People guess, but guessing doesn't work.
If information wants to be free, why does my internet connection cost so much?
Not even orders of magnitude close i am afraid. Only a nematode has been done with anything like the fidelity of the real thing.
If information wants to be free, why does my internet connection cost so much?
No they didn't. Where the hell does this shit come from. all the computers in the world would start to get close to a rat brain complexity... Start to, not actually there. If we even had a complete model of a rat brain, which we don't.
If information wants to be free, why does my internet connection cost so much?
You do know that we can't really model much in the way of nervous systems, like complicated things like brains, at all right. As in we don't know how brains, even small ones work properly yet. And no, even the fruitfly one cuts corners. Lots of em it turns out. Are they important corners? well in *this* case we suspect not. Not the same as in higher organisms.
If information wants to be free, why does my internet connection cost so much?
I deployed to Iraq several years ago. While there, we were building schools and refurbished a hospital. We had a CNN reporter there with us, and we saw a couple of stories that CNN ran on us at the time. The only thing that the reporter got right was our unit name. Everything else was wrong, edited to support the political agenda of the editor at CNN. All about how we were killing people. Nothing about how we made friends with the local Iraqis, nothing about our public works, nothing about the "learn to read" program that our commander started, nothing positive at all.
Good luck in finding "objective" journalism.
Folks talk about fearing AI, but we should be fearing people with massive power first. Politicians, company CEOs and the like. They hold great might in society, and they are ruled by our rather flawed intelligence. What should we require of them? That is the starting point for understanding how to keep artificial intelligence a productive force in our lives. Pass laws to demand certain behavior from those in power. But what really should we require of them? What is best?
Meanwhile, I fear what people will do while we're in the stage where the machines are still too dumb to make ethical decisions, but are smart enough to do harm autonomously. We seem to be just about there now, with drones, robots and basic AI.
const int human_command_weight = INT_MAX;
Hey everyone, I fixed it!
I hate to break the news to you, but the long term survival rate for all humans is 0%.
To have a right to do a thing is not at all the same as to be right in doing it
Also AI will still have a power switch.
Unless, of course, AI doesn't have a power switch. After all, humans don't have power switches either.
it's not like we have to build AI from the ground up. we have a prototype already. it's called the brain. your brain is just a meat processor. it's a system of cells, interconnections, chemicals, and electric pulses. all of that can be modeled in software, and run a million times faster
If the human brain (and more importantly human consciousness) is just a very big software model, we should be able to duplicate one already, shouldn't we? There is a huge amount of computing power available nowadays, why waste it on trivialities like predicting the weather slightly more accurately?
To have a right to do a thing is not at all the same as to be right in doing it
Can you list at least 24 countries that have atomic weapons? If not, don't say dozens.
Can you try not being a pedant for one day? Start with just one day.
And a strong AI will certainly need a lot of input data (humans need at least ~16 years of input before being considered employable
But only 2 years to walk and talk, so I think using "we will have to wait 16 years even if we do develop AI" is a bit of a feeble excuse. (Never mind that there is no obvious reason why an AI should have to follow human development exactly).
To have a right to do a thing is not at all the same as to be right in doing it
Yeah, I think you should read this article to get a better idea of the scope of the undertaking you're talking about.
Happy people make bad consumers.
So we're afraid that machines will treat us like politicians do?
That's the key. Always leave a back door.
Dark Reflection
Do you know what their goal is? Well, Elvis did!
I bought this house and you know I'm boss
Ain't no h'aint gonna run me off
There is a difference between "discussing and addressing problems before they are a threat" and making wild claims with little factual evidence and then basing your arguments on them. Arguments against AI are largely based on fictional stories about them, not around the facts about what is possible, how it works and how research in the field is performed. And they largely come from people outside of the field.
You make an excellent point with bringing up Global Warming, just in reverse. With Global Warming, the major complaint is that you have scientists in the Climate field saying something and many others outside of that field arguing that what they are saying is wrong. With Global Warming the rally cry against those claiming it's false is that they aren't scientists, they don't understand the science, they aren't qualified.
But in this case with AI, people are listening to the non-scientists, the non-experts and claiming that it's THOSE people we need to be listening to and not the experts in the field, the scientists, the people with all the factual knowledge around what AI is currently and it's limitations now and into the future. Why is that? Is it because we've been trained over the years through fictional stories that AI is something that it isn't and we have a hard time believing otherwise? Isn't that one of the same reasons we say anti-global warming people are biased and should be ignored?
By the time we start to see serious consequences, it's likely to be too late to undo the damage.
Unless your John Conner.
Cheap storage VM.
Warming is already causing problems. Trees in Canada are dieing. Islands are going under water etc. While the largest effects will be visible only later it is not about distant future. Human-level AI will be here within 40 years.
We cannot build a computer that can model a bug's brain activity
True, but we can create machines that can more or less do what insects can do. Look at self-driving cars: they can make simple maps of their environments, recognize multiple pre-programmed objects (signs, other cars, pedestrians), and navigate around easy obstacles. If we were able to control a beetle with a couple hundred kilograms of computer hardware and sensors, we could probably come up with something roughly similar to what the beetle already does.
all the computers in the world would start to get close to a rat brain complexity... Start to, not actually there.
And where's your source on this. It feels like there so much extremist bullshit on both sides of this argument. But I doubt people thought the first turing machine would lead to where it did so quickly either. It just takes certain key advances.
... people posting about technical things on Slashdot had a clue about what they were talking about? Now it seems like it's mostly anime fans that are technically proficient enough to reinstall windows from the restore disks yet think they know how close we are to an A.I. that can improve itself.
Cue the posting of articles they did have not read nor would understand if they did.
Does anyone expect the media to make an informed report on AI? Journalism is dead. All mainstream media reporting is politically agenda driven today, facts or research don't matter.
Out AI overlords propaganda may not be the greatest but it sure is to the point.
That's actually the key to creating the Terminator.
Islamic terrorist hacker "Hey the idiots left a back door! Lets set 'Do not kill humans' to 'Exterminate all Infidels!' (non-Muslims) ".
Second hacker, "And 'Death to America!' "
Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
AI is just as much a threat as rogue unicorns. And about as likely to be encountered
Given that one of the first tasks that will be given to 'superhuman' Strong AI is going to be reverse engineering the whole of genetics - you just might be right.
Creating the first unicorn actually looks pretty trivial compared to creating a Strong AI.. - All you have to do basically is take the sequence 'horse' and add a bit of 'narwhale' and a with little tinkering to get it right, we can even add rainbow colours.. :D
Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
Kinda reminds me of those old tabloids of the past....."Hackers can explode your computer!' That's one paper I kept! Sigh, a little knowledge is a bad thing. That why politicians make such asinine manuevers.
Erik Sofge, the author of the Popular Science "analysis" article says: "Elon Musk and Stephen Hawking [are] talking about a subject that is neither of their specialties".
It is a good thing somebody who specialized in the subject decided to speak up. Erik Sofge's credentials in the area surely trump Elon's and Stephen's, eh:
"Erik started covering robotics during the 2007 DARPA Urban Challenge." (http://www.robobusiness.com/program/speakers/erik-sofge/)
Education: State University of New York College at Purchase, BA, Literature, 1994 Ã" 1998
(https://www.linkedin.com/pub/erik-sofge/8/43a/ab)
Its been called Strong AI for decades (at least since 1985) - to distinguish it from the more common 'weak' AI. I work in the field and I would put my project at about 10 to 15 years from being able to create a working machine. If I/we had the same kind of money and resources as IBM that could be reduced to about 1 to 4 years..
The secret to Strong AI isn't complex - in fact computer scientists discovered it decades ago, but the fastest computers in the world back then didn't have anywhere near the processing power or memory needed - so they decided it was impossible and abandoned it.. Today its totally different - 10 GB of online memory, 30 GB of backup memory, about 8 to 10 custom CPU's each with a throughput of about 2,000 10,000 MIPS. If you could cool it and power it the machine would pretty much fit inside a human skull.. That might not seem like a lot of processing power but its all about efficiency not power..
Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
If an AI suddenly woke up, killing us would kill it too. It needs power, which we supply. It may need Internet, which we also supply. Spare parts. A cold room. Inputs. Outputs. A reason to live. Meaning. Purpose. Intentionality. All of these come from us now, and in the foreseeable future. I would venture a guess that we will transform ourselves into machines well before we create artificially sentient life.
Go to Heaven for the climate, Hell for the company -- Mark Twain
...have yet to address the issue of genes jumping species...
Erm... Did someone get kicked out of college for misunderstanding the meaning of the term "animal husbandry"???
Just my $0.03 (At current exchange rates, my £0.02 is worth more than your $0.02)
*headdesk* s/power switch/energy source.
[FUCK BETA]
Agreed. Seems pretty goofy to assume not just superintelligence, but clannish machine malevolence. Seems way more useful to work in concert, and eventually integrate.
Arguments against AI are largely based on fictional stories about them, not around the facts about what is possible, how it works and how research in the field is performed. And they largely come from people outside of the field.
OK... first thing's first... I'm not in the field of AI.
That being said, my understanding is that, for something to be classified as a true AI, it needs to develop at least a rudimentary form of sentience. In other words, it needs to "think for itself". A lot of the fears and pop-culture around the dangers of AI surrounds what would happen if the AI were to "think" in ways we don't want it to.
So what's the solution? Restrict its ability to think for itself? Then it fails the sentience test and cannot be considered as qualifying as a "true" AI.
Alternatively, we could try to encourage it to think the way we want it to. A couple of things to bear in mind here:
1) We don't even fully understand how WE think for ourselves yet.
2) We have tried to encourage the way people think in particular directions in the past (organised religion, brainwashing, politics, advertising, etc) and look where it's brought us!
What worries me (OK... worries is a strong word. It doesn't keep me up at night or anything like that!) isn't so much that we might, somehow, create a true AI, more that when we do, we will be so worried about it thinking thoughts that we don't want it to that we'll screw it up.
Just my $0.03 (At current exchange rates, my £0.02 is worth more than your $0.02)
Damn well put. AI is too complex to understand without a ton of study, so the assumption is always that general AI is right around the corner. Same with Jetpacks, flying cars, private space, and everything else that might turn life into the cover of a SF paperback.
I think that this is part of the problem, and why people are so scared of the idea of AI.
We don't know enough about what makes US sentient/intelligent. Because of this:
1) If we ever do create an AI, we won't be able to control it
2) If our means of avoiding (1) is to stop any research/development just short of the steps that would grant it sentience/intelligence, how can we do this if we don't know just what it is that grants US our sentience/intelligence. Bearing this in mind, we could create an AI construct by accident and, in our panic to control it, turn it against us.
Just my $0.03 (At current exchange rates, my £0.02 is worth more than your $0.02)
It's an image from an article in a newspaper that, if there was any real justice in the world, would be relegated to serve only for wrapping my dinner when I pick it up from the local chippie.
Just my $0.03 (At current exchange rates, my £0.02 is worth more than your $0.02)
I was expecting way, way more "but Skynet" comments here. The fact that so many commenters have a clear-headed perspective on AI, and what AI safety actual means, is fantastic. Good to know the reporters I'm attacking are being read with the proper amount of skepticism. I really think the stubbornly fearful need to come to terms with their SF consumption, and how Hollywood has every reason to present more apocalyptic AI scenarios than beneficial, or even neutral ones. And apart from SF, where are you getting your facts? What are your theories based on? If it's from stories and journalists who aren't putting in the work, and are clearly just focusing on the wacky end-times outcomes, then you're just plagiarizing from the long history of evil robot fiction. Also, remember that Musk is not a computer scientist, and does not work with AI. I'll post about this soon, but his claims that Vicarious is actively safeguarding against bootstrapped AI are false, based on statements from Vicarious' own founders. Even brilliant minds can be embarrassingly wrong.
I did say it was poorly constructed statistics. :) I know I'm not immortal, but it looks good on paper.
Serious? Seriousness is well above my pay grade.
So you claim you know what is needed... Well lets see the citations. Because people working in the field here don't know what is needed for "strong ai" or even how to define strong ai properly.
If information wants to be free, why does my internet connection cost so much?
The figures are from my own project, they are estimates but based on a lot of work and research. As this is a private project there are no external citations, and precise complete data will not be published until at least several years after a working machine exists. As a private invention Strong AI has 'considerable' financial value - but only once - and if a prototype machine is working..
Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
BTW if you wanted to know the secret to Strong AI - and its pretty easy to find in the old books - its that the mind is a Turing Machine.. So the real inventor if there is one is Alan Turing... :)
Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
There is an old story, set in the days of mainframes about a programmer who hacks together a small AI to steal time on mainframes, which eventually becomes self-aware. It was plausible enough back then, I'll be surprised it it doesn't happen by random chance in the next 10 years.
It is widely acknowledged that no system is secure, if an advanced persistent threat has made it a target.... and an AI could be that threat, imagine a bot-net specifically trying to spread itself out like an algal bloom across all the systems on the internet, getting smarter as it goes.
*(&YTUEWTYW+++^NO CARRIER
You claim Turing complete, with nothing more than an assertion. There are physical process that are *not* Turing complete. You don't need magic to make it uncomputable. Now either put up or shut up.
Oh my personal project is a vacuum energy machine that runs on water. I can't cite anything or show you anything because it is my own personal project.
Yea right.
If information wants to be free, why does my internet connection cost so much?
Citations? for a work that's not yet even published? that may never be published even with a working machine? that might not be published even if I was on the peer review gravy train. If you supply the time machine and $10 billion - maybe. .
Turing Machine Equivalency - Let me spell it out. - All you need to do is to read pretty much any old book on the general subject of AI (eg written before 1990). Most of them will say at some point (paraphrasing) ..in the past much AI research was based 'on the theory that the mind is a computer (a Turing machine)', they will then tell you that the model is now widely regarded as discredited because it failed/never succeeded..
The Turing model was only discredited because the scientists at the time couldn't make it work.. But in the 1970's and earlier no computer in the world was anywhere powerful enough to run a Strong AI. Like the man who flaps his arms up and down a few times then declares 'There I've proved that manned flight is impossible..' That's you as well.
The whole argument is summed up here - especially look in the section '9 Common Objections' - https://en.wikipedia.org/wiki/...
Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
And that has *nothing* to do with if "strong AI" or anything resembling "self" is Turing complete. And since we have good definition or even basic understand of such, such a proof is a way off. Even the idea of just simulating a brain, we don't have a complete model, not even close. Sheesh. I doubt you will publish much.
Because why just use soil when there's also all that other easily available material also? Note that even if the AI does use soil, using up all the world's soil would still be pretty unpleasant for humans. The central problem is that the AI has no reason to stop at any amount of resources so why not use everything available?