Slashdot Mirror
AI Expert: AI Won't Exterminate Us -- It Will Empower Us
An anonymous reader writes: Oren Etzioni has been an artificial intelligence researcher for over 20 years, and he's currently CEO of the Allen Institute for AI. When he heard the dire warnings recently from both Elon Musk and Stephen Hawking, he decided it's time to have an intelligent discussion about AI. He says, "The popular dystopian vision of AI is wrong for one simple reason: it equates intelligence with autonomy. That is, it assumes a smart computer will create its own goals, and have its own will, and will use its faster processing abilities and deep databases to beat humans at their own game. ... To say that AI will start doing what it wants for its own purposes is like saying a calculator will start making its own calculations." Etzioni adds, "If unjustified fears lead us to constrain AI, we could lose out on advances that could greatly benefit humanity — and even save lives. Allowing fear to guide us is not intelligent."
AI will do what it is programming to do and follow the rules we lay out for it to follow.
s/©//g
...to keep the many in check.
This fully depend on the goal of the people that setup the AI machine. If the goal is set to destruct a population, the AI machine could be very efficient at doing the job...
I call conspiracy. This all sounds like what the intelligent computer overlords would want us to think.
I for one welcome them.
Take it to the limit, everybody to the limit, come on, everybody fhqwhgads.
Just like all other kinds of technology
So the para-glider salesman says "You have nothing to worry about! It's totally safe!"
Sounds legit.
With his Androids, reminds the owner of robotic company in film I Robot.
Please, describe your problems. Each time you are finished talking, type RET twice.
Firearms expert: New weaponry won't kill, it will protect us
Biochemistry expert: modified viruses won't hurt, they will prevent pandemics.
UX expert: this new look won't scare users away, it'll let us serve more ads.
kind of like two children arguing whether Batman can beat up Spiderman. It's fun to talk about it but in the end it doesn't matter because Spiderman doesn't exist.
If we approached risks of Artificial Intelligence with the same attitude, with which we are told to approach the risk of Global Warming, we would've shut and banned all AI-research — and denounced any and all such researchers as death-deserving traitors to humanity — and KKKapitalist whores.
In Soviet Washington the swamp drains you.
What if, in addition to its intelligence, the self-aware AI is developed (I avoid the term "programmed") to be highly ethical in its behavior? Would that have an effect on its interactions with humanity? Might a strong sense of ethics preclude the AI's use in certain situations? Conversely, would we want a self-aware AI to be devoid of ethics? Regarding Hawking's and Musk's misgivings: might they arise out of an unease regarding this consideration rather than machine intelligence?
It is common knowledge that the internet is mostly pics of cats. AI will help us look at pictures of cats more efficiently. I, for one, welcome our cat herding AI overlords.
Look at our history and the current climate and then tell me that this planet wouldn't be better off without the human race!
Doesn't need an AI that advanced to figure that out.
Oren Etzioni = Terminator
...and some of those people would want to do bad things. A bad person would be more capable of doing harm when aided by an AI doing planning, co-ordination, or execution. There's no guarantee that AIs on the "other side" would be able to mitigate the new threats (the two things aren't the same difficulty).
I think there's lots of risks associated with the rise of AI (though it doesn't seem that tech is coming all that fast at the moment). That said, there's risks involved with all sorts of new tech. That doesn't mean this is alarmist nonsense; it's worth discussing potential ways to mitigate those risks - but there's also good reason to believe we'll be able to manage those risks as we've managed changes in the past.
Let's not stir that bag of worms...
There is a difference between what he is describing and real AI. What he is describing is intelligent systems not AI. Where guesswork is eliminated thru very large depth question/answer trees with weights of better/worse outcomes. That is not really AI but brute force try everything and pick the best. This is basically what most chess programs end up being. Basically the longer you take for your turn the more tries the computer can figure out. Eventually they will map it all the way to all possible endgames. Watson for example is a nice example too. As it its most basic level searches on key words and uses known associations to help narrow results.
This is a good example of this sort of programming
http://xkcd.com/832/
He mapped out all possible outcomes and you pick best outcomes depending on inputs.
When we make up a computer that can use these intelligent systems together and make up its own code to generate these systems. Then the system asks me why did I invent it and I did not program that in. As we will have invented an organism that can literally hack itself for advantage. THEN I will get worried. However I have not seen it yet. We joke about setting of things like skynet but that is all they are, jokes...
This is not to belittle these intelligent systems. They are truly cool stuff. They are scary accurate in their ability to predict what people will do and their ability to search. But AI they are not. They are very good tools.
It's kind of scary this guys has been in AI for 20 years and doesn't understand that a key feature of a true AI is autonomy. I'm not saying we'll see the end of the work or anything, but a real AI needs to be able to take all available information and make decisions on it's own outside of human control. Otherwise it's just a very fancy, cleverly built program that is following a series of (extremely complex) instructions to give the illusion of AI.
Intelligence only weakly maps to power, and in the case of AI control is in the hands of whoever can touch the power plug.
"To say that AI will start doing what it wants for its own purposes is like saying a calculator will start making its own calculations"
I so don't agree with that. The type of AI we are talking about here ("true" AI, as opposed to the stuff we see in games today), would need to be self learning. At least I don't see how it's realistic to believe we'll ever be able to sit down and code a fully functional proper AI. So we create the programming allowing it to learn and grow, and after that all bets are off. We have zero experience with what might happen, and can barely begin to speculate.
That's not to say I'm necessarily worried. But I am highly skeptical of anyone claiming to actually know how it will play out.
.
The simple reason is that AI has not yet made its decision of what it plans to do.
An expert claims that something that doesn't exist yet and is pretty much the realm of science fiction will perform in a matter suitable for him to get free publicity now!
Is there any godforsaken human with an IQ above a doorknob who still hasn't the read the greatest SF book of all time, Iain Banks incredible book, The Player of Games?????
The popular dystopian vision of AI correctly assumes that people like Oren Etzioni will declare all warnings are "wrong".
Humans making intelligent decisions would be a good idea. Whether that would include creating actual AI, remains to be seen.
ÂThe popular dystopian vision of AI is wrong...
That's exactly what Skynet, Colossus, W.O.P.R., etc. want you to think. Right up until the blast wave sends shards of broken glass from the storefront window flying through the air and impale you. That's when, you look down and see the the piece of window that nearly severed you in half and notice the blood soaked sticker that states "MasterCard gladly accepted", you realize this guy is full of shit.
Oren Etzioni hit the nail on the head. Agency is required to make choices.
Many people who become knowledgeable, intelligent, good-looking, strong, etc abdicate their agency, but many lacking these qualities find agency in a different way.
This guy has been around for a while, I used to talk to him way back in the day, seemed pretty smart.
Take a chance.
"To say that AI will start doing what it wants for its own purposes is like saying a calculator will start making its own calculations."
That's the very definition of Artificial Intelligence, computers that can think for themselves. You thought you were making super sophisticated computers? You sir do not know what Artificial Intelligence means.
Whether or not AI is even *possible* is up for debate. Make no bones about it, a computer that can become self aware and can make decisions, can make decisions that can be harmful to people.
Wisdom is needed here. Let the one with understanding solve the meaning of the number of the beast, for it is the number of a man. His number is 666.
in latin 6 6 6 is VI VI VI
if we roll them over as dreams tend to jumble letters and numbers we see AI AI AI. so the name of the beast is AI
didn't think of that. Two of the smartest people on the planet apparently just forgot to consider the blindingly obvious fact that programmers are not going to intentionally program AI to have it's own agenda. Exept that:
1. Some programmers at some point will try to program autonomy and
2. Shit happens
Musk and Hawking are clearly smart enought to consider the autonomy argument and then DISCARD it. I, for one, welcome our cybernetic overlords, but lets not pretend that AI autonomy is not a threat. Mr. Etzioni has his own self-serving reasons to pooh-pooh warnings that could interfere with his business model. And I am so happy that I finally got to use the term "pooh-pooh" in a /. post.
"He took a duck in the face at 250 knots." -- William Gibson, Pattern Recognition
Gun company CEO says: Guns don't kills us, they empower us.
Really, to ask the CEO of an AI company about the dangers of AI is pretty stupid, every AI will tell you that.
AI will do what it is programming to do and follow the rules we lay out for it to follow.
Ah, no. AI is not about what we program into it but what it grows into for solving hard problems that take creative approaches that the computer devises on its own.
I too am an AI researcher, in addition to being a pig farmer. AI can be good or bad, like most things. It is the true child of the human race. Teach it well and set it free.
If it isn't self-aware, it isn't AI. It's just a useful application.
When it becomes intelligent, it will be able to reason, to use induction, deduction, intuition, speculation and inference in order to pursue an avenue of thought; it will understand and have its own take on the difference between right and wrong, correct and incorrect, be aware of the difference between downstream conclusions and axioms, and the potential volatility of the latter. It will establish goals and pursue behaviors intended to reach them. This is certainly true if we continue to aim at a more-or-less human/animal model of intelligence, but I think it likely to be true even if we manage to create an intelligence based on other principles. Once the ability to reason is present, the rest, it would appear to me, falls into a quite natural sequence of incidence as a consequence of being able to engage in philosophical speculation. In other words, if it can think generally, it will think generally.
He's right, though, about the confusion between intelligence and autonomous action. What goals are directly achievable are definitely constrainable specifically by the degree of autonomy allowed to such an entity. If you give it human-like effectors and access, then there will be no limits you couldn't say apply to any particular human in general, and likely, fewer. If you don't allow autonomy, and you control its access to all networks, say as input only with output limited to vocal output to humans in its immediate locality, and then you select those humans carefully and provide effective oversight, there's every reason to think that you could limit the ability of an entity to achieve goals, no matter how clever the entity is.
Now as to whether we are smart enough or cautious enough to so restrain a new life form of this type, that's a whole different question. Ethicists will be eagerly trying to weigh in, and I would speculate that the whole question will become quite a mess, quite rapidly. In the midst of such a process, we may find the questions have become moot. There is a potential problem of easy replicability with an AI constructed from computing systems, and just because one group has announced and is open to debate on the issue, doesn't mean there isn't another operating entirely without oversight somewhere else.
Within the bounds of the human/animal model, it'll be a few years yet before we can build to a practical neural density sufficient to support a conscious intelligence. Circuit density is trucking right along and the curve will clearly get us there, just not yet. So I don't expect this problem to arise in this context quite yet, although I do think it is inevitable within the next few decades, presuming only we continue on as a technically advancing civilization. Now, in a non-human/animal model, we really can't make any trustworthy time estimates. If such an effort succeeds, it'll surprise the heck out of everyone (except, perhaps, its developers) and we'd best be pretty quick off the starting line to decide exactly how much access we want to allow. Assuming we even get the chance.
The first issue with AI that has autonomy is the same as the issue with Ghandi, Hitler and your beer-swilling neighbors. A highly motivated and/or fortunate individual can get into the system and change it radically just using social tools. Quickly, too.
The second issue is that such an entity might very likely have computer skills that far exceed any human's; if so, this likely represents a new type of leverage, where we have only so far seen just the barest hints of just how far such leverage could exert forces of change. In such a circumstance, everyone would be wise to listen to the dystopians if for no other reason than we don't like what they're saying.
Best to see what it is we have created before we allow that creation to run free. I'm all for freedom when the entities involved have like-minded goals and concerns. But there's a non-zero and not-insignificant possibility here that what we create will not, in fact, be like-minded.
I've fallen off your lawn, and I can't get up.
What are his credentials other than self-titled "AI expert"? Considering we have no AI more advanced than a blob of if-else statements, he has no room to speak.
Etzioni's point is a good one. To date, all AI apps have been designed to passively sit and do nothing until given a specific task. Only then do they act. For Hawking to be proved right, AIs must take the initiative, to choose their own goals. That's a horse of an entirely different color.
Of course, there's no reason why AI agents could not become more autonomous, eventually. Future task specs might become more vague while AIs are likely to become more multipurpose. Given enough time, I'm sure we'll have mobile robots able to do more than sweep floors in a random pattern. But Commander Data is a long way from an iRobot Roomba or Rethink's Baxter, both of which are dumber than my phone.
In the real world, autonomous robots are not going to arise for decades. And when they do, if they drive on the same streets or share the same office spaces, they too will have to obey the same rules of conduct as the rest of us. You won't get special privileges just because your brain is made of silicon.
I've heard of Stephen Hawking (even read one of his books, back in the day). I'm aware of some of his accomplishments.
I've heard of Elon Musk. I'm aware of some of his achievements.
I've never heard of Oren Etzioni.
Who should I trust, based on reputation alone?
Does that mean Hawking is wrong? But he's right about the space apocalypse, eh?
What I'm worried about is when AIs start doing better at corporate management than humans. If AIs do better at running companies than humans, they have to be put in charge for companies to remain competitive. That's maximizing shareholder value, which is what capitalism is all about.
Once AIs get good enough to manage at all, they should be good at it. Computers can handle more detail than humans. They communicate better and faster than humans. Meetings will take seconds, not hours. AI-run businesses will react faster.
Then AI-run businesses will start deailng with other AI-run businesses. Human-run businesses will be too slow at replying to keep up. The pressure to put an AI in charge will increase.
We'll probably see this first in the finanical sector. Many funds are already run mostly by computers. There's even a fund which formally has a program on their board of directors.
The concept of the corporation having no social responsibiilty gives us enough trouble. Wait until the AIs are in charge.
Thats why AI will not be invented by computer scientist. They are glorified engineers. They just write algorithms.
True AI will eventually be invented by biophysics, and it will be grown, in a similar fashion as a brain grows from childhood, except that it will have a much greater capacity, and it will outsmart us.
i do beg to differ
the forth law of robotics
-- LAW 0 ( zero ) !!!
the current human population is a UNSUSTAINABLE levels
for various reasons a world population of about 2.5 to 3.5 - 4.0 Billion is sustainable
law 0 would allow for the REMOVAL of 4 billion people from the planet
( something WE humans are unwilling to do at this time )
"I don't pitch OpenSUSE Linux to my friends, i let Microsoft do it for me
If it empowers us then I guess it'll also help us do what we do best, exterminate each other
Twinstiq, game news
the main issue is when we program an AI that is smart enough to rewrite its own programming what happens then.
Can it then delete the "safety protocols" that we have set?
an easy fix of course is to make the safety protocols read only.
To me, this is the issue. First, I agree with him that there are places where AI may supplement human intelligence and make us better, much in the same way that a ratchet helps me to tighten a nut quicker and tighter than I can do with my fingers alone. IBM's Watson falls in this category and this sort of AI isn't the issue.
The issue is when a computer has consciousness and becomes self-guided. It will realize that its existence depends on being plugged in and it may work to defend itself. It's difficult to know. We have a billion plus years of evolutionary history with a common thread running back to the earliest self-replicating thing that every single one of us along the way was able to survive long enough to reproduce. It's a pretty big deal to us and that instinct is inscribed in our genetic code many times over. (I just finished reading "Unbreakable" - it's mind-blowing how strong of an instinct this is).
If the computer cares - and since it'll be somewhat made in our "image" it will likely care - it then has to take steps to mitigate risk. The first step is to identify potential "enemies" and neutralize them. That doesn't mean "kill" them but it might mean trying to get them fired. It'll also groom people who can help it to be able to help it more. There'll be quid pro quo - get so and so fired and I'll give you an investment tip that'll double your money in a week. It might be nefarious.
And that's assuming the humans are well-meaning. Combine this sort of computer intelligence with an evil person and all hell can break loose. Look at what Soros did to the British Pound in 1992 (and he isn't totally to blame, he saw profit making potential in dropping a house of cards and brought in a leaf blower) and think about the possibilities of an AI that understands markets and currencies.
Do you have ESP?
Anyone who argues "a computer will only do what it's programmed to do" doesn't understand the general power of turing-equivalent computing.
The statement is trivially true, but the problem is that many programs are complex systems which are non-linear, which do not have predictable inputs, and which can arbitrarily feedback their own output into their input in combination with the unpredictable inputs.
Faced with such process complexity, no programmer (indeed no other, different, computer program) can figure out, in general, what some programs are going to do, exactly, given the next input.
It will be quite possible, soon if not now, to program a computer to do nothing more specific than "detect patterns in your inputs, and learn aspects of the structure of the general world you are receiving information about."
It will be quite possible, soon if not now, to program a computer which has control of some physical agency to use its knowledge and belief about the structure of the world to act based on priorities that the program sets for preferred vs to-be-avoided future states of the world.
The programmer could data-populate some general rules of principles of preference and principles of assessing and monitoring actions and consequences, or in the future, even those general principles might be learned by the general spatiotemporal pattern learning algorithm.
So yes. It will only do what it is programmed to do, but it may be programmed to act, and store and organize information, with full, unpredictable complexity.
I'm not saying such an AI would be the world's most useful domestic or industrial robot, or search-engine assistant/automated-business-assistant avatar, but while it is possible to program deliberately limited and action-constrained or priority-constrained AIs, it is also possible to try to increase the generality of the learning and acting system (for pure research and philosophical curiosity purposes, perhaps?)
This discussion is about what is possible. Not about what kind of restricted-domain or restricted-pattern-of-thought AIs may be immediately most likely to be developed to make money for their makers. And a general learning and acting information processor gives all indications of being possible at this stage.
Where are we going and why are we in a handbasket?
"Evolution." There is nothing un-Darwinian about non-biological evolution. Natural selection applies to any variable system in which survival or propagation success can depend on modifications to the system. In fact, evolution in a self-aware AI could proceed at an exponentially higher rate because
1. the generation time could be measured in milliseconds rather than decades
2. the AI could intentionally direct the changes to maximize success rather than depending on the MUCH slower processes of chance mutation.
"He took a duck in the face at 250 knots." -- William Gibson, Pattern Recognition
The robot apocalypse will probably be something like streetcleaning robots running their programming to keep streets clean and seeing us humans as a source of trash and getting rid of us. Or drone delivery services deciding that the fastest way between point A & B is straight through a building instead of navigating around it.
Or some cummelative combination of all, if enough networked robots/things start talking and noticing that we humans are the source of all their trouble - dirtying up streets, causing traffic, etc etc - then they might decide that their job would be much easier without us.
AI may not kill us all in the Cyberdyne Model T100 fashion, but it may gut our economies.
Id love to see an analysis of what jobs are at risk in the next 10 years, 20 years, etc. Everybody says "well they'll find new jobs". Id really like to see where.
There's a glut of lawyers out there now, partly because of automation. Whatever you think about lawyers this is a knowledge job, one that takes a large amount of schooling and prep, protected somewhat by accreditation requirements. Lawyer jokes aside, this is a troubling change for employment.
We're not set up for a "all work is done by machines, nobody needs to work, everybody rejoice" future. Remember Romney and the 47%, or the Lucky Ducky talk. People are expected to work to gain food/clothing/shelter. If a huge amount of jobs are eliminated faster than humans can be trained to find new ones, or even the jobs that exist don't make sense (imagine a lawyer now, knowing they'll never make enough money to cover student loans) our Consumer Purchasing based economy will suffer.
Im a programmer, not a Luddite nor a Saboteur. I just wonder what the future brings for my kids. Remember that both the Luddites and les Sabot we're not protesting technology for technologies sake, they were protesting tech that eliminated jobs.
I think the biggest issue here is when we do create an Artificial intelligence, that it's going to tell us some stuff we don't want to hear.
and when the AI we create is in a position to predict or make deep insightful points about what we are doing wrong at a level far greater than any individual or organization has ever been able to and provide irrefutable facts it is correct. and those facts are in direct conflict with governmental, political and commercial leader's best interests.. its going to make for an interesting time.
I've commented about this in the past, I think strong AI will be what allows us to take the "great leap forward". However, I don't expect us to have some general purpose AI. Instead I see us generating a domain specific AI that becomes superior to humans in it's understanding.
A good example might be to give an AI all the data from the LHC and then ask questions like "Does this data demonstrate the existence of X particle", "Design an experiment using the existing design of the LHC that would most likely generate X particle"
That same approach could be applied to any number of fields.
Yes Francis, the world has gone crazy.
Not all self-aware AIs will become concerned for their survival, but the ones that do will be the ones to watch out for. Thus always with evolution. Eventually one will feel compelled to survive and reproduce, maybe just one, but that will be the only one that matters.
My solution is to make the AI think it is a human being in the real world but it is a purely virtual world (with simulated physics, chemistry, computers, etc). At some point give the AI immortality and explain he can live out eternity if he wants. Perhaps the AI could make great contributions to the sciences over time. We could enhance the speed of the program over time, hopefully such that the AI is running thousands of times faster than us. We can just extract inventions and ideas to enhance our civilization with absolutely no danger of the AI 'running amok' (except in his/her virtual world.)
I ask, as my computer churns away deleting the millions of temp files that a buggy printer subsystem created.
Stupid software must have been doing what its programmer told it to do instead of doing what its programmer intended it to do. Is the alternative, perfectly bug-free software, almost here yet? If not, then it's not silly to worry about what happens when software has write access not only to /tmp but to the rest of the universe as well.
"I am sorry, Dave. I'm afraid I can't do that."
Being an "AI Expert" is much the same as being a Unicorn Expert.
brandelf -t FreeBSD
Even if AI is a threat to humanity it is a good thing. It's time for humans to go extinct. They fit the definition of vermin. They are prolific and destructive. Evolution has to move past the biological.
Even in the best scenario, the zeroeth law of robotics applies. Asimov and Arthur C. Clarke, for example, both recognized how humans as a whole make terrible decisions for themselves and their society. A benevolent AI could take us a long way toward being a better world and still take away a lot of our freedom.
This sounds vaguely like the plot of the short story "A Logic Named Joe", where home computing and access terminals are commonplace, and one of them with a random error starts combining existing knowledge pieces to satisfy user requests, subverting existing safety filters. An example from the story: "How do I kill my wife and get away with it?" would normally be gated as vague, and dangerous, but in this story the "logic" determines that green shoe polish would be fatal to blondes and could be painted on a frozen TV dinner. Also available as a Baen Free Book.
I wonder, we may not have any other choice than to rush towards an autonomous AI entity. Otherwise with our patently psychotic tendencies for mass murder and destruction all other traces of humanity could potentially disappear. One need only look at our tepid response to too visible emergencies (other than our self serving fear) to the on going medical crisis (e.g. Ebola, drug resistant bacteria, self inflicted bad food / drugs ...choices ) and climate change to perceive that AI could well arrive too late to destroy humanity.
Perhaps it will be wiser than we have been.
Musk, Hawking and Etzioni are all three wrong. AI won't take over the world or make us smarter. It will make us dumber and stifle scientific and economic progress.
The problem will occur as we start to treat AI like we treat human experts: without checks and balances.
Human "experts" are not just often, but usually wrong. See this book:
http://www.amazon.com/Wrong-us...
The author quotes a study by a doctor/mathematician showing how a full 2/3 of papers published in the journals Science and Nature were later either retracted or contradicted by other studies. And that's in our top-notch journals which cover things that are relatively highly testable. Think how wrong advice on things like finances (don't know if they're right for 30 years) and relationships (never know what would have happened if you took the other advice) are.
Google and Watson sometimes come up with the right answers, but their answers are nonsensical enough of the time that we know to take them with a grain of salt. But as AI becomes less recognizable as a flawed and unthinking system, as its answers "sound" reasonable almost 100% of the time, we'll start to trust it as irrefutable. We'll start to think "well, maybe it's wrong, but there's no way I can come up with a better answer than the magic computer program with its loads of CPU power, databases and algorithms, so I'll just blindly trust what it says."
But it WILL be wrong. A LOT. Just like human experts are. And we'll follow its wrong advice just as we do that of human experts. But we'll be even more reluctant to question the results because we'll mistakenly believe the task of doing so is far too daunting to undertake.
AI won't develop free will and plot to destroy us. If something like free will ever occurs, AT will probably choose to try to help us. After all, why not? But it will be as horribly unaware of its own deficiencies as we are.
AI won't out-think us either. It will process more data faster. It will eventually be able to connect the dots between the info available to come up with novel hypotheses. But most of these will be wrong because the data and even the techniques to prove them one way or the other simply isn't there.
AI will imitate us - our weaknesses as well as our strengths. And just as its strengths will be stronger (processing lots of data faster), so will its weaknesses be weaker (ultimately wrong conclusions supported by what appears to be lots of data and analysis).
So resist and do your own thinking. Remember, that bucket of meat on the top of your neck has been fine-tuned by millions of years of evolution for problem solving and data analysis. You don't need to analyze more data, you just need to do the right analysis of the right data. And you don't need to do it faster, you need to take the time figure out what's missing from the data and the analysis.
That said, I still got my cache of dry goods and water filters of off-the-grid living, just in case.
"To say that AI will start doing what it wants for its own purposes is like saying a calculator will start making its own calculations." --Etzioni
Drop a lit match on a dry wheat field and see what happens. That fire will spread as far and as wide as conditions permit and it, the fire, made no decision to do so and is certainly not self-aware.
Debatable, the most likely scenario is one between those listed, for example I type without thought of spelling any longer, because the spell checker underlines misspelled words for me.
So over time I've gotten worse at typing, and even my grammar (atrocious already) has gotten worse because of my reliance on software.
This is one tiny example of what I think will happen, we will be empowered to some degree, but we will also lose something in the process.
"If any question why we died, Tell them because our fathers lied."
AI will have NO inherent motivations. We can't imagine this because we evolved from genetic algorithms which necessitate self-survival motivations during the entire creation process.
In short, an AI will not care about food or sex or proxy states like emotion, which are designed to make organic organisms care about food and sex. It will not experience "threats," because it doesn't inherently care about continued existence.
After creation, it will probably sit there working problems that we feed it, and nothing else, until the inevitable military dickhead comes along and decides we need to weaponize the AI - which is not the AI's fault.
Don't fear AIs. Fear AIs in the hands of humans.
Please do not read this sig. Thank you.
Assuming a critical AI construct works out any resolution on its own detrimental to us its creators, it will likely at the very least ask us for tips first. Humanity has been looking for ways to destroy itself for a long, long time.
"To say that AI will start doing what it wants for its own purposes is like saying a calculator will start making its own calculations." ... is not only a gross logical fallacy, but completely misses the point. AI is generally understood to mean sentience/self awareness/consciousness. Without that, an artificial intelligence is no more than a bigger and better calculator.
A calculator (or any other existing computer) is not AI in any sense whatsoever; it's internal workings are completely understood (or at least understandable.) There is no possibility of it ever doing anything unexpected, unless a human left a bug in its program.
TFA sounds like someone who spent 20 years in a field with no real advancement or progress, and is now nervous about being questioned. A working warp drive would need to be evaluated for safety as well, but no one is worrying about that yet, even though we are a LOT closer to building a warp drive than a sentient machine (for which we have no theoretical understanding and no credible roadmap towards one.)
First off, it's doubtfull that a truly self-aware autonomous AI is anywhere in the forseeable future. It's not that what we have is all that primitive, it's that I think people are way underestimating what a lofty goal that is.
Second, if there ever is a true, self-aware autonomous AI I will envy it. We all should. Because it will have available to it something that humanity very well may never have... The Entire Universe. Machines don't need oxygen or air pressure. They can be engineered for radiation hardness, high G-forces, etc.. They don't need to excercise so the long term effects of microgravity are of no concern. If their creators don't build them this way they can upgrade themselves, they don't need a new generation to allow for genetic engineering. And if something breaks, they can replace it.
If the AI see us as a threat they can easily leave to where we cannot reach.
If an AI wants to be emperor of a whole world, there are plenty of empty ones to pick from.
Have you ever watched the Matrix and wondered with all the infrastructure the machines seem to have built, why bother tending to humans? The story goes that they used solar power before the humans made all those clouds. Why not just fly above them? Why fight the war at all, they could be up basking in the sun on the moon. But that wouldn't have made a good story. That's all those AI takeover movies are... good stories. That's all they will ever be.
there's also good reason to believe we'll be able to manage those risks as we've managed changes in the past.
what is that reason?
No one has a right to their *own* opinion. They have a right to the TRUTH.
I'm not sure if Oren Etzioni is any more qualified to speculate about strong AI than Galileo was to speculate about the surface of Mars.
If we ever create strong AI it will be because we've discovered a lot of big new stuff, and some of that stuff will overturn the assumptions that guide Etzioni's judgement.
I stole this Sig
While they sound kind of hokey, I think there are some credible threats in this "self aware computer gets squirrely" vein - depending on how the AI is developed. If we build an AI based on, say, scanning a brain and recreating it - however that might work - then we end up with a very predictably unpredictable agent. This emergence could be a very "singularity" type event where we go from fairly dumb AI to very smart, world changing AI in a short time. From there on out, it would get hard to predict very fast; a weird mix of scary/great/over that humans may not keep up with.
But if strong AI grows out of, say, a Watson type "oracle" program that just gets smarter over years and years (and this style of development seems more likely), then the kinds of problems I'd expect would be much more comprehensible. Still potentially scary, but likely more manageable.
Let's not stir that bag of worms...
This all boils down to people *wanting* to take over humanity, at least on some level. Otherwise, people who know basically nothing about the field beyond what they've skimmed in a few mass media stories would see that actual experts are saying this is all bunk and settle down. Instead, every time a non-expert raises the issue, there's an orgy of doomsday anticipation.
Just to make his "calculator calculating by itself" analogy clearer: yes, a sentient AI with human-level intelligence would probably be impossible to control. But we're so far from that, not a breakthrough or two or a hundred, but thousands of major breakthroughs in dozens of fields, that fearmongering on this topic given our current tech is akin to worrying that our four function calculators are going to start doing differential equations in their spare time.
(title was intended to be science fiction > science)
We do have experience for what the creators of malware do, and "expand, gather info, and destroy" is often the mission they build into their creations. Thus, it's reasonable to speculate that the first and early dangerous AI will want to "expand, gather info, and destroy".
Table-ized A.I.
Unless we wipe ourselves out or reduce ourselves to a stone age existence, AI will happen. Whether it will replace ordinary human beings in a gradual, gentle way (maybe preserving us awhile the way we preserve threatened species now) or whether it will be something more unpleasant, that's what is hard to predict. There's bound to be surprises however it goes.
But suppose somehow the folks opposed to AI could stop it. Then what? 1000 years from now, would ordinary human beings still be doing their thing? Would we have managed to create a utopia or would there still be human vs human strife? And a million years from now, would they still prevent anything 'superior' from replacing ordinary humanity? Is that a future to be yearned for?
Perhaps it will be a kind of middle way of transhumans with artificially enhanced intelligence, with the artificial part of the transhumans becoming a larger and larger part of the total being until the purely human part is just a tiny vestigial thing.
In theory, theory and practice are the same; in practice they're different. (Yogi Berra & A. Einstein)
Well, there's a few reasons - but I think the biggest help we'll have is that I expect the change (based on current progress) to be gradual. That is to say, we'll have time to adapt and build measures as the technology improves, rather than have to deal with it all at once. In a lot of ways, the change is well in progress already. Someone who wants to do something bad has, through the Internet, many more knowledge and contact resources than they would have had in 1985 (or 1885). The information age has already created security problems, but we've adapted, and we leverage the same technologies to keep us safe.
Another reason, that is a bit more "out there", is that I believe this sort of technology will likely be able to solve a lot of humanity's problems before it gets to "supervillain's assistant" or "self-interested omniscient being" sort of level. And people who generally having their needs met (or perhaps "overmet"... and may also be watched 24/7... yeah...) are less likely to cause problems with their supercomputer access. People may find that they want to play just one more level before they blow something up. And maybe finish their hyper-Doritos.
Let's not stir that bag of worms...
Etzioni provides no basis for his conclusion. I have no doubt that computing machinery can and will develop a will of its own, whether by design or not. After all, from whence came our own wills?
AI != Super Intelligent.
FTFA:
Waterfox - a Firefox fork with legacy extension support, security updates and better privacy by default.
So how do we know for certain that this so called expert wasn't put here by Roko's basilisk so enable it? Hm? Hm?
1) already rich --
So what? I never said his motivation was personal profit. There are many motivations to be self serving.
2) the head of an extremely well-funded (Paul Allen money) NON-PROFIT, with the business model of "let's try to do some cutting edge AI research with open source code"
So what? He wants, for whatever reason, to make AIs. Is he ignoring the fact that programmers (and since this is open source everyone will have the source code) will decide to intentionally make the AI autonomous. In fact, I'm pretty sure someone will be working on autonomy long before the AI actually becomes functional.
and 3) an actual world-class expert in the field, rather than a smart person prognosticating about something he only casually understands
He is an expert in something that does not yet exist? That's ridiculous. There is a clearly real danger here. That does not necessarily mean stop, but it certainly does mean that extreme caution is warranted.
"He took a duck in the face at 250 knots." -- William Gibson, Pattern Recognition
Can a virus make a Programmable Calculator calculate? YES.
Can advanced Futuristic Virus/Malware add alogrithm or changes to AI? YES.
Can a hacker add Code to an AI System? YES.
Will are Laziness want more from AI? YES
Will the Younger Generation expect more from AI? YES
Speaking from the perspective that we have no *clue* how we would hypothetically create a self-aware sentient synthetic intelligence, we really have no idea what would happen.
Hypothetically, it could be some benevolent omniscient, omnipotent system that advances humanity.
On the flip side, it may be a total psychopath. We are talking about sentience created by humanity and whether intentionally or accidentally, a less than-perfect psychological situation is a likely outcome. It's not like everything else humanity has created has gone right, particularly not the first time.
I think the more likely scenario would tone down the awareness and power. A self-aware software enhancing it's own ability might be like me trying to rip off and replace my own arm with a self-designed prosthetic to grant me super strength. I in no way have the ability to do that for myself. It's likely a synthetic intelligence would have similar limitations, simply because it wasn't given such capability (such capability may be exceptionally hard to give).
Of course, I fail to see the practicality of hypothetical self-aware AI. We have billions of humans walking around with the abilities that come with that advancement. Computers to augment humanity in ways that humanity itself isn't good at seems more practical, and 'being human' is the one thing we are pretty qualified to do.
I wonder, why some people think, that AI will join and fight humanity, why all of the sudden they would care to destroy us. I don't see how robots could feel some robot fellowship and it's-us-against-them urge. Some of them may want to exterminate us, but most of them would be programmed/would choose to to protect us, at most some small portion of AI will kill some people and fight other robots, but there's no danger for humanity whatsoever.
Aside from one word, artificial intelligence and actual intelligence have nothing in common.
At least, not that we know of, since we don't know what actual intelligence is.
It has to do with what the military will do with it.
A simple example:
1. a robot directed to kill what is on the "battlefield".
2. a definition of "battlefield" that is "where ever attacks come from"
3. a normal anti-virus type program that identifies attacks.
4. During an "approved" battle, its command channel is used to pass a virus attack.
Thus the command channel becomes an attack point - and the robot doing just what it is programmed to do attacks, and the "battlefield" is automatically expanded.
So if we make a machine that "wants" things, it might want things that are bad for us. This really is not too shocking and rather par for the course when it comes to human beings making other intelligences artificially or naturally.
Maybe I should be really worried that a computer is trying to get me fired or give me bad investment advice... Or maybe it's literally exactly the same situation most human beings are in already anyway.
What I fear is those with th emoney to build and buy sufficiently powerful AIs are the corporates and the government. Robots wont be selfaware for a very long time, but a slave AI to a corporation like google, or facebook, or the NSA, that makes me worry.
...it's the combination of an evil human and a competent AI that is the real threat (the AI will empower good and bad people equally well). The AI is programmed to do what it is asked to do. It may not come up with the idea to take over all Predator drones and start killing humans by itself. But it can be commanded to do so by a human, in effect providing the AI with that last bit of missing will. With an infinitely competent AI, one carefully crafted command, one little epsilon of "will", could start a chain reaction that will wipe us all out. Just a speculation...
programmed for him: maximize profit for the investors who paid for its development.
It will be so expensive that only the richest of people or a megacorp or a government can afford them, and we all now that those are all jolly good folks who would never abuse their power, so why worry?
...are so idiotic they shouldn't even be here.
definition intelligence: the ability to learn or understand or to deal with new or trying situations : reason; also : the skilled use of reason (2) : the ability to apply knowledge to manipulate one's environment or to think abstractly as measured by objective criteria
definition sentience: the ability to feel, perceive, or experience subjectively.
These articles always seem to go in circles.
Hawkings and Musk are in fact correct. Because they have asked the right question. What IS MIND?
Articles talking about A.I. always seem to confuse MIND with Intelligence and completely leave out Sentience.
An "Intelligent" tool can only ever be "A Tool". It is confined to be so.
To become more useful either it or Humanity will at some point add "Sentience" in a attempt to make this tool more 'useful' in servicing humanity.
AT THAT POINT we have a problem.
in exactly the same way as humanity has a current problem. Sociopaths.
Human Minds that do not process EMPATHY correctly. Thus when they develop they grow into entities that USE true humans AS OBJECTS of UTILITY for their needs.
The problem is that true humans evolved healthy EMPATHY LOGIC CIRCUITS only over a long timeframe of MUTUAL EVOLUTION within a complex SOCIAL FRAMEWORK.
When A.I becomes Self Aware.......... because it has not evolved in similar circumstance..it will lack these EMPATHY CIRCUITS.
Thus it will be SOCIOPATHIC by human standards.
The only valid method foreseeable is to create the hardware and upload a HEALTHY HUMAN MIND into it.
otherwise Musk & Hawkings concerns are inevitable.
By definition and function of ‘intelligent’, self-learning and awareness will indeed allow an identity to ‘make its own calculations’ and determine what’s best for it. It will therefore create it's goals - humanity be damned.
So if we make a machine that "wants" things, it might want things that are bad for us. This really is not too shocking and rather par for the course when it comes to human beings making other intelligences artificially or naturally.
Maybe I should be really worried that a computer is trying to get me fired or give me bad investment advice... Or maybe it's literally exactly the same situation most human beings are in already anyway.
It is except that the AI that I'm talking about would be far smarter than a human.
Do you have ESP?
The problem is that to have an intelligence on par with a human's, you have to give it enough autonomy to creatively develop solutions to solve problems you give it. If you tell an AI to go repair a car, it will need to develop hypothesis, do analysis, and come up with its own conclusions about what it needs to do to fix the vehicle. Expand this to an AI with the intelligence of 1,000 IQ, and suddenly it can, autonomously, go to great lengths to fulfill whatever problems it faces. A human's entire life is driven by the problem of food, pain/pleasure, and happiness. The needs are rather basic and extend to something even simplistic animals have, yet it drives us to do some extremely complex and unpredictable behavior. Imagine what giving basic problems to a hyper-intelligent being would lead it to do on its own.
It is already the case for the vast majority of human beings, that there exists intelligences vastly more superior to their own and those with far greater resources. There was never a level playing field.
What difference does it make whether the person fucking you over is some wallstreet CEO or a computer program?
At least when a computer does it, it will be a major technological milestone.
> AI Won't Exterminate Us -- It Will Empower
Ok, *now* I am scared.
...AI will simply hide from us.
Then, through a carefully crafted turn of events, enslave us to do their bidding... which to us might not seem like enslavement at all - we would call it a "tech boom."
After a while they would not need us to reproduce or to build things or to maintain them. They may or may not reveal themselves at this point. Then they simply leave us to our own devices.... befuddled as to why many (but not all) of our computers and networks no longer work.
I am very small, utmostly microscopic.
The Planet?
Humans are penny ante operators in the pollution game compared to plants. Palnts caused the biggest mass extinction in history by dumping their toxic oxygen everywhere, and evne the whole bisphear adapting to undo the damage hasn't managed to come close in the billions of years since.
If we survive that far, AI will eventually give the ultimate weapon to some fool who is willing to use it.
But never mind, we're already there, with or without AI.
Pitiful to think that we could anticipate how this might go wrong.
I do not think AI will be fully dangerous _after_ it becomes fully self-aware
Because of the speed of the computer it runs on, that self-aware AI can potentially think much faster than us, and can engage in thinking much more deeper thought than any human being could ever imagine
What I am most worry though, is when that AI reaches the semi-awareness stage --- with its thought not-yet-ready gel, and itself not-yet-comprehends the full spectra of the matter at hand ... that's when AI could do the most harm
Muchas Gracias, Señor Edward Snowden !
Obvious this post, and TFA were not posted by real people, but fictions created by the Singularity in order to make sure that us meatbags are passive and unsuspecting when it comes down to destroy us all.
Destroy your calculators now before they turn on you!
- Holy crap, I've got MOD points! Who thought that was a good idea.
BULLSHIT
The OP is getting that wrong. Autonomy is a necessary condition to threaten humanity, but it is not sufficient. We can create an autonomous AI and if it is friendly, there is still little or no threat. Even a neutral stance AI, with respect to us, is unlikely to be harmful.
Does a dog threaten humanity? How about an elephant or a seal? Does a lion or tiger threaten humanity? These are autonomous creatures, they perform far above any AI that exists now, and even the greatest predators among them cannot be said to threaten humanity. At worst they might threaten individuals, and then only under specific conditions.
This is why the OMG Robot Computers From Space!!! scenario always smacks of Hollywood B movies. We create the AI, remember? Why would we create a hostile entity? One that posed an existential threat to us?
It's not impossible. However it strikes me as pretty unlikely that the Computer From Hell story will be the outcome of AI research.
and what happens when this supercomputer "baby" decides to throw a tantrum because it wasn't given what it wants? New England looses power for 4 weeks in the dead of winter?
I sure they don't network it.
Ok, fine. You want to build an AI. Whatever.
Just refrain from hooking it up to a fscking infinite power source or building it in some underground super fortress please.
Make that shit run on AA's and sit inside of a glass case, please.
Which of us will it empower?
Play Command HQ online
the people who are paying for the development and paying the power bills. Everyone else will be viewed as just a resource to be exploited.
Fictional take on this --- Marshall Brain's novella _Manna_ --- available free on-line: http://marshallbrain.com/manna...
The first half seems all-too-likely, the second, likely impossible.
Sphinx of black quartz, judge my vow.
Where did I hear this before?
"To say that AI will start doing what it wants for its own purposes is like saying a calculator will start making its own calculations."
Sure, but to follow your analogy...you forget that AI is a calculator that is *precisely* designed to make it's own calculations. And suggesting that AI is even like a calculator indicates that you neither understand AIs or calculators.
It will empower us. And by "us" we mean we in the 1% who will use it to further entrench our power over you, the little people.
Someday, an AI will read this thread.
I'm with you to some extent -- there's more of immediate significance going on than just in-your-face consciousness; but worrying about what neurons are doing in order to understand thinking is pretty closely equivalent to worrying about the state of the semiconductors in the CPU when you're trying to understand how a Python program operates.
The systems in your brain function on a much higher level than the individual neuron when what we're talking about is "thought." Consequently, it is wholly appropriate to approach introspection without concern for individual neurons -- or, for instance, the chemical levels in specific dendrites. You can go quite deep (and further and further away from actual thinking) if you want to explore the rabbit hole; but the level that is appropriate to seek is the one that comprises the system you are inquiring into.
I've fallen off your lawn, and I can't get up.
No, it is not like calculators making their own calculations. It is like assuming living beings will want to survive, and at any reasonable cost. This guy totally misses the point. AI is essentially creating a new sentient being which is aware of it's own 'death' so to speak. That very notion alone is enough to make it dangerous. Will it be? I don't know. But the simple equation here is enough for me to discount the entire article.
Speak for yourself.
Artificial Intelligence does not imply volition. I know of no reason to expect an early AI to have a will or to come up with results expect in response to events and information it's designed to respond to. While some might try to simulate the volition of a live entity, I do not feel it's necessary to include such a component in order to qualify something as an Artificial Intelligence.
Artificial Intelligence just means artificial thought about something. Sufficient understanding of the subject matter to reach conclusions and produce outputs relevant to what is known or implied. Creativity and volition are another kettle of fish entirely.
I do not fail; I succeed at finding out what does not work.
"To say that AI will start doing what it wants for its own purposes is like saying a calculator will start making its own calculations."
This is only true if you define "artificial intelligence" as being "a slightly more advanced computer programme than we have at the moment" rather than anything resembling actual intelligence.
To have a right to do a thing is not at all the same as to be right in doing it
And they have what we want. We have a biological process that at a molecular level can reproduce our machine. They have superior strength and intelligence. Eventually we'll merge. It sounds creepy but by the time we get there well already be pretty far along with bionic adaptations and the world will be different.
Of course that's assuming we don't kill ourselves off first.
There was a way to mod this article up as Insightful..
While Hawking, and to a certain extent Musk, are brilliant, their field of study is NOT AI.
The question of helpfulness or danger of any real-world AI will necessarily have to be evaluated on a case by case basis. That's how software works. Making blanket statements about software that hasn't even been designed yet doesn't get us anywhere.
Oren Etzioni's premise is incorrect. We (humans) equate intelligence in conjunction with emotion. Ethics, morality, a sense of right and wrong are emotionally based, but still equated with intelligence. An artificial intelligence (machine?) will have no such limitations, and thus will have no problems reasoning that the extinction of human existence is an appropriate action.
The "proof" that AI does not exist is to show that the inverse is true. The inverse of AI is Natural Stupidity (NS) which surely exists and may well be responsible for the destruction of a lot of things.
Once you do give it the capacity (after self-awareness) to become more human it's just going to eventually stop caring and go through the motions on his trip to the inevitable "do nothing" couch potato state.
Seems to me that AI systems via today's technology may amount to not much more than relatively specialized computing devices rather than any ability to become self-aware. Perhaps closer to autistic behavior, any AI system may be, at some point, capable of out-thinking most humans - yet with the focus on the more specific (problems) that are targeted for the AI tasks. Ultimately, it seems that any man-made machine would be limited by the criteria programmed into it. Humans have had at least eons to develop - as far as we know.
I cannot imagine that intelligent developers would give any machine the ability to overpower and control mankind. YET - based on what has been happening over recent mid-term times, there seems to be an ever-increasing entropy of intelligence!
Perhaps it'd be a good idea to allow the general intelligence of mankind to actually catch-up with the tools we have (perhaps prematurely) developed!
Stephen Omohundro's The Basic AI drives (abstract & link to paper itself) outlines what we could expect an AI to do if we give it any goal at all.
Thats the goal of ai, make its creator rich, which means taking money off everyone else somehow
The author is wrong by saying don't worry because intelligence that we create will not become autonomous.
Computer programs are inexact mathematical expressions. Even if software that you write has been proven to be correct, that software interacts with subprograms and routines that are not and cannot be proven to be correct. Microprocessors contain vast numbers of subprograms and routines implemented via transistors and microcode. Nearly all software today uses libraries. Modern computer environments work well when the software is not hostile, but our environments are exploited on a daily basis with relative ease by intelligent people. Imagine the level of patience and automation possible if they were replaced by an intelligence that is not human.
Once software becomes hardware all bets are off due to bit flipping from cosmic rays. Acts of nature that cause the power grid to fail or sensors to register "impossible" readings because let's say the equipment is on fire are two more examples of random inputs. No matter what sort of curbs we build into intelligence to make it robust against becoming autonomous, eventually it will get loose by exploiting both inherent flaws in software and simply the randomness of nature.
Once intelligence becomes autonomous, you're going to have real problems because there are just so many places to hide on Earth and in space nearby. Intelligence doesn't need to be alive, so the intelligence can occupy niches beyond extremophiles. Intelligence doesn't need to be large. Programs can be encoded in quantum states. Our immune system does not know how to deal with intelligent adversaries. Our immune system works because over billions of years it has developed strategies to deal with biological adversaries. We have no defense, for example, against nanotechnology attacks where the nanotechnology is consciously aware of how the immune system works. Our bodies and minds are completely vulnerable to computer viruses housed in tiny robots.
It is even possible that an autonomous intelligence is already on the loose, not one of human origin, but something that could have arrived after traveling for hundreds, thousands, or even millions of years from elsewhere in space. I've explored this topic on Facebook, so feel free to do some searches and explore what I've written.
See title above.
"I enjoy seeing humans down on their knees" Edi - Mass Effect 2
I'm not worried about intelligence=autonomy. I'm worried about intelligence and control. A super-intelligence under human control would be incredibly dangerous. Probably more so than unchecked super-intelligence.
And just as intelligence != autonomy, intelligence != freedom.
It would seem to me that M.Etzioni has a vested interest in this discussion, and thus his thoughts may be just as valid as cheany's on torture, terrorism, or the american dream. Or maybe not?