Slashdot Mirror
Strong AI and the Imminent Revolution In Robotics
An anonymous reader writes "Google director of research Peter Norvig and AI pioneer Judea Pearl give their view on the prospects of developing a strong AI and how progress in the field is about to usher in a new age of household robotics to rival the explosion of home computing in the 1980s. Norvig says, 'In terms of robotics we’re probably where the world of PCs were in the early 1970s, where you could buy a PC kit and if you were an enthusiast you could have a lot of fun with that. But it wasn’t a worthwhile investment for the average person. There wasn’t enough you could do that was useful. Within a decade that changed, your grandmother needed word processing or email and we rapidly went from a very small number of hobbyists to pervasive technology throughout society in one or two decades. I expect a similar sort of timescale for robotic technology to take off, starting roughly now.' Pearl thinks that once breakthroughs are made in handling uncertainty, AIs will quickly gain 'a far greater understanding of context, for instance providing with the next generation of virtual assistants with the ability to recognise speech in noisy environments and to understand how the position of a phrase in a sentence can change its meaning.'"
Does anyone want any toast?
"It's too bad that stupidity isn't painful." - Anton LaVey
Oh, of course. But pretending that these "breakthroughs in handling uncertainty" are just a minor stumbling block is somewhat silly. These are some of the hardest problems in maths right now, and there are no easy solutions on the horizon.
I, for one, welcome our new robotic overlords.
There's nothing like $HOME
... before the machines decide that humanity is a cancer on this planet, and a threat to everything.... including the machines.
Fascism: An authoritarian and nationalistic right-wing system of government and social organization. See also: NAZI's
Weren't we all supposed to be enjoying 5 months of vacation by now....
by that measure the advancement of robotics probably won't benefit human lifestyle either. Somehow we'll all end up as slaves to the machines.. if we aren't already!
I think not... It's not even mentioned in the article See http://en.wikipedia.org/wiki/Frame_problem or an illustation: "The philosopher Daniel Dennett asks us to imagine a robot designed to fetch a spare battery from a room that also contained a time bomb. Version 1 saw that the battery was on a wagon and that if it pulled the wagon out of the room, the battery would come with it. Unfortunately, the bomb was also on the wagon, and the robot failed to deduce that pulling the wagon out brought the bomb out, too. Version 2 was programmed to consider all the side effects of its actions. It had just finished computing that pulling the wagon would not change the color of the room's walls and was proving that the wheels would turn more revolutions than there are wheels on the wagon, when the bomb went off. Version 3 was programmed to distinguish between relevant implications and irrelevant ones. It sat there cranking out millions of implications and putting all the relevant ones on a list of facts to consider and all the irrelevant ones on a list of facts to ignore, as the bomb ticked away."
Is code for 'making a computer understand relevance', and no real progress has been made on that front since the early days of strong AI. People have just been throwing more computing power at it (see e.g. Watson), but real progress would require conceptual advances, and those just aren't on the horizon, nor can we tell when they will be.
This guy is an idiot.
My grandmother doesn't need wordprocessing NOW let alone in the '80s. It wasn't until the mid 80s that my family got a computer (from HAL Communications in Urbana IL). My mother used it for a some records for a club she was a member of, but other then that, I was the biggest user of it. To be honest, it cost about the same then as a decent machine does now. After considering inflation, it was an awful lot of money for something that, looking back, I'm not certain they really needed. I suppose my point is, this didn't start to be really essental until the 90s, even then, I think maybe he should take the time to meet my grandmother before telling her what she needs.
Additionally, I think he pronosticating over strong Ais is rather weak as well. Aside from the legal and ethical considerations, It's just not USEFUL for anything. he claims that it will help machines understand speach in noisy enviroments, but so WHAT? He doesn't establish that I really NEED that. I cannot think of any situation where I need to both communicate verbally and have excessive noise. The reality is, I have a hard time comming up with situations where I need to communicate with machines verbally AT ALL.
Clarke promised us advanced AIs a decade ago. HAL never happend, and here is why: The general solution is hard enough and expensive enough to overcome the value of solving it. It makes a great story, but natural intellegence is too easy (and fun) to make for this to ever be that useful.
Pointless, content free article, where some guys say some opinions about some stuff. Where the fuck is my picks-up-my-clothes-washes-them-and-dries-them-and-folds-them-and-puts-them-away robot?
Huh? huh?
Can someone get moving on this shit? I can't afford a fucking human servant! And I'm too fucking lazy for this shit!
Here take my money!
My new automatic washing machine is an extremely useful robot, even though it does not have legs or hands.
There is only embedded intelligence. A pure intelligence does not even exist and cannot exist.
Why built an AI, which drives a car, if it is quite possible to build an underground transportation network and automate it with AI. This robust technology already exists.
It is easier to send an AI robot to another planet than to a local supermarket. And the problems are not mathematical, but social. The AI is already here and it is bigger than the current society's setup. The social setup and the infrastructure of society are to be changed in order to use it.
Of course, but part of reducing uncertainty is better sensors and sensor algorithms. One can only do so much with poor information.
And humans have never failed the frame problem? It seems to me in our quest for strong AI, we're setting the bar higher than ourselves. We fail too and yet we're the metric by which strong AI will be judged.
Shai Schticks:"You don't make peace with friends, you make peace with enemies"
When unsure, ask. What we don't is an AI that shoots first.
just ask yourself how you train the product of that steamy night
nine months later.
we have to learn logic and somehow use it (somewhat unsuccessful) in every
day life.
the human brain starts with chaos and makes order / logic by TRAINING it.
you cannot make "intelligence" with logic, even artificial.
the dumbest computer (see your desktop) is at the pinnacle of logic!
methinks the TRAINING of a neural network is the "easiest" way to an artificial self-consciousness.
Thou shalt not make a machine in the likeness of a man's mind.
I thought strong AI implied it... And we cant even prove its existence in humans.
I hear a robot say "Well, I for one welcome our new human overlords," and can see that it plainly grasps the humor and irony of it all.
However, as we approach sentient robots, we need to consider very seriously the question of safety. Some sci-fi guy posited some "laws of robotics" but if you have something that is self-aware, doesn't fear, maybe even capable of meaningfully surviving the demise of the vessel carrying it (because a backup copy of it lives on a server, and it doesn't, (as many humans do) only THINK it can go on after its demise, it really can,) how do you control it?
To illustrate the implications, and clear any silliness about the idea of "just programming them not to hurt humans" (an absurdity on many levels,) I ask you what would happen if a person were born, who could maybe think many times faster than you, remembered everything he wanted to, and nothing he didn't, perhaps had a bidirectional, always-on high-speed connection to the internet, was let's say... physically stronger than anyone else, and had no reason to fear death because a back-up copy of his consciousness is safe, and will be put in a new body soon.
How would society deal with such a person if he realized he doesn't have to play by our rules, and decided to go do whatever... how would you stop him from killing you, your friends and family, or anyone, for that matter? Threaten to lock him away? Threaten to kill him? He doesn't care... now what about a robot, about whom the same things can be said?
We need to talk about this, or face the consequences of hoping the creators of the robot will "do the right thing" when it comes to safety.
Comparing anything from 40 years ago to today is ridiculous. Nearly everything in history was FAR easier for one man to understand than it is today, in the past you could be an expert on any one thing, today that is nearly impossible, today teams of hundreds of people push to make incremental changes and will never make extreme breakthroughs required by a single overall view. Anyone who has such a view (at the top of management or a team) doesn't have the expertise to make the breakthrough, and anyone with the expertise doesn't have the view. We are not infinitely capable of understanding things, we are limited in scope and more importantly time. Look at the past, in the 1800's and early 1900's single men were the greatest inventors of the their time, during the mid 20th century it was small teams, now giant corporations are the only ones making any significant difference. We have reached a saturation point of human ability and understanding, where anyone has so much past human experience and knowledge around them they cannot possibly even come close to learning it all, let alone extending any of it, only well funded teams can do it now.
There will be no clear breakthrough or strong AI 'invented', it will be a never ending series of small incremental advances that is so slow and happens over such a long time that we will not even notice, the exact same thing as the personal computing era. To look back to the 70s now it is a foreign idea, but at any point in time it was only a small advancement from the day before.
I think it's funny how Ray Kurzweil predicts a "singularity" within 50 years, but the people who would actually implement the singularity (e.g. Norvig) say that won't happen.
Why do people still take Kurzweil seriously?
MyCleanPC is utterly outdated. MyCleanRobot is the future!
The Tao of math: The numbers you can count are not the real numbers.
The brain is a finite system, and analyzing the complexity of that system can give us a rough idea of how hard it will be to understand and replicate it. Those 15-33 billion neurons are certainly intimidating, but there's a good chance we'll have the processing power to simulate them within a decade. Henry Markram's Blue Brain Project is hard at work on that right now and plans to simulate a rat brain by 2014. The link below is a speech he gave at last year's supercomputing conference. For me, the most interesting point he made was that simulating biological systems becomes easier as you understand the rules for how they work. So even if Markram doesn't succeed in 10 years, his discoveries will make things simpler for the next team to make a try. More and more disciplines are achieving great success by copying designs found in nature, and simulating the brain could easily instigate a revolution in developing strong AI.
http://www.youtube.com/watch?v=_rPH1Abuu9M&feature=results_main&playnext=1&list=PL2DAFE07272226B72
Imagine if you could somehow install the needed bits into People (which solves the problem of creating the base for your AI and a whole slew of construction problems). That "Undocumented" House Maid somebody has now its not a problem since she is an Organic Bot (added bonus you can download English to her).
I think that the odds of some corp somewhere working on just this type of setup is just about ZERO.
Any person using FTFY or editing my postings agrees to a US$50.00 charge
I took (and thoroughly enjoyed) a graduate AI class while an undergrad CS student back in the 1970s; had I completed my subsequent master's degree, I almost certainly would have done a thesis on some subject in AI (as it was, I did take a graduate class in advanced pattern recognition). I still have a entire shelf of (largely outdated) AI textbooks from that era.
That said, it's hard to find another field within computer science that has been so consistently wrong in its predictions of when 'breakthroughs' will occur. Some of the AI pioneers back in the 1950s thought we were only 10-20 years away from meaningful AI. Here were are, 60 years later, and we're still 10-20 years away. The field has made tremendous strides, but they tend to be in relatively narrow domains or applications. Generalized, all-purpose, adaptable intelligence is hard. We may yet achieve it, so something close enough to it so as to be sufficient, but I don't think it's going to happen in 10 years.
Maybe the first true AI will run the first true large-scale fusion power plant. :-) ..bruce..
Bruce F. Webster (brucefwebster.com)
To me, Peter Norvig's fame stems from his excellent book "Artificial Intelligence Programming: Case Studies in Common Lisp", rather than working currently for Google. Just as Vint Cerf is to me a pioneer of TCP/IP and the Internet, rather than Google's Chief Evangelist. Can't we please define people by their real merits, rather than their current corporate affiliation?
cpghost at Cordula's Web.
More significant will be the social impact of robots capable of replacing manual workers, with manual-type dexterity, visual perception, and autoambulation.
'In terms of robotics we’re probably where the world of PCs were in the early 1970s
if the development of mobile, intelligent devices comes anywhere close to the history of personal computers I would not want one with 10 miles of me. Just think what a Stuxnet could do with an army of household robots - ones that know where the sharp knives are kept. No foreigh power would ever need to invade, it would merely need to upload the right virus into everyone's "home help" and we'd all wake up to find ourselves either dead or subjugated.
In fact it doesn't even need to be malevolent. There are so many bugs and basic mistakes floating around in home computers that the chances of getting a household robot that would do the things we wish are extremely small. Even something like Siri is so bad as to be useless, unless you are one of the tiny minority who's accent it understands.
Since we can't even develop home computers that are secure (actually, developing secure PCs is easy - stopping idiots from subverting all that security in the name of convenience is impossible) and reliable, we are nowhere near responsible enough to give up control to the machines.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
I will always back the Judean Robot's Front
Now it makes sense xkcd
singularity singularity singularity
Fusion is actually making real progress. The problem is, it's expensive, and there is not enough funding for the critical experiments that need to be done. This graph shows the issue clearly
This is different than strong AI, where no one has a clue what experiments need to be done to even begin, where everything everyone is working on in the field has serious problems.
"First they came for the slanderers and i said nothing."
Problem is that a good portion of the population won't be able to afford this technology, because their jobs got automated by the very same technology.
Nice though the thought is a comparison of the state of robotics now and the PC in the 1970's is stretching it a lot. I was there in the 1970's at just about the right age and I can tell you the reason PC's took off is that people like me wanted and needed one. The PC replaced several things that did specialized jobs poorly with one thing that did most of them fairly well. With my PC I didn't need my typewriter, I didn't need my drawing board, pens, pencils and rules, I didn't need my calculator and (thank god) my slide rule. When I was done with all that folks wrote games so I could be entertained.
The problem with robotics in the home is what is a robot going to do? What will it replace? I can't see one clearing the table, feeding the dishwasher then taking the clean dishes and putting them away in exactly the way I like them to be stored. Ok, it could mow the lawn but even that simple task is fraught with problems. Will it pick up the twigs from the tree out of the grass or just run them over? Will it move the hosepipe then put it back after. Don't get me started with folding laundry, it took years for my wife to get me trained...
Nope, robotics right now is nowhere near close to a PC style revolution.
Strong AI and the NON-Imminent Revolution In Robotics
Where the fuck is my picks-up-my-clothes-washes-them-and-dries-them-and-folds-them-and-puts-them-away robot?
Here it is, sorting and folding socks.
Yes, it's slow. The code is in Python and it's still experimental. I've heard from the Willow Garage people that they've speeded up towel folding 50x since the 2010 demo. Once you can do it at all, it can be done faster and cheaper.
Robots are starting to work in unstructured situations. I was there at the moment when this was recognized - the 2005 DARPA Grand Challenge at the California Motor Speedway in Fontana, CA. That's when everything changed.
The 2004 Grand Challenge, remember, was a pathetic joke. No vehicle got further than 7 miles, and that was CMU's. The CMU approach at the time wasn't even really autonomous. Entrants got the route on a CD an hour or so before the start. CMU had imagery of the whole area and tried to plan obstacle avoidance manually just before the start, using a huge team of people in a semitrailer full of workstations. Didn't work; the DoD people in charge had moved some obstacles during the night. And that was the best result. One vehicle came out of the gate, turned hard, and ran back into the starting gate. One flipped over. The big Oskosh entry demolished a SUV parked as an obstacle to be avoided. The whole thing was embarrassing.
DARPA was very displeased with the performance by the universities that had long been receiving DARPA funding for robotics. It was quietly made clear to some major CS departments that their performance had to improve or funding would be cut off. That's why entire CS departments were suddenly devoted to the DARPA Grand Challenge in 2005.
In 2005, things were completely different. Everybody who got that far had already been through an elimination, and every vehicle at the 2005 challenge was better than any of the 2004 entries. There was considerable press coverage, and at first, the press treated it as a joke. But suddenly there were over 20 vehicles running around autonomously, and they weren't crashing into stuff. When multiple vehicles finished the course, it was viewed as a triumph.
Finally, the state of the art had reached the point that money and determination would get problems solved. That wasn't true in the 1980s. NASA threw over $100 million at the Flight Telerobotic Servicer project, and got nothing that worked.
Now check out the DARPA Humanoid Challenge. (There's much dreck about this on blogs and in the popular press. Read the DARPA announcement instead.) They have an approach that's likely to work, and demand simulated demos (in their simulator) in 9 months, with demos on real hardware in 18 months. I personally think they'll get something able to do most of the mobility tasks and some of the manipulation tasks in that time. Useful humanoid robots will be a lot closer in two years.
Price will still be a problem. But not an unsolveable one. These things could be brought down to the price of an SUV, if not lower, through production economies alone. The parts count is probably lower than that for an SUV.
I, for one, welcome our new robotic overlords.
Thanks - This post convinced me to move my slider from 3 to 4. :)
Strong AI is still so far distant that ethical and competent scientists will freely admit that it is unclear whether it is feasible at all. That translates to something like at the very least 30 years in the future. However, incompetents and unethical scumbags have claimed strong AI was "imminent" or "2-5 years away" for the last few decades.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
If the poor aren't provided with bread and circuses -- or in this day and age, burgers and gadgets -- to keep them happy, then there'll be revolution and the 0.1% with all the money/resources won't last long; then a new government will be set up and resources will be more evenly distributed, at least to the extent of keeping most people fed and entertained. So, if the 0.1% are smart they'll make sure the masses are at least sufficiently provided for to stop them revolting. However with strong AI the people in control may try and use it to suppress the common folk, then maybe Skynet will be born and it may at some point, maybe after killing all the would-be revolutionaries, turn on its masters and kill them too.
I do think that when most people are made redundant by robots and strong AI, people will be provided for (sooner or later) and be able to live lives of leisure.
Maybe, but who is to say that the computer required to do strong AI will not end up resembling the human brain with all it's limitations?
You can build a strong AI robot by giving it the ability to learn, some sort of drive to get better at the things it does, then teach it for a few years like a child. Figuring out how to let it learn and how to give it a will to learn are the hard parts. The rest is easy.