OSDev.org

Just popping around youtube to come across titan the robot, and other android bots.

There good, but how much do you think they'll improve? And How Fast?
I personally think development have been going fast. We now have bots that can walk across any terrain, and bots that reconginise objects, which is pretty darn knarly.

I Think in the future, once they get the gist of farming body parts, there with be syborg android's which use muscles and skin instead of plastic and motors.

And what about the Ai sides of things? I think AI will be the biggest problem, as it requires technology that we simply don't understand.
"Um I'm hungry" , Why are you hungry? "Wow, that looks good!" Why does that look good?
"4+4=8" how does it equal 8?

and what about imagination? That would be cool. robots designing the cheapest, best design buildings - Learning sciences 8,000X the rate we can.
Understanding situations far more precise then we can.
Imagine on the battle field - robot's will get perfect Headshots every hit.

Technology has jumped massively with Kinect. (well at least on consumer side of things.)
We can get games consoles now that recognise are body from the environment, and can recognise who we are.

I say, were doomed when robots are designing other robots.

Anyone fancy starting Skynet industries?

I see compatibility and standards as a big problem. There's to much re-invention of the wheel going on for each project. If they where all based on the same tech and languages it would be far easier to reuse parts and get a better do-it-all robot out from the smaller parts developed.

As for AI, I think we'll get there at some point, but self-aware..? Dunno about that. More probable it will be a great faker in different areas put together to look like AI.

naf456 wrote:I Think in the future, once they get the gist of farming body parts, there with be syborg android's which use muscles and skin instead of plastic and motors.

That doesn't sound too useful, does it? We generally use automatons for tasks that people cannot perform or can only do so with difficulty. No use in reproducing the characteristics we are running away from. Well, maybe they'll make companions for humans who can't get a date.

naf456 wrote:And what about the Ai sides of things? I think AI will be the biggest problem, as it requires technology that we simply don't understand.
"Um I'm hungry" , Why are you hungry? "Wow, that looks good!" Why does that look good?
"4+4=8" how does it equal 8?

Right, but these example questions are terrible because we already have a very deep understanding for all of them. I hope you understand what makes you hungry and why 4 + 4 = 8. Arguably, the one about something looking good is not even related to intelligence; rather, human opinion.

naf456 wrote:and what about imagination? That would be cool. robots designing the cheapest, best design buildings - Learning sciences 8,000X the rate we can.

Imagination, which is mostly a random combination of past experiences, is an imperfect mechanism that humans developed for coming up with answers that are "good enough." Depending on what we find out about each complexity class (quantum, non-quantum, or of a different nature), we might learn that there are more efficient solutions to the sets of problems for which we humans use imagination.

naf456 wrote:Imagine on the battle field - robot's will get perfect Headshots every hit.

Would they? If they're so smart, shouldn't they also be able to avoid these hits (whatever the meaning of the word "hit" is in this case)?

naf456 wrote:Technology has jumped massively with Kinect. (well at least on consumer side of things.)
We can get games consoles now that recognise are body from the environment, and can recognise who we are.

Software which recognizes faces and other patterns has existed for ages. As far as motion detection goes, people have been building this sort of stuff with webcams in their homes for quite some time. Kinect is merely something designed for entertainment, nothing more.

naf456 wrote:As for AI, I think we'll get there at some point, but self-aware..? Dunno about that. More probable it will be a great faker in different areas put together to look like AI.

Consciousness, or self-awareness, is just a term we invented for a characteristic that humans manifest. We're just as "fake" as machies can be---that's the whole point of the Turing test.

The biggest obstacle, as usual is economics. There's almost always a lot of really cool things happening in some lab somewhere, but getting to market affordably is the biggest challenge. In terms of consumer usage, I think we might see a few more roomba style robots that do automated housework tasks and are (relatively) small and cheap. Tasks like mowing lawns and raking leaves comes to mind.

There will probably continue to be a strong expansion into areas less constrained by market costs (military and medicine come to mind). The US military is already making extensive use of drones and DARPA has been pouring money into getting vehicles automated so they can drive a convoy through a hotspot without worrying about an IED killing a bunch of soldiers. There's a push to get surgical robots into rural hospital here in the US as well, they'd be human operated by allow surgery in these areas without the costs and time associated with transporting the paitent or surgeon.

Other than that I think we'll continue to see more robots going into environments that humans just can't operate safely or easily in: deep sea, space, perhaps under ground mining operations. Generally in these areas it comes down to the extraction of valuable resources that justify the high per unit cost of a robot. While that doesn't sound super eco friendly, we'll also probably see the expansion of robots further into automated sorting and processing lines for the sake of recycling materials.

In general I doubt the trend of automation is going to be slowing down any time soon, but most robots will continue to be produced to do manual labor or assist human operators in doing tasks. The last estimates I saw for anything resembling a thinking machine was around 2030ish for supercomputers that should be able to accurately model the functioning of the human brain if current trends continue. There's been a lot of speculation about the human mind actually having the ability to design software complex enough to model itself, etc. But I don't think it will matter much, simply because we'll definitely have the ability to just gather raw data and piece it together. There were some scientists who basically used this approach to model depth and motion perception on a computer based off data they pulled from an insects brain has it was show a movie simulating movement through an environment. I don't know that people will just vanish or cease to be necessary though, personally I subscribe more to the idea of transhumanism where humanity will just evolve by integration with it's technology and understanding of our own biological code.

I'm sorry, I have to disagree with you Love4Boobies.

Love4Boobies wrote: That doesn't sound too useful, does it? We generally use automatons for tasks that people cannot perform or can only do so with difficulty. No use in reproducing the characteristics we are running away from. Well, maybe they'll make companions for humans who can't get a date.

I wouldn't like the streets to be filled with shiny Plastic humanoid people, and creating robots that look like humans helps them to fit in more. My Initial intention for Robots, was not to serve as slaves, that work in factories - we already have an adequate solution for that. What is my intention for robots? That I have no opinion on.

Love4Boobies wrote:Right, but these example questions are terrible because we already have a very deep understanding for all of them. I hope you understand what makes you hungry and why 4 + 4 = 8. Arguably, the one about something looking good is not even related to intelligence; rather, human opinion.

Yes, dreadful example's.
Better examples would be, How we can tell what in a lie and what is the truth? How can we tell the difference between possible and impossible? to answer these questions, we calculate the pattern of existence - this pattern being reality.
Whether something is believed to exist depends on are understanding, which we get from are experiences, from smell, touch, sight, hearing, etc...

The Next time you find an answer to an equation, think about how you got there... We got there because we learn patterns from are environment... What are numbers at the end of the day? yes, unit's which we assign nodes of a particular pattern to.
But what are numbers REALLY? where do they come from? Unlike Languages, Numbers are fundamental.

Imagination, which is mostly a random combination of past experiences, is an imperfect mechanism that humans developed for coming up with answers that are "good enough." Depending on what we find out about each complexity class (quantum, non-quantum, or of a different nature), we might learn that there are more efficient solutions to the sets of problems for which we humans use imagination.

How I feel so sorry for you, that you find imagination to be so inadequate.

Imagination is everything that turns us Remarkable, Human beings, into something more then a piece of metal that can form calculations of an equations 2.7 billion times a second.

'Your an idiot?!'
yes, I am. but I sure know that nothing in this universe so far is better then the Human Brain, And That is largely consisting of imagination.

"Friends, Romans, countrymen, lend me your ears; I come to bury Caesar, not to praise him"

tell me? What Produces such words? The out put of an algorithm? no.

When you close your eyes, and try to think of the most Beautiful place in the world, with the People you love the most. or try to conjure a Story, maybe a quest to A kingdom to bring revenge on a king , or think of the most hideous monster you can ever imagination...

Are Imagination gives use to power to visualise and emulate solutions to a problems, Lets us Create pattern sound, which changes are emotions as we capture it's data from the air.
Are Imagination allows us to build machines, that can process imagines and colours, and produce sound, which in turn can create virtual Universes which we can explore.
We can send data 3000 miles away - a second later, we have that data rendered and displayed on are TV - a solid plane of liquidised crystals that change colour through different voltages.

Frankly I am utterly outraged by your Comment. not with angry, but with laughter.

if i was a computer, I would out most reach for imagination...Then Again, I would Need imagination for that....


Your kinect statement is also invalid- or not accurate.

Yes i can create a program that takes pattern of peoples faces and recognises them .
What kinect does, is do this more accurately.
I wouldn't be able to recognise people faces 6 feet away with my webcam, as Microsoft does with kinect - the difference is like a robot walk 4 steps then falling over, and a robot running a marathon, through woodland terrain.

naf456 wrote:Better examples would be, How we can tell what in a lie and what is the truth? How can we tell the difference between possible and impossible? to answer these questions, we calculate the pattern of existence - this pattern being reality.
Whether something is believed to exist depends on are understanding, which we get from are experiences, from smell, touch, sight, hearing, etc...

Yep. But not knowing how to tackle these problems mathematically---at least not sufficiently well---isn't a dead end. For such purposes, the usual approach is to copy nature as is the case with artificial neural networks (ANNs) and machine learning algorithms. Just like humans, these approaches are prone to error and not entirely accurate, but they're good enough.

naf456 wrote:The Next time you find an answer to an equation, think about how you got there... We got there because we learn patterns from are environment... What are numbers at the end of the day? yes, unit's which we assign nodes of a particular pattern to.
But what are numbers REALLY? where do they come from? Unlike Languages, Numbers are fundamental.

Numbers are an abstract concept we've come up with, not nature. We map them onto real patterns and assign them a meaning but what they typically mean is "how many (divisions) of something are there" or an entirely different convention that we set up (see barcode scanners). E.g., when we measure velocity, we measure it in multiples of a standard unit of the same type; when we count apples, we measure in terms of how many single apples there are. This is what they really are.

naf456 wrote:

Imagination, which is mostly a random combination of past experiences, is an imperfect mechanism that humans developed for coming up with answers that are "good enough." Depending on what we find out about each complexity class (quantum, non-quantum, or of a different nature), we might learn that there are more efficient solutions to the sets of problems for which we humans use imagination.

How I feel so sorry for you, that you find imagination to be so inadequate.

I never said imagination was inadequate; it's perfectly adequate as proven by natural selection. When I say imperfect, I mean the solutions it brings forth are suboptimal, mathematically speaking---they're approximations good enough to be useful. I see you are offended by my comment when you shouldn't be. I too enjoy being a human being but that's not the point I was trying to make. In science and engineering it is a good idea to be objective and not bring our emtions in---history teaches us that just doesn't work (see how Christianity opressed science for hundreds of years). Before you answer to this part, please also read the next paragraph as it further explains what I mean.

naf456 wrote:Imagination is everything that turns us Remarkable, Human beings, into something more then a piece of metal that can form calculations of an equations 2.7 billion times a second.

'Your an idiot?!'
yes, I am. but I sure know that nothing in this universe so far is better then the Human Brain, And That is largely consisting of imagination.

Sure but I described things materialistically, as they really are, rather than from the point of view of a human being. In this framework, there's no such thing as remarkable, nor is a "piece of metal that can perform calculations" in any way superior or inferior to the human brain.

naf456 wrote:"Friends, Romans, countrymen, lend me your ears; I come to bury Caesar, not to praise him"

tell me? What Produces such words? The out put of an algorithm? no.

Humans are the system produced by their physical composition and their environment. It's truly amazing what we can do but it doesn't contradict the laws of physics. If we can do it, there's no reason to think machines can't; the most naïve method to implement this would be to just build a machine or piece of software that mimics the physical interactions that happen inside our brains.

naf456 wrote:Are Imagination gives use to power to visualise and emulate solutions to a problems, Lets us Create pattern sound, which changes are emotions as we capture it's data from the air.
Are Imagination allows us to build machines, that can process imagines and colours, and produce sound, which in turn can create virtual Universes which we can explore.
We can send data 3000 miles away - a second later, we have that data rendered and displayed on are TV - a solid plane of liquidised crystals that change colour through different voltages.

Yes. These are all imperfect solutions---this is the reason we constantly have to improve on them. That's good enough for us but remember that we are describing things materialistically.

naf456 wrote:Frankly I am utterly outraged by your Comment. not with angry, but with laughter.

That's because you didn't really understand it; sorry, maybe I should've been more verbose. Either way, next time don't be so quick to judge.

naf456 wrote:if i was a computer, I would out most reach for imagination...Then Again, I would Need imagination for that....

I guess it's time to explain imagination is suboptimal. Apart from what I've stated earlier about having to improve on our ideas (if they were perfect, they would need no improvement), it's feasible to think that a process which considers all possibilities (a problem of combinatorial nature) will find our solutions too. However, it would also find better solutions and even one or more best solutions (since all possibilities are considered). Remember what I've said about mimicking nature? Unfortunately, as far as our current abilities go, implementing such a system would not be feasible because (a) we're not certain about how certain things like quantum mechanics could accuarately be simulated as we don't fully understand it, and (b) even if we were to only consider classical mechanics, it would require vast amounts of time and enormous amounts of memory to solve more complicated problems. I've mentioned ANNs earlier---they are an abstraction of and at the level of the nervous system.

naf456 wrote:Your kinect statement is also invalid- or not accurate.

Yes i can create a program that takes pattern of peoples faces and recognises them .
What kinect does, is do this more accurately.
I wouldn't be able to recognise people faces 6 feet away with my webcam, as Microsoft does with kinect - the difference is like a robot walk 4 steps then falling over, and a robot running a marathon, through woodland terrain.

Yeah, but that's a matter of development rather than research---we already knew we could use more processing power and better cameras to do this type of work. However, the techniques have been played with by enthusiasts around the world, usually in the form of programs that do gesture or face recognition.

naf456 wrote:And what about the Ai sides of things? I think AI will be the biggest problem, as it requires technology that we simply don't understand.
"Um I'm hungry" , Why are you hungry? "Wow, that looks good!" Why does that look good?
"4+4=8" how does it equal 8?

There are two books of Roger Penrose, well known mathematician, - "The Emperor's New Mind" and "Shadows of the Mind", where he subsequently prooves the idea that modelling of intellect is impossible, from the physical and mathematical points of view. That means that AI which is close to what we estimate is unrealizable. Very interesting books, I highly recommend them for reading.

There is even another, lesser problem, that theoretically could be solved, but in practice still doesn't have an adequate solution. I mean body kinetics that looks natural. For human bodies in all animation movies it is motion capture widely used. All other creatures - dynosaurs, tigers, etc. move bad.

naf456 wrote:Better examples would be, How we can tell what in a lie and what is the truth? How can we tell the difference between possible and impossible? to answer these questions, we calculate the pattern of existence - this pattern being reality.
Whether something is believed to exist depends on are understanding, which we get from are experiences, from smell, touch, sight, hearing, etc...

I think that our consciousness (and even consciousness of other creatures, mammals) operates on the basis of internal modelling. All that we see around from the very birth to death, constantly builds up our internal models. Beginning with simplest ones - falling of the items, swinging the trees on the wind, and ending up with the most complex - mathematical theoremes, evolution of Universe, etc. But every external manifestation of things that we don't yet know slightly improve our internal models of that things. All our answers about why I'm hungry and why 4+4=8 are extrapolation of our internal models. We are capable of answering to those question for that we have a proper model and we won't understand questions for that our models are just absent. AI programs won't provide adequate answers because in reality they don't have internal models of subject being discussed at all. They just simulate sintactical analysis of common speech, nothing more. That's why I prognose that in fact the AI task even if could be theoretically solved (if we disbelieve Roger Penrose) is far beyond the capabilities of modern computational power. AI must have complicated internal models of EVERITHING around us and in scope of our communication. Especially such complicated and informal models as emotions, humor and so on.

Yoda wrote:

naf456 wrote:And what about the Ai sides of things? I think AI will be the biggest problem, as it requires technology that we simply don't understand.
"Um I'm hungry" , Why are you hungry? "Wow, that looks good!" Why does that look good?
"4+4=8" how does it equal 8?

There are two books of Roger Penrose, well known mathematician, - "The Emperor's New Mind" and "Shadows of the Mind", where he subsequently prooves the idea that modelling of intellect is impossible, from the physical and mathematical points of view. That means that AI which is close to what we estimate is unrealizable. Very interesting books, I highly recommend them for reading.

Prove is the wrong word here---he merely presents some highly controversial arguments in favour of this hypothesis so as to stimulate the reader. It's generally agreed amongst computer scientists that AI is possible in Turing machines and that quantum computers are not necessary, unlike Penrose suggests in the first book (he subsequently changed his mind about that too). Nevertheless, the books are quite entertainig.

Love4Boobies wrote:Prove is the wrong word here---he merely presents some highly controversial arguments in favour of this hypothesis so as to stimulate the reader.

Yes, you are right . Better to say tries to proove with very strong arguments, although doesn't provide 100% formal proof.

Love4Boobies wrote:It's generally agreed amongst computer scientists that AI is possible in Turing machines and that quantum computers are not necessary

Their "agreement" worth nothing unless such AI will be made in practice.

I wonder what use Fuzzy logic has in this case.

Even beyond that by what convention did we make 4+4=8 and not 6?
It is simply a bunch symbol we use to recognize the value of well something.

There is IBM's Watson, no mere human can win a game of jeopardy from that machine.

Yoda wrote:

Love4Boobies wrote:Prove is the wrong word here---he merely presents some highly controversial arguments in favour of this hypothesis so as to stimulate the reader.

Yes, you are right . Better to say tries to proove with very strong arguments, although doesn't provide 100% formal proof.

I still have the feeling that you're overhyping his theory. It's not only that he didn't provide a full formal proof, he didn't provide any sort of proof. He only presented plausible arguments with which most researchers in the field of AI disagree (I don't either but I don't think that's enough to convince anyone ). If you don't believe me, just Google around to see the huge amounts of heavy criticism.

Yoda wrote:

Love4Boobies wrote:It's generally agreed amongst computer scientists that AI is possible in Turing machines and that quantum computers are not necessary

Their "agreement" worth nothing unless such AI will be made in practice.

Agreed but you have to do your part and be patient.

CrypticalCode0 wrote:I wonder what use Fuzzy logic has in this case.

Not sure what this is in answer to.

CrypticalCode0 wrote:Even beyond that by what convention did we make 4+4=8 and not 6?

We also invented the mathematical tools to create such systems. They are just as logical but the one we are used to maps better to how we discern real-world situations.

berkus wrote:Jeopardy isn't really an intellect game, just memory, no?

It does some associations and picks up on some subtle jokes AFAIK but I haven't read anything on its implementation, nor do I know whether such information is public.

Love4Boobies wrote:Agreed but you have to do your part and be patient.

What part?
In light of my concept of "adaptive internal modelling" (and I'm sure that it is the only possible approach) of everything that surrounds us, - both physical models and abstract concepts, - this task becomes incredibly complex. Beyond capabilities of modern computational power and algorithms.
Yes, that's valuable addition - adaptive. I forgot to introduce it in my previous post. So, I don't see any even small part of this task which could be programmed.
I hear talks about AI and expert systems from 80'th and everyone states, - "right now, just a small effort, and we'll solve this! If not this year, it will be definitely solved in a couple of years!" But where all that? All researchers agreed but with no result .

I think eventually robotics will perform most low wage labour intensive jobs instead of humans. I imagine even in the long term, robots won't be conscious, be able to reason in philosophical discussions, etc. Instead I think they'll be specifically programmed and good at performing one job. For example; a farmer would buy a corn harvesting robot, and the robot only requires the technology for traversing the farm terrain, recognizing corn, and picking corn and putting it into a bucket.

Yoda wrote:

Love4Boobies wrote:Agreed but you have to do your part and be patient.

What part?
In light of my concept of "adaptive internal modelling" (and I'm sure that it is the only possible approach) of everything that surrounds us, - both physical models and abstract concepts, - this task becomes incredibly complex. Beyond capabilities of modern computational power and algorithms.
Yes, that's valuable addition - adaptive. I forgot to introduce it in my previous post. So, I don't see any even small part of this task which could be programmed.

As for adaptive, we're speaking in too general terms as to what intelligence really means. If you're talking about solving problems the way humans do, I agree. As Norbert Wiener suggested with the introduction of cybernetics, feedback is a powerful tool for automatization. This is what we've invented approaches such as genetic algorithms for.

Yoda wrote:I hear talks about AI and expert systems from 80'th and everyone states, - "right now, just a small effort, and we'll solve this! If not this year, it will be definitely solved in a couple of years!" But where all that? All researchers agreed but with no result .

Exactly, you're not being patient. Think of Fermat's last theorem---it took the world's greatest mathematicians 358 years to solve what was seemingly a simple conjecture: There's no n > 2 such that a ^ n + b ^ n = c ^ n. This lead many to believe that such a proof was impossible because it was outside of the standard axiomatic systems; but they were eventually proven to be wrong. Given how many people tried to solve the N vs. NP problem, they now hypothesize the same thing about that. So given that the AI problem is one that we don't fully understand yet, how can one hope for a solution in just a short amount of time? We will probably not live to see it but that doesn't mean it won't happen.

MessiahAndrw wrote:I think eventually robotics will perform most low wage labour intensive jobs instead of humans. I imagine even in the long term, robots won't be conscious, be able to reason in philosophical discussions, etc. Instead I think they'll be specifically programmed and good at performing one job. For example; a farmer would buy a corn harvesting robot, and the robot only requires the technology for traversing the farm terrain, recognizing corn, and picking corn and putting it into a bucket.

Depends on how long the period of time we're talking about is. No doubt that the time you speak of will come but there's no reason to think we'll stop there. If you think it's impossible, you need to provide a reason why we wouldn't at least be able to simulate the nervous system programmatically (using something similar to ANNs).

IBM's Watson is mostly a glorified search engine, with large parts of for example wikipedia stored on local disks. Obviously it has some algorithms for language understanding as well, but as seen in the videos it's not perfect by any means.

MessiahAndrw wrote:I think eventually robotics will perform most low wage labour intensive jobs instead of humans. I imagine even in the long term, robots won't be conscious, be able to reason in philosophical discussions, etc. Instead I think they'll be specifically programmed and good at performing one job. For example; a farmer would buy a corn harvesting robot, and the robot only requires the technology for traversing the farm terrain, recognizing corn, and picking corn and putting it into a bucket.

Thats basically what I believe too, but thats just because the "internal model" as yoda calls it is basically hardcoded information about for example corn and fields. No matter how much code you pour into that to make it be the perfect corn picking robot it's never going to become self aware or able to preform well on other general tasks. Thats how robotics and AI work right now, and I think it's a waste of time and resources. We'll never be able to program for every possible condition and even with self learning additions to the base code, it feels mostly like ugly hacks to me.

With a much more flexible and self learning/expanding ("adaptive internal model") aka the holy grail of algorithms - what I consider true AI with self awareness, imagination and ideas shouldn't be impossible at all. And i don't think hardware is the problem here, just that we're stuck with a lousy way of tackling the software difficulties.