Indeed. I first began pondering the question after writing a simple robot program that caused a robot to drive away from obstacles. I attempted to trap the robot with some boxes but it moved in a way I didnt expect and escaped. Were it an animal, I'd think it clever.
Of course, I knew after some thought precisely why it happened, since I designed it - but it made me think, is the appearance of intelligence different than actual intelligence? One of the many human AI issues we see is that it'll never be good enough, because we tend to raise the bar higher and higher for 'intelligence' as milestones are met.
I don't like that though. I think intelligence should be a very loosely defined term. Certainly many animals have more demonstrable intelligence than say, a human infant or toddler. Anyone with a pet can say they find their pet intelligent, to an extent.
I don't think a machine will ever truly replicate a human to a satisfactory degree. There will always be some nuance of it that isn't quite natural. Sort of like those psychologists who've done studies on robot physical shapes, particularly faces. Humans tend to find them uncanny and disconcerting because they're not quite right. I think thats because in order to be human enough, you need to be human.
No, the point is that such a test does not necessarily demonstrate intelligence. Its definition of intelligence is limited to the ability to clone a particular human answerer.
You just haven't been around much. The CS program I am in (one of the largest in the US) I would say is mostly of asian, indian and middle eastern descent. Those 3 combined probably make up 2/3rds the program. At least in my experience here... I could be wrong. Women are there, though in much fewer numbers.
I can think of several gay CS majors I'm aware of, and undoubtedly there are many I am not.
I work in intelligent robotics. The turing test is nonsense. It tests if you can create a electronic clone of the current human answerer, not 'intelligence'.
Consider this, a human 4 year is intelligent by most people's measures. However, if you were to replace me with a 4 year old in a turing test, it would be obviously not myself and thus, not 'intelligent'.
Similarly, if the turing test was conducted in chinese, and you asked me to fill in the part of the computer, I would also fail it.
I don't think mechanical memory precludes it from digital computation. Many early machines that are considered digital computers used mercury delay line memory - soundwaves transmitted through a meter long tube filled with mercury.
Then there is Magnetic drum memory which was distinctly electromechanical in nature, requiring motors and physical movement of parts to work. This was found on machines until core memory became cheap enough to replace it.
Lilienthal developed gliders, but not powered aircraft (which is what the wright brothers did).
Though Lilienthal's wing designs and calculations did influence the brothers Wright. Though they actually discovered his were in fact, wrong, after having troubles getting them to work. This inspired them to build a miniature wind tunnel in their workshop to investigate wing designs and make proper measurements.
I believe the ENIAC also worked in decimal instead of binary.
All the early machines were typically created for specific purposes that only a few parts and concepts of which might still be seen in currently existing machines, but as a whole do not represent a clear lineage to what we have today. In that, I think they should all be considered firsts.
I like xilinx stuff. I know a guy who built a simple processor using one of their FPGAs, as well as a video card. Using a few more chips, it was possible to actually boot and write some simple software on the setup.
They're getting better, but its still years away from usefulness. This past week I checked out some new work that used temporal data streaming to fill in whats between the discrete frames (much more like how human vision works). This allowed much more detail to be considered, with lower noise errors that plague stereo vision traditionally.
However it still required structured light to work well, meaning outside the lab it wont work well.
One problem that I see with stereo vision research at the moment is that its still focused on turning cameras into fancy rangefinders. Though its nice to be able to determine the entire range of objects in a camera's vision area (as opposed to a singular laserbeam endpoint), it still tells you very little about the nature of those objects which is a great deal more useful than their range.
and high level nuclear waste can often be reprocessed and turned back into fuel for other reactors thus further decreasing the few kilos produced per year of actual waste.
The 'problem' is mainly one of politics. Reactors create plutonium, of which Pu-239 can fuel other reactors great. 239 can also be used in bombs though, so the worry is there regarding security and possible theft. Pu-238 is also created, and can power RTG's but not used in bombs (its too unstable, generates too much heat)
And of course the widespread ignorance that equates nuclear reactors with nuclear bombs and radioactivity with death. I wonder how people would react if they realized their smoke detectors contain a highly radioactive alpha particle emitter...
Similarly, should we be using natural gas or gasoline?
I bet the list of horrific accidents involving those two items (natural gas of course, generates much of the US' electricity) is quite a bit longer than nuclear fission's accident list.
And considering that a new nuclear power plant hasnt been built in the US in *20 years*, reactor designs have advanced considerably in the time.
The most modern of fission reactor designs are passively safe, meaning they require little to no active safety systems (ie, working cooling pumps) in order to operate.
Toshiba is working a design which requires no crew even. You build a housing, put the reactor in the ground, and in 30 years replace its core fuel element. Several of these put together can power entire towns.
but as for iRobot, the majority of the research robots in our lab (aside from the ones we built) were made by iRobot. The roomba is a curious toy from them, but most of their work is in much more advanced machines.
And I dont think electrolux has done anything particularly *new* with the trilobite, aside from giving it near-consumer level pricing. Its still using sonars and has the capabilities research robots saw 10-12 years ago. Still though, the equipment to do that used to be outrageously expensive and is now merely, expensive. That is noteworthy in and of itself.
To an extent, any individual robot model needs significant research/engineering into properly getting SLAM (simulataneous localization and mapping) working. Sometimes different robots with common elements (such as sensor/effector networks used by a manufacturer) can bypass some of this, but you still need to tune. And compromise is a big deal with limited portable computing power limits what algorithms you can use..
To build a consumer robot with a well functioning SLAM is very difficult, because of the unpredictability of environments it will have to function in. Just last week for example I was hard at work running some of our lab bots across a variety of floor surfaces to properly take into account the errors the wheel encoders suffer when running on carpet, hardwood, linoleum, concrete etc. It's nontrivial work and significant mathematical modeling - and has to be done on a individual robot model basis.
Image vaults are now available today for about $300. They contain a harddrive (often on the order of 30 gigs) and a cardreader. They're about the size of an mp3 player. Just plug your card in and dump the images. Some of them can even burn a CD.
I imagine they're trying to recoup research costs. As pointed out, it can find its charger station on its own. This requires the ability to map and localize within that map.
Doing that has been the focus (and still is!) of robotics research for almost 20 years now. Sonars are tricky to do it (I'd prefer a laser, but a decent laser scanner costs much more than the trilobyte..), but you can with a bit of computation. The hardware requirements for that probably drive up the cost too, as roomba has very little. Besides being able to find the charger base, this would also allow the trilobyte to map out a room and clean it faster and better than roomba's blind spiralling technique.
And of course, electrolux are well known makers of high end appliances. The roomba is a fairly cheap plastic device.
btw, im a researcher in an intelligent robotics lab...
The trilobyte is probably the closest thing to the autonomous research robots I work with, thats really aimed at the consumer (albeit, the wealthy consumer...). From my understanding, trilobyte uses sonars for navigation and has internal mapping and localization. Those are necessary for it to find its charger base automatically, but it also can make cleaning faster and more efficient than say, roomba's spiralling and wall following algorithms.
Its kind of high priced, but not too bad considering the amount of research that localization and mapping has had over the years, and the hardware requirements to support it. Sonars are tricky devices to use and require quite a bit of computation to do accurate mapping with.
When I was a younger, I had a few meccano sets. This was maybe 14-15 years ago though. I wasn't too impressed with them as the parts tended to come loose and were quite wobbly.
As I recall the software for the current mars rovers wasn't even complete before it was launched. They had to upload it remotely while it was in transit.
And then of course when they filled up their flash ram and crashed the computers, that required another remote software upgrade...
and then turn 90 degrees sideways and go fast enough that you fall back at the same rate the earth is falling away from you.
And have a magical engine (and inertial dampening system) that can accelerate to orbital speed fast enough and without tearing the ship apart. Actually orbital craft angle backwards as they accelerate through the atmosphere instead of going straight up.
Slashdot Mirror
Indeed. I first began pondering the question after writing a simple robot program that caused a robot to drive away from obstacles. I attempted to trap the robot with some boxes but it moved in a way I didnt expect and escaped. Were it an animal, I'd think it clever.
Of course, I knew after some thought precisely why it happened, since I designed it - but it made me think, is the appearance of intelligence different than actual intelligence? One of the many human AI issues we see is that it'll never be good enough, because we tend to raise the bar higher and higher for 'intelligence' as milestones are met.
I don't like that though. I think intelligence should be a very loosely defined term. Certainly many animals have more demonstrable intelligence than say, a human infant or toddler. Anyone with a pet can say they find their pet intelligent, to an extent.
I don't think a machine will ever truly replicate a human to a satisfactory degree. There will always be some nuance of it that isn't quite natural. Sort of like those psychologists who've done studies on robot physical shapes, particularly faces. Humans tend to find them uncanny and disconcerting because they're not quite right. I think thats because in order to be human enough, you need to be human.
No, the point is that such a test does not necessarily demonstrate intelligence. Its definition of intelligence is limited to the ability to clone a particular human answerer.
You just haven't been around much. The CS program I am in (one of the largest in the US) I would say is mostly of asian, indian and middle eastern descent. Those 3 combined probably make up 2/3rds the program. At least in my experience here... I could be wrong. Women are there, though in much fewer numbers.
I can think of several gay CS majors I'm aware of, and undoubtedly there are many I am not.
I work in intelligent robotics. The turing test is nonsense. It tests if you can create a electronic clone of the current human answerer, not 'intelligence'.
Consider this, a human 4 year is intelligent by most people's measures. However, if you were to replace me with a 4 year old in a turing test, it would be obviously not myself and thus, not 'intelligent'.
Similarly, if the turing test was conducted in chinese, and you asked me to fill in the part of the computer, I would also fail it.
I don't think mechanical memory precludes it from digital computation. Many early machines that are considered digital computers used mercury delay line memory - soundwaves transmitted through a meter long tube filled with mercury.
Then there is Magnetic drum memory which was distinctly electromechanical in nature, requiring motors and physical movement of parts to work. This was found on machines until core memory became cheap enough to replace it.
this is OT but regarding your sig...
HTML renderers do that, not slashdot. You won't see more than one space on any other page unless the text is enclosed within a pre tag.
Lilienthal developed gliders, but not powered aircraft (which is what the wright brothers did).
Though Lilienthal's wing designs and calculations did influence the brothers Wright. Though they actually discovered his were in fact, wrong, after having troubles getting them to work. This inspired them to build a miniature wind tunnel in their workshop to investigate wing designs and make proper measurements.
All the early machines were typically created for specific purposes that only a few parts and concepts of which might still be seen in currently existing machines, but as a whole do not represent a clear lineage to what we have today. In that, I think they should all be considered firsts.
I like xilinx stuff. I know a guy who built a simple processor using one of their FPGAs, as well as a video card. Using a few more chips, it was possible to actually boot and write some simple software on the setup.
They're getting better, but its still years away from usefulness. This past week I checked out some new work that used temporal data streaming to fill in whats between the discrete frames (much more like how human vision works). This allowed much more detail to be considered, with lower noise errors that plague stereo vision traditionally.
However it still required structured light to work well, meaning outside the lab it wont work well.
One problem that I see with stereo vision research at the moment is that its still focused on turning cameras into fancy rangefinders. Though its nice to be able to determine the entire range of objects in a camera's vision area (as opposed to a singular laserbeam endpoint), it still tells you very little about the nature of those objects which is a great deal more useful than their range.
cable broadband these days costs about $44.95 (in austin, for the full 3 mbps service)... so $40-50 is a really good price for a brand new service.
and high level nuclear waste can often be reprocessed and turned back into fuel for other reactors thus further decreasing the few kilos produced per year of actual waste.
The 'problem' is mainly one of politics. Reactors create plutonium, of which Pu-239 can fuel other reactors great. 239 can also be used in bombs though, so the worry is there regarding security and possible theft. Pu-238 is also created, and can power RTG's but not used in bombs (its too unstable, generates too much heat)
And of course the widespread ignorance that equates nuclear reactors with nuclear bombs and radioactivity with death. I wonder how people would react if they realized their smoke detectors contain a highly radioactive alpha particle emitter...
Similarly, should we be using natural gas or gasoline?
I bet the list of horrific accidents involving those two items (natural gas of course, generates much of the US' electricity) is quite a bit longer than nuclear fission's accident list.
And considering that a new nuclear power plant hasnt been built in the US in *20 years*, reactor designs have advanced considerably in the time.
The most modern of fission reactor designs are passively safe, meaning they require little to no active safety systems (ie, working cooling pumps) in order to operate.
r _reactr.html
Toshiba is working a design which requires no crew even. You build a housing, put the reactor in the ground, and in 30 years replace its core fuel element. Several of these put together can power entire towns.
http://www.nuclear.com/n-plants/index-Small_modul
but as for iRobot, the majority of the research robots in our lab (aside from the ones we built) were made by iRobot. The roomba is a curious toy from them, but most of their work is in much more advanced machines.
And I dont think electrolux has done anything particularly *new* with the trilobite, aside from giving it near-consumer level pricing. Its still using sonars and has the capabilities research robots saw 10-12 years ago. Still though, the equipment to do that used to be outrageously expensive and is now merely, expensive. That is noteworthy in and of itself.
and then in order to recoup the research and new manufacturing plant costs required to design and build these new components, charge $1800 for it.
To an extent, any individual robot model needs significant research/engineering into properly getting SLAM (simulataneous localization and mapping) working. Sometimes different robots with common elements (such as sensor/effector networks used by a manufacturer) can bypass some of this, but you still need to tune. And compromise is a big deal with limited portable computing power limits what algorithms you can use..
To build a consumer robot with a well functioning SLAM is very difficult, because of the unpredictability of environments it will have to function in. Just last week for example I was hard at work running some of our lab bots across a variety of floor surfaces to properly take into account the errors the wheel encoders suffer when running on carpet, hardwood, linoleum, concrete etc. It's nontrivial work and significant mathematical modeling - and has to be done on a individual robot model basis.
Image vaults are now available today for about $300. They contain a harddrive (often on the order of 30 gigs) and a cardreader. They're about the size of an mp3 player. Just plug your card in and dump the images. Some of them can even burn a CD.
here are several reviews of many varieties
I imagine they're trying to recoup research costs. As pointed out, it can find its charger station on its own. This requires the ability to map and localize within that map.
Doing that has been the focus (and still is!) of robotics research for almost 20 years now. Sonars are tricky to do it (I'd prefer a laser, but a decent laser scanner costs much more than the trilobyte..), but you can with a bit of computation. The hardware requirements for that probably drive up the cost too, as roomba has very little. Besides being able to find the charger base, this would also allow the trilobyte to map out a room and clean it faster and better than roomba's blind spiralling technique.
And of course, electrolux are well known makers of high end appliances. The roomba is a fairly cheap plastic device.
btw, im a researcher in an intelligent robotics lab...
The trilobyte is probably the closest thing to the autonomous research robots I work with, thats really aimed at the consumer (albeit, the wealthy consumer...). From my understanding, trilobyte uses sonars for navigation and has internal mapping and localization. Those are necessary for it to find its charger base automatically, but it also can make cleaning faster and more efficient than say, roomba's spiralling and wall following algorithms.
Its kind of high priced, but not too bad considering the amount of research that localization and mapping has had over the years, and the hardware requirements to support it. Sonars are tricky devices to use and require quite a bit of computation to do accurate mapping with.
Everyone knows a computer is a young woman sitting behind an adding machine.
(at least until the 60s, thats exactly what was meant by a 'computer')
When I was a younger, I had a few meccano sets. This was maybe 14-15 years ago though. I wasn't too impressed with them as the parts tended to come loose and were quite wobbly.
As I recall the software for the current mars rovers wasn't even complete before it was launched. They had to upload it remotely while it was in transit.
And then of course when they filled up their flash ram and crashed the computers, that required another remote software upgrade...
Yeah but in theory, another civilization may in fact broadcast a "here we are, anyone listening?" beacon. Thats what SETI is looking for.
The liklihood of finding a civilization by receiving some alien cell phone call is unlikely.
and then turn 90 degrees sideways and go fast enough that you fall back at the same rate the earth is falling away from you.
And have a magical engine (and inertial dampening system) that can accelerate to orbital speed fast enough and without tearing the ship apart. Actually orbital craft angle backwards as they accelerate through the atmosphere instead of going straight up.