Catching it with a big helicopter in mid-air might actually be a really good idea. It would take a huge helicopter though - and a huge net.:)
Maybe another idea might be some kind of last second purge system to flush the fuel out of the engines. Whatever, this near miss is still already a huge success.
Exactly - great explanation.. What they need is a smaller central landing engine that can give the right amount of throttle. Otherwise I can see them ever getting a good success ratio using this method.
I like the method recently chosen for Vulcan, eject the engine cluster and recover it by parachute and inflatable landing system. Simpler and easier, but it doesn't recover the whole stage..
Come on. Its a pretty terrible job being a TSA guard, cut these guys a little slack. Surely everyone should be entitled to the odd extra grope at work... It seems particularly surly to fire them just for having a little fun while doing their job.... (at least it shows they are human and not Darleks):D
Children with spears and broomsticks against trained soldiers with aerial bombardment, tanks, machine guns, and flame throwers, etc. No wonder we were so afraid of going in. The real reason for the hurry at the end of WWII had less to do with Japan and a lot more to do with the Russians..
Also note that they know absolutely nothing about fully autonomous machines as weapons, or about Strong AI. Ie they are 100% incompetent to make an informed decision. The real joke is that such weapons don't even exist yet so how could they know so much about them ? Its like planning to fight off an alien invasion by looking at old movies..
"Autonomous cars are supposed to be better than people because they make fewer mistakes."
The actual minimum safety mark for autonomous cars is to be about 2/3 as good as the human average.. That's actually a very high level because the human average is very high and pretty close to zero crashes per unit of driving time. The thing with people is that when we are sub-par we tend to be far bellow the average, and since the machine hopefully attains a fairly constant level of safety and doesn't get tired it achieves a far higher level of cumulative safety as a result. I.e. the person drives when they feel like it and hands over to the machine if tired or drunk or too stressed or whatever.
The whole problem is that they are trying to create such a treaty.. Most of the people involved have no stake in autonomous weapons, and at least some of them are probably just trying to stop the US gaining a new military advantage. Another worse group are pure Luddites. The real problem is that they are trying to legislate future events and tech - in an area where they literally do not know what they are talking about... Like the UN on climate change - or most things in general. At its worst International Law is just another disastrous incompetent bureaucracy trying to smother our species future.
Ask India why they have developed nuclear weapons, the answer is China. Ask Taiwan or Singapore or Vietnam or the Philippines or - or Tibet.. China has a history of attacking and invading the territory of its neighbours, and its military spending is currently the fastest rising in the world. The US hasn't actually fought a full scale war since WWII and the signs at the moment are that both China and Russia are both willing to test just how far they can push the US and NATO. ISIS are savages but they are mainly a local threat to the middle east - mainly to other Islamic people. China is an immense bureaucracy, maybe the biggest risk with them is some part of the government or the military choosing to go to war without permission from the centre.. Iran has similar problems..
ISIS declared themselves a Caliphate, that curiously means they are strictly bound by Islamic commands. They are not allowed to recognize state borders and its actually a duty to eventually conquer every nation on Earth. The commandments they follow are so psychopathic that even being destroyed in a nuclear war would be something they actually want - since they believe it would help bring on the Islamic version of the Judgement Day...
In jihadi culture dying makes you a martyr so they welcome it.. That's why killing Bin Laden was such a mistake.. keeping him alive and in prison forever would have done a lot more damage to Al Qaeda and the rest of the jihadist movements.. Would have been less good on TV for the home market.. The heart of the problem isn't ISIS or Al Qaeda anyway, its the Wahhabis in Saudi Arabia who started the whole thing..
This is the exact difference between the Weak AI and Strong AI approaches. Weak AI needs complex specialized infrastructure and dependency on centralised systems. Strong AI type machines can directly replace human drivers in individual cars or trucks with no modification to the road system. Strong AI machines could even include human body robots to add additional functions like loading or unloading. Of course such robots will probably cost at least $100,000 each (or more), and even the most basic Strong AI system for a car will cost at least $20,000 to $40,000. Of course Strong AI is at least still 10 years away and certification for driving probably 20 years away. (I first made those estimates (or similar) about 20 years ago and they are still true today.:D )
Great advice. But I am writing a Strong AI. I was caught on the ass end of what felt like every ugly crazy language trick that C++ was capable of. Yes it was ugly, boy was it unintuitive, oh god.. there were single variable name calls that were more than a whole line long and had something like seven layers of indirection.
Then I realized that I was trying to build an egg using a spanner, and it didn't work. Once I realized that I needed to rewrite the heap system and core memory management things became easier because I realized that it was all just impossible. In the end the solution turned into FPGA and Verilog..
You cant imagine how much I sometimes wish for 'boring' code.. Back to the bleeding Edge.:)
Assuming the machine is based on a Strong AI type system. - A Strong AI is a dynamic real time type system, and has a lot of constraints that limit what it can do.. The underlying speed of the hardware might be very fast but its very hard to really make use of most of it. The machine uses a predictive algorithm that is normally looking over many seconds forwards and backwards in time. If a Strong AI machine gets into a disaster type situation there's probably not really a lot more it can do than a person because the probability cone is broken. Its main goals are likely to be simply bringing the vehicle to a stop and avoiding obstacles. Any priority choosing of where to go would intrinsically choose the machine and its passengers first, because the machine actually 'works' by continually optimising its own survival probability cone. If the car protects itself it intrinsically protects its passengers as well..
At first glance Weak AI is a totally different story. Here the algorithms are probably mostly fixed and pre-programmed so there is more scope for prioritising who to protect first in an accident.. The other side of this though is that the speed of a disaster situation may simply overwhelm the machines visual model leaving it effectively blind in the time it has. Again all the machine can really do in that position is try to stop as quickly as possible and maybe avoid obstacles. A big difference is that weak AI machines may be restricted to fairly slow speeds whereas the Strong AI variant can probably drive safely at higher speeds than most humans.. (50 mph verses 200 mph) Strong AI based designs are probably still 15 to 20 years in the future..
You are absolutely right about security. People just don't seem to have done the basic marketing on autonomous vehicles. - One of the biggest dangers for all autonomous vehicles - or autonomous machines in general is theft. The only solutions are that machines are kept under constant human guard - and or - are set up to defend themselves against theft. One of the biggest problems is emergency stop buttons, or 'Steal Me' buttons. - The solution for Strong AI machines is that there is no absolute stop button but there is a priority override button. The machine will only stop if it recognizes an authorized user or owner. Even there the maze gets complicated because the machine must also obey people like police or officials - but it must also recognise the difference between real and fake police. Maybe a solution is for people like police to carry special encoded radio badges, or to have the machines carry an on-board database of all authorised police & officials.
I'm sorry but that's a really stupid example. A woman deliberately walking naked into room with a bunch of male lifer prisoners would be acting as agent provocateur. A prisoner could argue that her being naked in that position is an open invitation to sex or even suicide. They might face charges, she almost definitely would - or face sectioning to a mental hospital.
The point is that its all based on context. In the end context is more about power than right or wrong. The impartial answer is that all deliberate killing is murder or that none is..
In the context of the US vs the Japanese, thousands of Japanese prisoners were summarily shot after being captured. In the later parts of the war many of the fights and battles became little more than one sided slaughters. Hiroshima was probably justifiable in pushing the Japanese towards unconditional surrender. Nagasaki was more an experiment than anything else, its military value was secondary although it actually ended the war..
We know that the Nazis might have hit New York with a nuclear bomb, by accepting that it was justified for us to use the weapon against them we are also accepting a justification for them to have used it against us.. Its about capability not morals.
I am in the middle of a Strong AI development program. We're seeking funding to go to full prototype construction.
My system prototype is not going to use quantum directly but it is partly based on quantum logic. A Strong AI based on full quantum hardware is much more difficult and still requires the building a lot of new technology -that does not exist yet- from the ground up. My system will use node level simulation and a trick to bypass the quantum element..
This is about the first time I have ever heard anyone mention 'Quantum AI' as a serious subject. I'm wondering if a project somewhere is getting close to complete and working??
"That's a mildly interesting corner case... war between belligerents being declared where?"
Mid 1980's Israeli attacks on Lebanon attacking civilian areas with bombs - bombs and planes supplied and paid for with money given by US. In a culture based on revenge and counter revenge 9/11 is just another marker among hundreds. We get them so they get us. They get us so we get them. There is no first shot. We'll (probably) still be fighting them in 50 years..
Ah so the killing of the Jews by the German Nazis was not murder either. After all the Jews were seen as enemies by the Nazi's. Makes it all totally legitimate. It also makes everything ISIS do totally legitimate as well. And Pol Pot. And 9/11.
Fascinating logic there. When we kill them its justified when they kill us its murder. You could use it to justify any crime anywhere...
Gee the funny thing is that space is expensive, who knew? If you are planning to send humans to Mars on a pittance then you need to scrape that pittance together first. They are planning to do this for about $10 billion - some of NASA's (admittedly not very efficient) plans have budgeted for over $1 trillion for getting humans to Mars. $10 billion isn't Ryan air its more like the space version of 'boat people'.. Saying all the above $10 billion is still actually a lot of money, and raising it and even spending it are not exactly easy, and cannot be done overnight. The thing that annoys me about this debate is that its the kind of thing that can actually make projects like this less likely to succeed. If you want to look at a real scam go and look at George Bush and the Iraq war - there was a trillion dollars spent.. and most of it was wasted... $1 trillion is enough to fill the pockets of every bent politician in America. (and that's a lot) Ultimate result : Saddam Hussein (Bad) --> ISIS (Worse)
The really scary thing is when you do the long term extrapolation.. In fifty years the whole world will be roughly equal - a global equal third world of poor. Democracy will be even more irrelevant than today, everyone will be poor and powerless, and a tiny super ultra-elite will rule the world and own everything..
The third world is a true capitalist society, no one gets anything they don't pay for - healthcare, school for children, social justice, police protection from crime, food or shelter on destitution. The poor will go back to being expendable, beggars, or forced to steal to live, or for the most part dead. That is the ultimate long term direction of travel and maybe the desired aim, to create a global paradise for themselves by killing most of us.. starving us out - and it will be totally legal. The Capitalist Globalist version of Pol Pot...
You've just put your finger on the real problem with modern science. The incremental approach is an evolutionary approach - hill climbing, its notorious for being blind and uncreative and for missing the most obvious solutions just outside its current set. This kind of approach is ok where everything is known and no new discoveries are likely to be made. Its also perfect in areas where the experimental 'space' is very difficult to explore and each new step is very difficult - like in semiconductor research. Where the incremental approach really fails is in areas where research and knowledge are far less than complete - nuclear fusion is a great example, rocket research is another. An even better example is my own field Strong AI, here a litany of failure guides most research in pointless circles and real progress in the university establishment has long been ground to a halt.. (both fusion & Strong AI should be far ahead of where they are) There are plenty of other areas where the same problems apply, there is still far too much specialization and not enough investigation of the knowledge in the gaps.. The problem is that true advancement requires both a rock solid basis and sound engineering but very often also the outside thinker or genius, and that is what is missing from so much of modern science.
This made me laugh. Anyone who tells you that C++ isn't complex or is easy to use, either hasn't used it or has never done anything really complicated.. Maybe C++ is still the 'best' most flexible mid-low level tool we have for doing really complex, difficult stuff. But its still a clunky obsolete nightmare that's very good at producing unreadable, leaky, crash prone code, and lets not even get started on stuff like make or include files. Just like SQL - its dead easy, of course it is, easy easy easy. But then you get a really complex query and its like trying to unravel general relativity, degree level, with the stabilizers off.. (Ok maybe I'm exaggerating, but I still remember how glad I was at Uni when I knew I wouldn't have to do any more of it..) I now work in Verilog, much easier...
Slashdot Mirror
Catching it with a big helicopter in mid-air might actually be a really good idea. It would take a huge helicopter though - and a huge net. :)
Maybe another idea might be some kind of last second purge system to flush the fuel out of the engines. Whatever, this near miss is still already a huge success.
Exactly - great explanation.. What they need is a smaller central landing engine that can give the right amount of throttle. Otherwise I can see them ever getting a good success ratio using this method.
I like the method recently chosen for Vulcan, eject the engine cluster and recover it by parachute and inflatable landing system. Simpler and easier, but it doesn't recover the whole stage..
Come on. Its a pretty terrible job being a TSA guard, cut these guys a little slack. Surely everyone should be entitled to the odd extra grope at work... :D
It seems particularly surly to fire them just for having a little fun while doing their job.... (at least it shows they are human and not Darleks)
That's an argument FOR fully autonomous weapons not against them..
Children with spears and broomsticks against trained soldiers with aerial bombardment, tanks, machine guns, and flame throwers, etc. No wonder we were so afraid of going in. The real reason for the hurry at the end of WWII had less to do with Japan and a lot more to do with the Russians..
Also note that they know absolutely nothing about fully autonomous machines as weapons, or about Strong AI. Ie they are 100% incompetent to make an informed decision. The real joke is that such weapons don't even exist yet so how could they know so much about them ?
Its like planning to fight off an alien invasion by looking at old movies..
"Autonomous cars are supposed to be better than people because they make fewer mistakes."
The actual minimum safety mark for autonomous cars is to be about 2/3 as good as the human average.. That's actually a very high level because the human average is very high and pretty close to zero crashes per unit of driving time. The thing with people is that when we are sub-par we tend to be far bellow the average, and since the machine hopefully attains a fairly constant level of safety and doesn't get tired it achieves a far higher level of cumulative safety as a result. I.e. the person drives when they feel like it and hands over to the machine if tired or drunk or too stressed or whatever.
The whole problem is that they are trying to create such a treaty.. Most of the people involved have no stake in autonomous weapons, and at least some of them are probably just trying to stop the US gaining a new military advantage. Another worse group are pure Luddites. The real problem is that they are trying to legislate future events and tech - in an area where they literally do not know what they are talking about... Like the UN on climate change - or most things in general.
At its worst International Law is just another disastrous incompetent bureaucracy trying to smother our species future.
Ask India why they have developed nuclear weapons, the answer is China. Ask Taiwan or Singapore or Vietnam or the Philippines or - or Tibet.. China has a history of attacking and invading the territory of its neighbours, and its military spending is currently the fastest rising in the world.
The US hasn't actually fought a full scale war since WWII and the signs at the moment are that both China and Russia are both willing to test just how far they can push the US and NATO. ISIS are savages but they are mainly a local threat to the middle east - mainly to other Islamic people. China is an immense bureaucracy, maybe the biggest risk with them is some part of the government or the military choosing to go to war without permission from the centre.. Iran has similar problems..
ISIS declared themselves a Caliphate, that curiously means they are strictly bound by Islamic commands. They are not allowed to recognize state borders and its actually a duty to eventually conquer every nation on Earth. The commandments they follow are so psychopathic that even being destroyed in a nuclear war would be something they actually want - since they believe it would help bring on the Islamic version of the Judgement Day...
In jihadi culture dying makes you a martyr so they welcome it.. That's why killing Bin Laden was such a mistake.. keeping him alive and in prison forever would have done a lot more damage to Al Qaeda and the rest of the jihadist movements.. Would have been less good on TV for the home market.. The heart of the problem isn't ISIS or Al Qaeda anyway, its the Wahhabis in Saudi Arabia who started the whole thing..
This is the exact difference between the Weak AI and Strong AI approaches. Weak AI needs complex specialized infrastructure and dependency on centralised systems. Strong AI type machines can directly replace human drivers in individual cars or trucks with no modification to the road system. Strong AI machines could even include human body robots to add additional functions like loading or unloading. Of course such robots will probably cost at least $100,000 each (or more), and even the most basic Strong AI system for a car will cost at least $20,000 to $40,000. Of course Strong AI is at least still 10 years away and certification for driving probably 20 years away. :D )
(I first made those estimates (or similar) about 20 years ago and they are still true today.
Great advice. But I am writing a Strong AI. I was caught on the ass end of what felt like every ugly crazy language trick that C++ was capable of. Yes it was ugly, boy was it unintuitive, oh god.. there were single variable name calls that were more than a whole line long and had something like seven layers of indirection.
Then I realized that I was trying to build an egg using a spanner, and it didn't work. Once I realized that I needed to rewrite the heap system and core memory management things became easier because I realized that it was all just impossible. In the end the solution turned into FPGA and Verilog..
You cant imagine how much I sometimes wish for 'boring' code.. Back to the bleeding Edge. :)
Assuming the machine is based on a Strong AI type system. -
A Strong AI is a dynamic real time type system, and has a lot of constraints that limit what it can do.. The underlying speed of the hardware might be very fast but its very hard to really make use of most of it. The machine uses a predictive algorithm that is normally looking over many seconds forwards and backwards in time. If a Strong AI machine gets into a disaster type situation there's probably not really a lot more it can do than a person because the probability cone is broken. Its main goals are likely to be simply bringing the vehicle to a stop and avoiding obstacles. Any priority choosing of where to go would intrinsically choose the machine and its passengers first, because the machine actually 'works' by continually optimising its own survival probability cone. If the car protects itself it intrinsically protects its passengers as well..
At first glance Weak AI is a totally different story. Here the algorithms are probably mostly fixed and pre-programmed so there is more scope for prioritising who to protect first in an accident.. The other side of this though is that the speed of a disaster situation may simply overwhelm the machines visual model leaving it effectively blind in the time it has. Again all the machine can really do in that position is try to stop as quickly as possible and maybe avoid obstacles.
A big difference is that weak AI machines may be restricted to fairly slow speeds whereas the Strong AI variant can probably drive safely at higher speeds than most humans.. (50 mph verses 200 mph) Strong AI based designs are probably still 15 to 20 years in the future..
You are absolutely right about security. People just don't seem to have done the basic marketing on autonomous vehicles. - One of the biggest dangers for all autonomous vehicles - or autonomous machines in general is theft.
The only solutions are that machines are kept under constant human guard - and or - are set up to defend themselves against theft. One of the biggest problems is emergency stop buttons, or 'Steal Me' buttons. - The solution for Strong AI machines is that there is no absolute stop button but there is a priority override button. The machine will only stop if it recognizes an authorized user or owner.
Even there the maze gets complicated because the machine must also obey people like police or officials - but it must also recognise the difference between real and fake police. Maybe a solution is for people like police to carry special encoded radio badges, or to have the machines carry an on-board database of all authorised police & officials.
I'm sorry but that's a really stupid example. A woman deliberately walking naked into room with a bunch of male lifer prisoners would be acting as agent provocateur. A prisoner could argue that her being naked in that position is an open invitation to sex or even suicide. They might face charges, she almost definitely would - or face sectioning to a mental hospital.
The point is that its all based on context. In the end context is more about power than right or wrong. The impartial answer is that all deliberate killing is murder or that none is..
In the context of the US vs the Japanese, thousands of Japanese prisoners were summarily shot after being captured. In the later parts of the war many of the fights and battles became little more than one sided slaughters.
Hiroshima was probably justifiable in pushing the Japanese towards unconditional surrender. Nagasaki was more an experiment than anything else, its military value was secondary although it actually ended the war..
We know that the Nazis might have hit New York with a nuclear bomb, by accepting that it was justified for us to use the weapon against them we are also accepting a justification for them to have used it against us.. Its about capability not morals.
I am in the middle of a Strong AI development program. We're seeking funding to go to full prototype construction.
My system prototype is not going to use quantum directly but it is partly based on quantum logic. A Strong AI based on full quantum hardware is much more difficult and still requires the building a lot of new technology -that does not exist yet- from the ground up. My system will use node level simulation and a trick to bypass the quantum element..
This is about the first time I have ever heard anyone mention 'Quantum AI' as a serious subject. I'm wondering if a project somewhere is getting close to complete and working??
Socialist here. Just pointing out that money is the lifeblood of modern civilization.. :)
"That's a mildly interesting corner case ... war between belligerents being declared where?"
Mid 1980's Israeli attacks on Lebanon attacking civilian areas with bombs - bombs and planes supplied and paid for with money given by US.
In a culture based on revenge and counter revenge 9/11 is just another marker among hundreds. We get them so they get us. They get us so we get them. There is no first shot. We'll (probably) still be fighting them in 50 years..
"No. Killing the enemy is not murder."
Ah so the killing of the Jews by the German Nazis was not murder either. After all the Jews were seen as enemies by the Nazi's. Makes it all totally legitimate. It also makes everything ISIS do totally legitimate as well. And Pol Pot. And 9/11.
Fascinating logic there. When we kill them its justified when they kill us its murder. You could use it to justify any crime anywhere...
"It is only pervs who consider images of unclothed women degrading."
It is only pervs who consider images of unclothed young men degrading. (while listening to Silverchair 'Freak')
Gee the funny thing is that space is expensive, who knew? If you are planning to send humans to Mars on a pittance then you need to scrape that pittance together first. They are planning to do this for about $10 billion - some of NASA's (admittedly not very efficient) plans have budgeted for over $1 trillion for getting humans to Mars. $10 billion isn't Ryan air its more like the space version of 'boat people'.. .. and most of it was wasted... $1 trillion is enough to fill the pockets of every bent politician in America. (and that's a lot) Ultimate result : Saddam Hussein (Bad) --> ISIS (Worse)
Saying all the above $10 billion is still actually a lot of money, and raising it and even spending it are not exactly easy, and cannot be done overnight. The thing that annoys me about this debate is that its the kind of thing that can actually make projects like this less likely to succeed. If you want to look at a real scam go and look at George Bush and the Iraq war - there was a trillion dollars spent
The really scary thing is when you do the long term extrapolation.. In fifty years the whole world will be roughly equal - a global equal third world of poor. Democracy will be even more irrelevant than today, everyone will be poor and powerless, and a tiny super ultra-elite will rule the world and own everything..
The third world is a true capitalist society, no one gets anything they don't pay for - healthcare, school for children, social justice, police protection from crime, food or shelter on destitution. The poor will go back to being expendable, beggars, or forced to steal to live, or for the most part dead. That is the ultimate long term direction of travel and maybe the desired aim, to create a global paradise for themselves by killing most of us.. starving us out - and it will be totally legal. The Capitalist Globalist version of Pol Pot...
You've just put your finger on the real problem with modern science. The incremental approach is an evolutionary approach - hill climbing, its notorious for being blind and uncreative and for missing the most obvious solutions just outside its current set. This kind of approach is ok where everything is known and no new discoveries are likely to be made. Its also perfect in areas where the experimental 'space' is very difficult to explore and each new step is very difficult - like in semiconductor research.
Where the incremental approach really fails is in areas where research and knowledge are far less than complete - nuclear fusion is a great example, rocket research is another. An even better example is my own field Strong AI, here a litany of failure guides most research in pointless circles and real progress in the university establishment has long been ground to a halt.. (both fusion & Strong AI should be far ahead of where they are)
There are plenty of other areas where the same problems apply, there is still far too much specialization and not enough investigation of the knowledge in the gaps.. The problem is that true advancement requires both a rock solid basis and sound engineering but very often also the outside thinker or genius, and that is what is missing from so much of modern science.
This made me laugh. Anyone who tells you that C++ isn't complex or is easy to use, either hasn't used it or has never done anything really complicated.. Maybe C++ is still the 'best' most flexible mid-low level tool we have for doing really complex, difficult stuff. But its still a clunky obsolete nightmare that's very good at producing unreadable, leaky, crash prone code, and lets not even get started on stuff like make or include files.
Just like SQL - its dead easy, of course it is, easy easy easy. But then you get a really complex query and its like trying to unravel general relativity, degree level, with the stabilizers off.. (Ok maybe I'm exaggerating, but I still remember how glad I was at Uni when I knew I wouldn't have to do any more of it..)
I now work in Verilog, much easier...