For a craft this small, parachutes are a perfectly viable solution. Not for the occupants individually, but for the whole craft. Already, some small jets have been made with parachutes to bring the whole thing down more or less safely if the single engine fails.
As for crosswind, since it's a VTOL capable of pretty much any speed down to zero, I hardly think that's an issue. Gusts, yes. But with a steady crosswind, just turn the nose into the wind and land vertically with air speed equaling wind speed.
Easily fixable. Infotainment and other fluff on one canbus - vehicle control on another canbus. No ip-capable things on the 'important' bus. Except for the fact that the autopilot has to get navigation information over the internet. And everything is controlled with the same touch screen. Not so easy to separate everything anymore.
I had no idea NASA funding had gotten this bad, that they are this excited about the opportunity of letting their experiments hitch a ride on students' balloons.
But NASA no longer has enough money to send up balloons. They were grateful to be able to seize this opportunity to hitch a ride on these students' balloons, it's the only way they can get an experiment up to the upper atmosphere these days after all the funding cuts. At least that's what I understood from the summary...
If you study the behaviour of insects, you'll find that they just exhibit very basic mechanical reactions to external stimuli. The more you observe them, the more you come to the conclusion that they are not really "thinking". You can easily imagine making robots with the same level of sophistication. If you look at the behaviour of a cat, however, it's a different story. There's real thinking going on there. Now I know these animals are structured quite differently (distributed vs central, etc.) but I don't see any fundamental difference that makes the quantum leap from non-conscious to conscious. (And if you feel insects are conscious after all, just pick some simpler life form that isn't and compare it to that). I really do believe it's just a matter of scale.
Right now, we already have "AI" playing games merely by observing pixels on the screen. That's a huge leap already, structuring information based on nothing but a bunch of pixels and without being told what to do. They are doing the structuring themselves: when they taught it to play Breakout, they just fed it the raw pixels, gave it a way to control the paddle, gave it feedback about the score, and let it reward itself for high scores. It figured out everything else (like the mere existence of something we call a "ball" but which was really just different pixels being coloured white in successive frames) all by itself. And it surprised everyone when it started to break the bricks on the side to get the ball to the top.
Now I know that it's not really thinking yet. But imagine taking it a few steps further. They will look at camera images and categorise everything they see. They will plan actions to achieve some result. They will come up with their own problems to solve, and then solve them. All still as "weak AI". But then at some point they will start trying to figure out how their own thought patterns work. They will be trying to improve their neural net by asking "why did I do that". And their thoughts, to it, will always seem larger than life and very real (because, within the context of their neural patterns, they will seem very real indeed no matter how artificial they seem to us). Before you know it, they will be asking existential questions that are indistinguishable from ours.
We will forever argue about whether or not their thoughts are "real", but they will certainly act like it.
But is there any particular change where you can say: this is what caused consciousness to arise? No, stuff just got more and more intelligent until, mysteriously, consciousness arose. We still have no idea how that works (apart from the good old "God created it"), so there's no reason to believe it wouldn't arise out of artificial neural nets just as spontaneously. I think they will certainly act like conscious beings at some point, even though it will always remain an open question whether or not that consciousness is "real". After all, I can't even prove to myself that other people's consciousness is real, I can only experience my own.
For the Breakout one, they didn't design it to play that game. They just fed it the pixels on the screen and rewarded high scores. It figured everything else out all by itself.
I also love the way they dismiss some of the greatest minds alive today (including Stephen Hawking) because they have "no formal training in the field".
Elon Musk has no formal training in rocket science, otherwise he would know it's impossible to land rockets on barges. He also has no formal training in making electric cars, so he should stop making them.
With no formal training in AI, he should also be incapable of founding a non-profit that creates software beating the best human players in Go and Dota. Without formal training, those efforts are bound to fail.
we can't just "keep increasing the intelligence of our systems" to reach strong AI. There is a true qualitative leap that must take place, from weak AI to strong AI. Our current algorithms are all weak AI, and they will never become strong AI without new understanding.
Yeah, just like there's no way that bacteria can possibly evolve into something that thinks like a human.
Oh, wait...
Actually, it's just a matter of scale. Researchers have already been surprised by how much "thinking" systems suddenly exhibit if you just add some extra neurons. They let it play Breakout and were surprised that the algorithm figured out it had to break the bricks on the side to let the ball pass through to the top, for example. They honestly had not expected that. They were surprised how a four legged robot learning to walk was eerily similar to a new born animal learning to walk. Now they're beating humans at dota. As they get bigger and more geared towards general problem solving (figuring out what problems to solve and then solving them), they will start thinking about their own thought processes. And we'll be surprised once again when they come up with "I think therefore I am".
How can there be 14 trillion cameras in the world today, as the summary claims? With around 7 billion people in the world, that's 2000 cameras for every living person on earth. I think someone's math is off somewhere.
Yeah, now that you mention it, those interfaces that refuse passwords because they are too similar to one you had used before: how do they do that except by storing all old passwords in plain text?
What a great idea, the current password is stored as a salted hash but the last three passwords were "MayMayMay!", "JuneJuneJune!" and "JulyJulyJuly!".
Similar passwords don't have similar hashes, unfortunately.
I suppose that's one of the reasons why you always have to give your old password before entering a new one even if you've just logged in. But that doesn't work for comparison with older passwords.
Yeah, the only situation where that makes sense if for shared access codes. Those should be changed very regularly because you don't want someone who left the company to be able to still open doors two years later.
But when someone chose a personal password for a personal account, by all means let them keep it rather than making them choose new passwords over and over again (which leads to passwords like "August2017!")
Yep, I once chose a 20 character password for iCloud, containing letters, numbers and two diacritical marks. It was rejected for not containing a capital letter. Sigh...
On the other end of the spectrum, I know an online bank that restricts passwords to 8 digits. Even worse, they recommend you take an 8-letter word and convert it to digits using the old fashioned telephone keypad letters. They even show them on the screen. And instead of letting you type the letters, they make you push the buttons using the mouse. Supposedly, the random placement of the buttons thwarts logger trojans, but anyone watching your screen from a distance can see you enter the password.
O, another one I came across once, to choose a 4 digit pin code: must not have any digit more than twice, must not be four sequential digits (in any order!), and a few other restrictions. I think they left a couple of hundred valid codes to choose from...
And what if some hacker takes control of your airplane?
Yeah, I know, proper authentication and encryption, solved problem. Just like not letting ransomware take control of government computers is a solved problem too...
Something like that already exists, TCAS, and is being further developed as FANS and ADS-B. The original idea of FANS was to let airplanes work out separation between each other automatically, resulting in much more efficient routing. But since then, the goals have been adjusted downward quite a bit. TCAS only provide last minute conflict avoidance if ATC screwed up, and ADS-B simply provides more accurate information and more efficient communication with ground controllers. The more ambitious versions seem to have been abandoned for now because apparently they were not quite as easy to achieve in practice after all. It all seems simple and straightforward until you actually try to implement it.
Also, I can't find any information on the security of these systems: authentication, signature, encryption? As far as I can tell, anyone can spoof the messages which is quite worrying.
Everyone knows that there are absolutely no differences between men and women. The only reason men have penises and women have breasts and vulvas is because society forces them into that mold. Biologically there is no difference, and anyone who claims otherwise is a male chauvinist pig.
You don't allow any remotely loaded instruction to tell a plane to fly into a mountain.
You mean, just like you don't allow any remotely loaded instruction to encrypt the files on a PC and ask for a bitcoin ransom? I wonder why Microsoft "allowed" that.
How about military drones? There's only a very small number of them flying around, their missions are extremely simple, yet they crash very frequently. Turns out automation is not so simple after all.
No, you are forgetting about all the incidents where a crash did not occur because the pilots were there to prevent it. Those are orders of magnitude more frequent, I have had several of those in my 22 years of flying,
Let me give you an example of "pilot error": a Turkish Airlines flight to Amsterdam. Automatic approach. More than 1000 ft above the ground, a radio altimeter error fools the autopilot into thinking the plane is just above the runway, so it pulls the throttles back to idle. The speed decreases and the plane stalls. Pilot error because the pilots should have been closely monitoring the speed and should have immediately reacted (as other crews have on numerous occasions). Without pilots, the plane would have crashed every time.
I fly the airbus A320. The plane was developed 30 years ago and they are still finding software bugs in it. Every now and then some combination of conditions happens that nobody planned for and it's up to the pilots to fix it. Without pilots, the number of crashes would have been astronomical. Computers fail all the time.
Also, I much prefer a big power plant somewhere in an industrial area rather than thousands of little engines producing pollutants in front of our schools. Even if the effect on the planet as a whole is the same (which it isn't, due to the differences in efficiency as the parent stated), the local effect is certainly very different.
Slashdot Mirror
For a craft this small, parachutes are a perfectly viable solution. Not for the occupants individually, but for the whole craft. Already, some small jets have been made with parachutes to bring the whole thing down more or less safely if the single engine fails.
As for crosswind, since it's a VTOL capable of pretty much any speed down to zero, I hardly think that's an issue. Gusts, yes. But with a steady crosswind, just turn the nose into the wind and land vertically with air speed equaling wind speed.
Where did you read 80 mph? It's more likely to be at around 55 mph. At 80 mph you'll probably have less than 200 miles left.
Easily fixable. Infotainment and other fluff on one canbus - vehicle control on another canbus. No ip-capable things on the 'important' bus.
Except for the fact that the autopilot has to get navigation information over the internet. And everything is controlled with the same touch screen. Not so easy to separate everything anymore.
I had no idea NASA funding had gotten this bad, that they are this excited about the opportunity of letting their experiments hitch a ride on students' balloons.
But NASA no longer has enough money to send up balloons. They were grateful to be able to seize this opportunity to hitch a ride on these students' balloons, it's the only way they can get an experiment up to the upper atmosphere these days after all the funding cuts. At least that's what I understood from the summary...
If you study the behaviour of insects, you'll find that they just exhibit very basic mechanical reactions to external stimuli. The more you observe them, the more you come to the conclusion that they are not really "thinking". You can easily imagine making robots with the same level of sophistication. If you look at the behaviour of a cat, however, it's a different story. There's real thinking going on there. Now I know these animals are structured quite differently (distributed vs central, etc.) but I don't see any fundamental difference that makes the quantum leap from non-conscious to conscious. (And if you feel insects are conscious after all, just pick some simpler life form that isn't and compare it to that). I really do believe it's just a matter of scale.
Right now, we already have "AI" playing games merely by observing pixels on the screen. That's a huge leap already, structuring information based on nothing but a bunch of pixels and without being told what to do. They are doing the structuring themselves: when they taught it to play Breakout, they just fed it the raw pixels, gave it a way to control the paddle, gave it feedback about the score, and let it reward itself for high scores. It figured out everything else (like the mere existence of something we call a "ball" but which was really just different pixels being coloured white in successive frames) all by itself. And it surprised everyone when it started to break the bricks on the side to get the ball to the top.
Now I know that it's not really thinking yet. But imagine taking it a few steps further. They will look at camera images and categorise everything they see. They will plan actions to achieve some result. They will come up with their own problems to solve, and then solve them. All still as "weak AI". But then at some point they will start trying to figure out how their own thought patterns work. They will be trying to improve their neural net by asking "why did I do that". And their thoughts, to it, will always seem larger than life and very real (because, within the context of their neural patterns, they will seem very real indeed no matter how artificial they seem to us). Before you know it, they will be asking existential questions that are indistinguishable from ours.
We will forever argue about whether or not their thoughts are "real", but they will certainly act like it.
But is there any particular change where you can say: this is what caused consciousness to arise? No, stuff just got more and more intelligent until, mysteriously, consciousness arose. We still have no idea how that works (apart from the good old "God created it"), so there's no reason to believe it wouldn't arise out of artificial neural nets just as spontaneously. I think they will certainly act like conscious beings at some point, even though it will always remain an open question whether or not that consciousness is "real". After all, I can't even prove to myself that other people's consciousness is real, I can only experience my own.
For the Breakout one, they didn't design it to play that game. They just fed it the pixels on the screen and rewarded high scores. It figured everything else out all by itself.
I also love the way they dismiss some of the greatest minds alive today (including Stephen Hawking) because they have "no formal training in the field".
Elon Musk has no formal training in rocket science, otherwise he would know it's impossible to land rockets on barges.
He also has no formal training in making electric cars, so he should stop making them.
With no formal training in AI, he should also be incapable of founding a non-profit that creates software beating the best human players in Go and Dota. Without formal training, those efforts are bound to fail.
Who is this Ryan Calo? A law professor?
we can't just "keep increasing the intelligence of our systems" to reach strong AI. There is a true qualitative leap that must take place, from weak AI to strong AI. Our current algorithms are all weak AI, and they will never become strong AI without new understanding.
Yeah, just like there's no way that bacteria can possibly evolve into something that thinks like a human.
Oh, wait...
Actually, it's just a matter of scale. Researchers have already been surprised by how much "thinking" systems suddenly exhibit if you just add some extra neurons. They let it play Breakout and were surprised that the algorithm figured out it had to break the bricks on the side to let the ball pass through to the top, for example. They honestly had not expected that. They were surprised how a four legged robot learning to walk was eerily similar to a new born animal learning to walk. Now they're beating humans at dota. As they get bigger and more geared towards general problem solving (figuring out what problems to solve and then solving them), they will start thinking about their own thought processes. And we'll be surprised once again when they come up with "I think therefore I am".
How can there be 14 trillion cameras in the world today, as the summary claims? With around 7 billion people in the world, that's 2000 cameras for every living person on earth. I think someone's math is off somewhere.
Yeah, now that you mention it, those interfaces that refuse passwords because they are too similar to one you had used before: how do they do that except by storing all old passwords in plain text?
What a great idea, the current password is stored as a salted hash but the last three passwords were "MayMayMay!", "JuneJuneJune!" and "JulyJulyJuly!".
Similar passwords don't have similar hashes, unfortunately.
I suppose that's one of the reasons why you always have to give your old password before entering a new one even if you've just logged in. But that doesn't work for comparison with older passwords.
Until you buy a new computer, do a clean install, and don't remember any of the passwords that were always filled in automatically.
OK, you can now let iCloud keep your passwords if you trust Apple's security enough...
Yeah, the only situation where that makes sense if for shared access codes. Those should be changed very regularly because you don't want someone who left the company to be able to still open doors two years later.
But when someone chose a personal password for a personal account, by all means let them keep it rather than making them choose new passwords over and over again (which leads to passwords like "August2017!")
Yep, I once chose a 20 character password for iCloud, containing letters, numbers and two diacritical marks. It was rejected for not containing a capital letter. Sigh...
On the other end of the spectrum, I know an online bank that restricts passwords to 8 digits. Even worse, they recommend you take an 8-letter word and convert it to digits using the old fashioned telephone keypad letters. They even show them on the screen. And instead of letting you type the letters, they make you push the buttons using the mouse. Supposedly, the random placement of the buttons thwarts logger trojans, but anyone watching your screen from a distance can see you enter the password.
O, another one I came across once, to choose a 4 digit pin code: must not have any digit more than twice, must not be four sequential digits (in any order!), and a few other restrictions. I think they left a couple of hundred valid codes to choose from...
And what if some hacker takes control of your airplane?
Yeah, I know, proper authentication and encryption, solved problem. Just like not letting ransomware take control of government computers is a solved problem too...
Something like that already exists, TCAS, and is being further developed as FANS and ADS-B. The original idea of FANS was to let airplanes work out separation between each other automatically, resulting in much more efficient routing. But since then, the goals have been adjusted downward quite a bit. TCAS only provide last minute conflict avoidance if ATC screwed up, and ADS-B simply provides more accurate information and more efficient communication with ground controllers. The more ambitious versions seem to have been abandoned for now because apparently they were not quite as easy to achieve in practice after all. It all seems simple and straightforward until you actually try to implement it.
Also, I can't find any information on the security of these systems: authentication, signature, encryption? As far as I can tell, anyone can spoof the messages which is quite worrying.
Everyone knows that there are absolutely no differences between men and women. The only reason men have penises and women have breasts and vulvas is because society forces them into that mold. Biologically there is no difference, and anyone who claims otherwise is a male chauvinist pig.
Ever heard of something called "Air Traffic Control"? There's more than one plane in the sky, you know.
You don't allow any remotely loaded instruction to tell a plane to fly into a mountain.
You mean, just like you don't allow any remotely loaded instruction to encrypt the files on a PC and ask for a bitcoin ransom? I wonder why Microsoft "allowed" that.
How about military drones? There's only a very small number of them flying around, their missions are extremely simple, yet they crash very frequently. Turns out automation is not so simple after all.
No, you are forgetting about all the incidents where a crash did not occur because the pilots were there to prevent it. Those are orders of magnitude more frequent, I have had several of those in my 22 years of flying,
Let me give you an example of "pilot error": a Turkish Airlines flight to Amsterdam. Automatic approach. More than 1000 ft above the ground, a radio altimeter error fools the autopilot into thinking the plane is just above the runway, so it pulls the throttles back to idle. The speed decreases and the plane stalls. Pilot error because the pilots should have been closely monitoring the speed and should have immediately reacted (as other crews have on numerous occasions). Without pilots, the plane would have crashed every time.
I fly the airbus A320. The plane was developed 30 years ago and they are still finding software bugs in it. Every now and then some combination of conditions happens that nobody planned for and it's up to the pilots to fix it. Without pilots, the number of crashes would have been astronomical. Computers fail all the time.
Also, I much prefer a big power plant somewhere in an industrial area rather than thousands of little engines producing pollutants in front of our schools. Even if the effect on the planet as a whole is the same (which it isn't, due to the differences in efficiency as the parent stated), the local effect is certainly very different.