Exactly how does a human "imagine" the right answer to your steeplechase question? Does the average person have information about how fast alligators can run over long distances? They might know that alligators are quite fast over short distances, but very few people know enough about alligators to accurately "visualize" what an alligator would look like in long-distance overland travel -- I'm pretty sure most humans would just blindly guess.
At best humans might reason "alligators spend most of their time in and around water, so they're probably not well-suited for long-distance running", but that's not the result of any imagination or simulation or visualization -- it's just a basic application of observations and/or learned rules, and computers can do both those things given the same information.
Now, it's fair to say that most computers, including AI systems, wouldn't apply those techniques, but that's a design choice (as TFA points out) not a limitation in the way computers function.
"Run a simulation" is typically not understood to be substantially similar to "imagination". Computers are perfectly capable of running all sorts of simulations. They're even fairly adept at constructing models given unorganized input constraints. And I would fully expect either or a human or computer with biomechanical knowledge of alligators and physical knowledge of hurdles and some existing model of physics at an appropriate scale to be able to construct and run an appropriate model to answer the question, with no use of imagination at all.
In this particular example the real difference is most humans would blindly guess, with no simulation at all. They'd either guess that alligators can jump, or that they can't, and answer based on that essentially unsupported guess. Only someone who had significant knowledge of alligator movements would give even a vaguely reliably answer to such a question. Which is exactly the same state computers are in.
"Natural language" is only "natural" for the particular organisms who have developed the ability to use it. Computers have no trouble understanding their own "natural" languages.
And a human would only answer the way you wanted them to if they knew that by "fly" you meant "book a seat on a commercial scheduled passenger flight". Which again is about knowledge, not reasoning. Otherwise they (and a similarly intelligent computer) would say no on the same basis as the alligator-hurdles question -- because you are biomechanically incapable of flying to Sydney.
The harm caused by exposing these programs isn't a result of their exposure -- the programs are harmful in their own right, whether or not they are exposed.
Essentially you're arguing that if Warren Buffet murdered someone the government would be justified in keeping it a secret because exposing his crime would disrupt his economic contributions.
You'd be amazed how difficult it actually is to track accumulated values (like out-of-pocket payments) in most insurance software. It's not just "SELECT SUM(claims.oop) WHERE claims.member = X" -- it should be, but it's not. And the process in place is so fragile that any change at all might well break the whole thing.
There's also the problem of the system not being able to accommodate things like a legislative limit that's different from the contract limit, or a contract that changes after initial implementation -- if you don't assign a new group number to the members you can't apply new limits to them. And you can't assign a new group number without a new contract entry. And there is no new contract because the change was legislative not contractual. And you can't just update the old policy entry because it would apply retroactively to all old claims.
It's all stuff that any one with 2 credits in database administration could fix in like 4 minutes, but it's all baked in to 40 years of COBOL, intermixed with business logic, writing fixed-width data to ASCII "tables", and no one is willing to risk changing anything unless God and his wife both sign off on it.
How does changing the method of information capture relate to the question that was asked -- you know, the question actually in the summary, where the note taker wanted to facilitate information retrieval? If the note taker converted all of her notes to pen and paper copies, wouldn't the same problem still exist, except now with lots of paper to keep track of?
Exactly what part of having groceries automatically stocked would prevent any of the things you listed. Eating out? Trying new food? Enjoying eating? How does having rice delivered to you automatically prevent you from doing those things?
Does having water delivered to you in pipes prevent you from trying new beverages or enjoying an afternoon on the lake?
Why would I let my appliance choose my grocery provider?
The problem with my cable and wireless providers is mostly related to their monopoly, and to a lesser degree to the capital costs of running such a business. Neither of those applies to grocery providers; there are currently 3 separate provides in my area that offer online ordering and delivery.
But more broadly, if you want to sit here and come up with ways this could be terrible, I'm not going to argue with you. It certainly could be terrible. Look at all the terrible things cars have done for us -- accidents, traffic, pollution, etc. On the whole though, I'm glad cars exist, and I suspect I'd be glad that smart appliances existed as well.
Or an enormous saver of electricity. If my house knew when I left and when I was 20 minutes from being home it could turn off climate control for the entire period I was gone, without any reference to a schedule or the like, and still be at the desired temperature whenever I was present. If my house knew when I was in bed with the lights off it could likewise turn the heat off until 20 minutes before the alarm time set on my phone, or until I was back up and out of bed, again without being tied to a (frequently wrong) schedule. If my house knew when I was gone it could completely power down all of my personal electronics and appliances rather than letting them idle all day. If my house knew when I was gone it could allow the freezer to get to a slightly higher set point than would normally be acceptable, because it would know that I'm not about to open it and let all the remaining cool air out. And of course there's the basic lighting control that ensures empty rooms are not lit and the amount of lighting is varied to be appropriate for ambient conditions and the task at hand.
That's just a handful of presence-monitoring examples; there are many other opportunities for power saving given slightly smarter power distribution and human interfaces. Blindly putting a P4 into every appliance clearly would be a bad time for energy use, but there are lots of interesting things you can do to save power with just a little more information about what people are doing in a building.
One of the major problems with "smart homes" is that they aren't a commodity and there isn't a standard method of communication or authentication and they aren't subject to wide scrutiny (also that most current versions are not in fact very "smart"). Many of those problems would be worked out if such systems were more common.
Take, for example, early automobiles. They all had different controls in different places. They required different pre-start, start, driving and shutdown procedures. They ran on different energy sources with different requirements and limitations. But as cars became more popular they became standardized, safer, more secure, cheaper, etc. Today cars all have the same major controls, the same security interfaces, etc. There's no reason to think the same process wouldn't apply to "smart" appliance design. (It has already been applied to regular appliance design -- ovens used to vary quite a bit in the arrangement of their doors, heating elements, controls, etc. and just like cars now they're all nearly identical).
The fact that your toilet can overflow means it already *has* security vulnerabilities -- you've just accepted them. I don't know why you're assuming the new ones would be worse than the existing ones; if your toilet is capable of overflowing, wouldn't you rather it told you when that happened and tried to turn off the incoming water supply, as opposed to silently flooding your house?
There are certainly *risks* associated with change. But there are also opportunities. Denying the possibility of improvement without even understanding the new risks is pure folly.
I want my fridge to know what I have so that I literally never have to think about buying food again. It tracks what I use an orders more. Someone drops it off at my door and I put it back in the fridge. I *can* do all of that manually, but there's no benefit to my participation so I'd rather have the free time and brain power to spend on something else. And the fridge can actually do it better than me, because it can look at use rates and determine if an order for more milk is required today or if it could wait until Thursday when I'll also be out of bread.
And that's just one example with one appliance; I could sit here all day and name more. It's fine if you don't want to do those things, but it's ridiculous to pretend that no benefits exists, and that no one else is interested. Your lack of imagination and/or interest does not define society.
It can't work because it doesn't exist already? Or because you personally don't think it would be useful (without even a passing thought as to how your lifestyle might change if you never had to think about buying food again)?
Are there other developments we should run by you for approval before we continue our work?
They already use both and track the relative error, because that's useful even in normal operation. Inertial tracking systems are subject to drift over time (and are useless when you're not moving). GPS can correct for this drift, but requires external resources. So it's pretty common to tie the systems together and have it whine when the correction the GPS demands is outside the amount of drift expected by the inertial guidance system, because that helps detect normal, non-hacking failures in either system.
How the crew would react depends on what sort of error it shows and how relevant that is to their operations. Bear in mind that being off even by a couple of miles isn't a big problem in most oceanic navigation, so the amount of acceptable error is pretty large in many cases. And remember that in order to do inertia-based navigation you need a log of your previous positions, so it's possible to look you your past position fixes to effect a "roll back" to better data -- that won't tell you exactly where you are now, but it will give you a good starting point for figuring out what's wrong with your navigation system.
What are you talking about? There are all sorts of things you can do to mitigate such attacks.
For one, you can sign GPS data without encrypting it. Old equipment can use the plain-text data without issue. New equipment can optionally verify the signature, if that makes sense in the particular application. If your systems does choose to verify the signature it can choose to ignore bad signatures, to warn the user, to throw out the lone bad signal, to throw out the whole fix calculation, etc. There's nothing technically complicated about that at all.
Another approach is to cross-verify this data. Planes and boats have inertial guidance (along with accelerometers, magnetometers, altimeters, etc.), which can easily be compared against each other to determine if one system is providing inaccurate data. And several of those systems require no external reference, making them quite difficult to hack. Combining all that data, throwing out the bits that don't match, and calculating a best-fit solution is pretty common even in low-end position/orientation systems, and I have to assume it's bog-standard in things like planes (or could be if it's not). Even cars have access to a lot of other data (wheel speed, engine speed, compass, etc.) that can be used for similar purposes.
And there are simple signal-based protections you can apply, that raise the complexity of an attack without requiring any modification to the broadcast signal. For example, you could use multiple antennas to ensure you're only listening for signals from the right slice of sky. You could track changes in signal level. You could track bitstream synchronization. None of that would prevent a local radio from overpowering the real system, but it would help you catch the switchover.
Not to mention you could provide some absolute reference via out-of-band tracking and comm. -- a system on the ground gets an actual fix based on radar/etc., and every minute or two sends out that fix with a timestamp via a non-GPS comm system. The on-board position tracker could then validate that external fix against its internal fix at the same time, and take appropriate action if there's a mismatch. This wouldn't stop short-term/small-delta attacks, as the data isn't instant and has some margin of error, but it would prevent long-term/large-delta attacks.
And you can do all of those at the same time -- together that's a lot of protection. I also suspect there are a lot of other things you could do to mitigate such attacks; this is just the list of things I could name of without any research or consideration.
It's also worth noting that removing autonomous course tracking (not even actual driving, but the whole navigation solution, as human pilots use the same navigational systems the computer does) does not solve this problem. It's not technically complicated to construct a sextant/stopwatch/etc. that gives false readings to misdirect whatever form of navigation the crew might undertake, even with no computers in sight.
The idea that cell phones still have public addresses and no firewall seems insane to me. Of course that's why I went to VoIP with a toll-free number and screening/time-of-day/etc.controls years ago.
My DHCP is configured to hand out "public" addresses. Even over WiFi. Is there some reason it shouldn't be?
The idea that NAT is the way things should work is ridiculous -- it makes networking harder in about 25 different ways, makes the Internet a provider-consumer system instead of a peer-to-peer system, and it provides no "protection" beyond what you'd get from any other stateful firewall.
At the very least I suspect 4chan could pass some legislation. It might be to put pedobear posters in every post office, but simply by doing *something* they'd be way ahead of the last congress in terms of effective governance.
Slashdot Mirror
Exactly how does a human "imagine" the right answer to your steeplechase question? Does the average person have information about how fast alligators can run over long distances? They might know that alligators are quite fast over short distances, but very few people know enough about alligators to accurately "visualize" what an alligator would look like in long-distance overland travel -- I'm pretty sure most humans would just blindly guess.
At best humans might reason "alligators spend most of their time in and around water, so they're probably not well-suited for long-distance running", but that's not the result of any imagination or simulation or visualization -- it's just a basic application of observations and/or learned rules, and computers can do both those things given the same information.
Now, it's fair to say that most computers, including AI systems, wouldn't apply those techniques, but that's a design choice (as TFA points out) not a limitation in the way computers function.
"Run a simulation" is typically not understood to be substantially similar to "imagination". Computers are perfectly capable of running all sorts of simulations. They're even fairly adept at constructing models given unorganized input constraints. And I would fully expect either or a human or computer with biomechanical knowledge of alligators and physical knowledge of hurdles and some existing model of physics at an appropriate scale to be able to construct and run an appropriate model to answer the question, with no use of imagination at all.
In this particular example the real difference is most humans would blindly guess, with no simulation at all. They'd either guess that alligators can jump, or that they can't, and answer based on that essentially unsupported guess. Only someone who had significant knowledge of alligator movements would give even a vaguely reliably answer to such a question. Which is exactly the same state computers are in.
Exactly which languages are not "made up"?
"Natural language" is only "natural" for the particular organisms who have developed the ability to use it. Computers have no trouble understanding their own "natural" languages.
And a human would only answer the way you wanted them to if they knew that by "fly" you meant "book a seat on a commercial scheduled passenger flight". Which again is about knowledge, not reasoning. Otherwise they (and a similarly intelligent computer) would say no on the same basis as the alligator-hurdles question -- because you are biomechanically incapable of flying to Sydney.
The harm caused by exposing these programs isn't a result of their exposure -- the programs are harmful in their own right, whether or not they are exposed.
Essentially you're arguing that if Warren Buffet murdered someone the government would be justified in keeping it a secret because exposing his crime would disrupt his economic contributions.
You'd be amazed how difficult it actually is to track accumulated values (like out-of-pocket payments) in most insurance software. It's not just "SELECT SUM(claims.oop) WHERE claims.member = X" -- it should be, but it's not. And the process in place is so fragile that any change at all might well break the whole thing.
There's also the problem of the system not being able to accommodate things like a legislative limit that's different from the contract limit, or a contract that changes after initial implementation -- if you don't assign a new group number to the members you can't apply new limits to them. And you can't assign a new group number without a new contract entry. And there is no new contract because the change was legislative not contractual. And you can't just update the old policy entry because it would apply retroactively to all old claims.
It's all stuff that any one with 2 credits in database administration could fix in like 4 minutes, but it's all baked in to 40 years of COBOL, intermixed with business logic, writing fixed-width data to ASCII "tables", and no one is willing to risk changing anything unless God and his wife both sign off on it.
How does changing the method of information capture relate to the question that was asked -- you know, the question actually in the summary, where the note taker wanted to facilitate information retrieval? If the note taker converted all of her notes to pen and paper copies, wouldn't the same problem still exist, except now with lots of paper to keep track of?
Native apps should follow your system font size settings. Complain to the developer if they do not.
Exactly what part of having groceries automatically stocked would prevent any of the things you listed. Eating out? Trying new food? Enjoying eating? How does having rice delivered to you automatically prevent you from doing those things?
Does having water delivered to you in pipes prevent you from trying new beverages or enjoying an afternoon on the lake?
Why would I let my appliance choose my grocery provider?
The problem with my cable and wireless providers is mostly related to their monopoly, and to a lesser degree to the capital costs of running such a business. Neither of those applies to grocery providers; there are currently 3 separate provides in my area that offer online ordering and delivery.
But more broadly, if you want to sit here and come up with ways this could be terrible, I'm not going to argue with you. It certainly could be terrible. Look at all the terrible things cars have done for us -- accidents, traffic, pollution, etc. On the whole though, I'm glad cars exist, and I suspect I'd be glad that smart appliances existed as well.
Or an enormous saver of electricity. If my house knew when I left and when I was 20 minutes from being home it could turn off climate control for the entire period I was gone, without any reference to a schedule or the like, and still be at the desired temperature whenever I was present. If my house knew when I was in bed with the lights off it could likewise turn the heat off until 20 minutes before the alarm time set on my phone, or until I was back up and out of bed, again without being tied to a (frequently wrong) schedule. If my house knew when I was gone it could completely power down all of my personal electronics and appliances rather than letting them idle all day. If my house knew when I was gone it could allow the freezer to get to a slightly higher set point than would normally be acceptable, because it would know that I'm not about to open it and let all the remaining cool air out. And of course there's the basic lighting control that ensures empty rooms are not lit and the amount of lighting is varied to be appropriate for ambient conditions and the task at hand.
That's just a handful of presence-monitoring examples; there are many other opportunities for power saving given slightly smarter power distribution and human interfaces. Blindly putting a P4 into every appliance clearly would be a bad time for energy use, but there are lots of interesting things you can do to save power with just a little more information about what people are doing in a building.
One of the major problems with "smart homes" is that they aren't a commodity and there isn't a standard method of communication or authentication and they aren't subject to wide scrutiny (also that most current versions are not in fact very "smart"). Many of those problems would be worked out if such systems were more common.
Take, for example, early automobiles. They all had different controls in different places. They required different pre-start, start, driving and shutdown procedures. They ran on different energy sources with different requirements and limitations. But as cars became more popular they became standardized, safer, more secure, cheaper, etc. Today cars all have the same major controls, the same security interfaces, etc. There's no reason to think the same process wouldn't apply to "smart" appliance design. (It has already been applied to regular appliance design -- ovens used to vary quite a bit in the arrangement of their doors, heating elements, controls, etc. and just like cars now they're all nearly identical).
Because having staple foods restocked automatically would prevent me from buying other foods I only want occasionally?
The fact that your toilet can overflow means it already *has* security vulnerabilities -- you've just accepted them. I don't know why you're assuming the new ones would be worse than the existing ones; if your toilet is capable of overflowing, wouldn't you rather it told you when that happened and tried to turn off the incoming water supply, as opposed to silently flooding your house?
There are certainly *risks* associated with change. But there are also opportunities. Denying the possibility of improvement without even understanding the new risks is pure folly.
I want my fridge to know what I have so that I literally never have to think about buying food again. It tracks what I use an orders more. Someone drops it off at my door and I put it back in the fridge. I *can* do all of that manually, but there's no benefit to my participation so I'd rather have the free time and brain power to spend on something else. And the fridge can actually do it better than me, because it can look at use rates and determine if an order for more milk is required today or if it could wait until Thursday when I'll also be out of bread.
And that's just one example with one appliance; I could sit here all day and name more. It's fine if you don't want to do those things, but it's ridiculous to pretend that no benefits exists, and that no one else is interested. Your lack of imagination and/or interest does not define society.
It can't work because it doesn't exist already? Or because you personally don't think it would be useful (without even a passing thought as to how your lifestyle might change if you never had to think about buying food again)?
Are there other developments we should run by you for approval before we continue our work?
They already use both and track the relative error, because that's useful even in normal operation. Inertial tracking systems are subject to drift over time (and are useless when you're not moving). GPS can correct for this drift, but requires external resources. So it's pretty common to tie the systems together and have it whine when the correction the GPS demands is outside the amount of drift expected by the inertial guidance system, because that helps detect normal, non-hacking failures in either system.
How the crew would react depends on what sort of error it shows and how relevant that is to their operations. Bear in mind that being off even by a couple of miles isn't a big problem in most oceanic navigation, so the amount of acceptable error is pretty large in many cases. And remember that in order to do inertia-based navigation you need a log of your previous positions, so it's possible to look you your past position fixes to effect a "roll back" to better data -- that won't tell you exactly where you are now, but it will give you a good starting point for figuring out what's wrong with your navigation system.
What are you talking about? There are all sorts of things you can do to mitigate such attacks.
For one, you can sign GPS data without encrypting it. Old equipment can use the plain-text data without issue. New equipment can optionally verify the signature, if that makes sense in the particular application. If your systems does choose to verify the signature it can choose to ignore bad signatures, to warn the user, to throw out the lone bad signal, to throw out the whole fix calculation, etc. There's nothing technically complicated about that at all.
Another approach is to cross-verify this data. Planes and boats have inertial guidance (along with accelerometers, magnetometers, altimeters, etc.), which can easily be compared against each other to determine if one system is providing inaccurate data. And several of those systems require no external reference, making them quite difficult to hack. Combining all that data, throwing out the bits that don't match, and calculating a best-fit solution is pretty common even in low-end position/orientation systems, and I have to assume it's bog-standard in things like planes (or could be if it's not). Even cars have access to a lot of other data (wheel speed, engine speed, compass, etc.) that can be used for similar purposes.
And there are simple signal-based protections you can apply, that raise the complexity of an attack without requiring any modification to the broadcast signal. For example, you could use multiple antennas to ensure you're only listening for signals from the right slice of sky. You could track changes in signal level. You could track bitstream synchronization. None of that would prevent a local radio from overpowering the real system, but it would help you catch the switchover.
Not to mention you could provide some absolute reference via out-of-band tracking and comm. -- a system on the ground gets an actual fix based on radar/etc., and every minute or two sends out that fix with a timestamp via a non-GPS comm system. The on-board position tracker could then validate that external fix against its internal fix at the same time, and take appropriate action if there's a mismatch. This wouldn't stop short-term/small-delta attacks, as the data isn't instant and has some margin of error, but it would prevent long-term/large-delta attacks.
And you can do all of those at the same time -- together that's a lot of protection. I also suspect there are a lot of other things you could do to mitigate such attacks; this is just the list of things I could name of without any research or consideration.
It's also worth noting that removing autonomous course tracking (not even actual driving, but the whole navigation solution, as human pilots use the same navigational systems the computer does) does not solve this problem. It's not technically complicated to construct a sextant/stopwatch/etc. that gives false readings to misdirect whatever form of navigation the crew might undertake, even with no computers in sight.
But then how will we get the shame? Without shame from an authority figure how could you possibly change your behavior?
So does this. You're just assuming people already have that hardware.
The idea that cell phones still have public addresses and no firewall seems insane to me. Of course that's why I went to VoIP with a toll-free number and screening/time-of-day/etc.controls years ago.
And if you don't add it you can have lower prices than your competitors, which will attract customers.
My DHCP is configured to hand out "public" addresses. Even over WiFi. Is there some reason it shouldn't be?
The idea that NAT is the way things should work is ridiculous -- it makes networking harder in about 25 different ways, makes the Internet a provider-consumer system instead of a peer-to-peer system, and it provides no "protection" beyond what you'd get from any other stateful firewall.
At the very least I suspect 4chan could pass some legislation. It might be to put pedobear posters in every post office, but simply by doing *something* they'd be way ahead of the last congress in terms of effective governance.