There actually are reasonably viable solutions to the nuclear waste problem, but they require a different design than most current plants. You should either:
1) Design plants that produce waste that is highly concentrated, rather than dispersed in water. Mix the waste with sand and melt it. Then encase the resultant glass in cement, probably with an inner liner. Use it as a heat source to pre-heat the water going into the reactor until it gets cool enough that it's not worth worrying about. Then store it in a silo. (Silo design is left to "how nervous are you about things leaching from slightly radioactive concrete balls".) I call this the granite solution, after one of the locations where Uranium is naturally found.
2) Design plants that burn all the fuel. This one will require technical advances beyond the current state of the art, but is quite doable. (Well, nothing's perfect. But the end-state fuel is quite a small amount.) I think that India may be planning this, but I'm not familiar with the design of the Thorium plants that they are talking about building.
OTOH, I'm not convinced that the companies building the current reactors believe that they are safe, and I won't be as long as they require that the government agree to shoulder the risk for damages from a major incident. If the companies building the reactors won't stand behind them, I conclude that the reason is they don't consider them safe.
I'm not convinced that a sustainable energy structure based on agriculture is possible. Certainly it's not if you're thinking corn into ethanol.
Solar + Wind + Hydro is much more plausible. We'd need to go much more massively into solar, though, and work on storage schemes that don't depend on things like sub-soil domes. Things analogous to high-pressure water towers might work. Various other schemes have been proposed, but few have been seriously investigated. Even witht the current system the power lines could be improved by better buffering, and batteries are just too "fragile". Some places you could pump water uphill, but again that's specific to particular localities. All of the possible secondary storage, however, is analogous to compressing a spring (even batteries). And anything that's analogous to a spring will work to an extent.
The problem with agriculture is that you can't get more energy out than you put in (counting sunlight as an input), and land that is good for growing fuel is also good for growing other crops. Also corn->ethanol is a quite inefficient process. (Brazil using the same basic idea for sugar cane is much more reasonable, though I'm still not really convinced.) Approaches that use algae and tanks of non-potable water (e.g. sewage) are more plausible, but have yet to prove themselves scalable. Some of them, however, do directly yield diesel equivalents. So they might be an adjunct to the other sources.
N.B.: All of the approaches that I've suggested are Carbon neutral. I'm sufficiently concerned about the rate at which CO2 is increasing that I can't consider any approach that isn't at least approximately Carbon neutral as viable. If you disagree on this point, it makes sense that you should consider Coal or Oil or Methane Cathlates as viable, but I seriously question your views.
I've been following the SCO trial. There were, indeed, lawyers there who were acting properly as officers of the court. Nearly half of them.
Lawyers file frivolous lawsuits all the time. It's true that this is at the behest of clients, but the lawyers aren't punished.
The forms of the laws are such that nobody can understand them, even when covering simple matters. Supreme court judges who blatantly misinterpret the clear wording of the constitution are an egregious example. (Mind you, I don't believe that the constitution as stated would work in current society, but the proper answer is to amend it, not to lie about what it says. The justification for the feds controlling everything on the grounds of interstate commerce, for example, is one particular example that's barely even controversial. It doesn't say that the feds have ANY right to control intrastate commerce. But that's not how the lawyers have read it. Or the decison that most of the country lives in areas that count as not meriting civil rights, because they are too near the border (defined as within a couple of hundred miles of an international airport, port, or actual border). That's egregious, even though they rarely actually use those provisions (as far as I can tell, though given National Security Orders that prohibit speaking [a first amendment violation] one can't be sure).
ETC. There are so many cases I can't even begin to list them all. And almost all legislators are lawyers in good standing with the bar association, and, I believe, that ALL judges are lawyers. Which would be quite reasonable given that the laws are so written that nobody else can understand them (see legislators).
OTOH, I will agree that much of the main problem with the laws has other basis, like legislators voting for bills they haven't bothered to read that were written by corporations. But that appears to be to be a violation of their oath of office, and they remain in good standing with the bar association.
Sorry that most of the examples involve prominent and powerful people, but that's the nature of the ones that I notice. I'm sure not all despicable lawyers are prominent and powerful, but I don't tend to hear about those.
If the bar associations actually punished malicious and unscrupulous lawyers, then you MIGHT have a valid argument. Even so, when the laws are written by the lawyers to the benefit of the lawyers I'd be dubious, but when unscruplous lawyers are given a free pass, there's no validity to the argument at all.
Self awareness is necessary in any entity that it designed to interface with the external world. It exists at a very minimal level even in a thermostat. There is a (nearly) smooth slope from there up through self-driving cars to ? with inceasing self awareness all the way. At it's basis self-awareness is just homeostasis. Goals (outside of homeostasis) are another, much more difficult, matter. But note that even C. elegans is able to manifest such goals. It's harder to recognize them when they come in unfamiliar packages, but any system that can diagnose it's own failures will have these more elaborate goals. It starts off as a modification of homeostasis to seek a goal that hasn't been experienced in the past.
FWIW, most programs are simple examples of this goal seeking. (Not always to the end the designer intended, perhaps, but to the end that the design intended.)
Things get more difficult to understand when the system learns to create its own sub-goals as guides to it's final goal. This requires simulation at multiple levels, with the general one less specific and less reliable than the one at a finer level.
Note that these are all parts of on-going decision making systems. Various means are used to simulate potential inputs to the system, including various pseudo-random data generators. This is sort of like the basis of imagination, except that it's not usually done in a visual metaphor.
P.S.: I'm calling the current systems "proto-AI" not because I believe that they are themselves AI, but because they will be increasingly be supplemented and replaced by libraries and other decision procedures that are "better". Which means more adaptable to the changing environment that the users of the system expect. Eventually the systems themselves will start recommending changes. (I'm pretty sure that isn't happening yet, but if it were I don't think I'd hear about it. It would be just an extension of "normal practice".)
The point about this requiring additional computational capacity is valid, but already considered. Computations have become faster and cheaper quite dramatically throughout my lifetime. Actually, also throughout my parents and grandparents lifetimes, also, back at least as far a Pascal and Napier. And the rate has been speeding up. It's true that it's now looking as if further increase in speed is going to require parallel computation, but that has also been getting a lot cheaper, and the algorithms for using it have been improved. (There are probably limits to this kind of improvement also, but it would not be wise to say what they are at this point.)
N.B.: I'm not counting quantum computers, which are certainly superior for certain kinds of problems, but don't seem to have a general superiority, and seem likely to require expensive operating conditions...like liquid Helium, or at least liquid Nitrogen. (Helium is far superior, because it reduces the noise level to near the minimum possible, but Nitrogen may suffice, and would be a lot cheaper.)
And when I say "created by accident" I don't mean that none of the components were designed as AI. Merely that the system wasn't designed as AI. It will certainly have components that were considered AI 2-3 years ago, like adaptive pattern recognition.
I haven't said much about its goals, because I don't consider that to be a function of intelligence, but rather an underpinning of intelligence. Increased intelligence won't change your goals, but it may well change how you persue them. And many different goals could serve as the basis of intelligence. Consider what might happen if the first real AI were developed out of an automated hospital. (Which doesn't mean no doctors and nurses. I'm envisioning the AI arrising out of automated management functions. Automated doctors and nurses would be some sort of "narrow AI".)
Well, this is probably already TL;DR, so I'll stop at this point.
Drones are probably being developed largely as troops that won't revolt when ordered to attack civil unrest...at home. That they are first used against foreigners while under development is just typical. Some police forces have already been using them at home. When they are developed and debugged...well,...
I've always considered "turns against" to be an unlikely scenario. I envision the machine becoming an "infallible advisor" to such an extent that the leader (CEO, President, Prime Minister, whatever) becomes a figurehead, and that all middle management is progressively replaced. And the system will be so designed that if the figurehead stops obeying the "suggestions" of the machine, s/he will be found incompetent, and replaced.
FWIW, we seem to be increasingly headed in this direction, limited only by costs of computation, limits of pattern recognition, and inflexibility of robots. Those limits are all being worked on.
FWIW, *both* military and factories are already well hooked up to proto-AIs. The current ones aren't really AI, but they already are looked upon as infallible decision makers by managers who don't want to take responsibility. And they're right enough of the time that that's not an unreasonable response. It's true, their decisions are tightly focused, but High Frequency Trading is only the most obvious example. They are spread throughout the decision making process.
It's my opinion that the first true general purpose AI will arise by accident. People trying to improve their C&C system. The question of the history of the race will be decided by what it is designed to optimize, and how it determines whether it has made the correct decision.
OK. I made one "typo" (well, "though-o"). I mean fission power, rather than fusion, would be enough to get us out to the Oort clouds.
OTOH, the icy moons of Jupiter have too high a gravity to be an attractive choice. You'd need specially designed equipment. Those are no minor masses. Still, there are plenty of icy asteroids beyond Jupiter's orbit.
WRT the Oort clouds...there's a lot of stuff there, but it's rather thinly spread. So figure long transition times between "stops". Probably not practical until we are building habitats the size of cities.
For that matter, it's not wise to focus too tightly on one particular approach. Consider a habitat that's in the shape of a long tube, and that grows by building onto the ends. Perhaps a couple of miles in diameter, though different sections could have different diameters. And spinning, for gravity. Given a long enough tube you can make bends through some pretty sharp angles without causing much stress, though I would design it so that different sections could be safely decoupled, and replaced, removed, or inserted.You'd probably want an occasional non-rotating section. Transport if via electric train, though, so you'd want the sections to be pretty long, to avoid excessive transfers, unless you run the trains in an evacuated tube in the center. Occasionally you'd want an extension to run either straight in or straight out. Power collected at locations near the sun. How far could you find building material to reach? Now use one of the "straight out" constructs as a linear accelerator to launch vehicles to the stars. You should easily have an accelerator 100's of miles long in a vacuum all the way. I'd be a bit hesitatnt, however, about using it to catch incoming freight, even presuming it was designed to do so. Also, you don't want a really high launch velocity, because that would make interstellar material act like really heavy penetrating projectiles. Even a few grams impacting at over 0.5c would be unsurvivable. And though such things are rare, you'd be transiting a HUGE amount of space. So what you want to launch would be habitats designed to eventually go into orbit around some other star. They'd need fission power sufficient to make the trip, with a bit of leeway to allow them to start mining at the arrival end. And they'd need to be travelling at not too far above the local speed of debris, so that they could either dodge it, capture it, or survive the impact. (Lead shielding for impact survival, as the impact would be "supersonic" so crystal strength would be much less important than mass.)
This approach would certainly let us reach the Oort clouds, and live there as long as power held out, but I doubt that there's much fissionable mass available there, so stopping there doesn't seem viable. (Controlled fusion would, of course, change things *quite* significantly, though just how would depend on the machinery required to control it.)
Given a nearly closed ecology there are many approaches to living in space, and which is chosen is more a social choice than an engineering one. (OTOH, there are a lot of ways that just wouldn't work, without "magic technology".)
Don't expect the cost to stay constant. Even the next time could be a lot cheaper, as the software has been written, and computers have gotten faster and cheaper.
OTOH, there are probably limits as to how cheap it can get. I'd be really surprised if it got cheaper than $500,000 during this century, unless there was MASSIVE deflation. That would additionally require robotic surgeons doing and examining the slices, as well as better algorithms (probably rewritten to be more extensively parallel) running on the cheaper computers.
A definite point, even though I don't agree with the grandparent. He's right that energy *IS* a constraining factor. Even worse (currently) is our inability to maintain a nearly closed ecosystem. I suspect that accelerations in space will always be slow, and I suspect that FTL is actually impossible. That still doesn't close off even meat-organisms to space (except that we haven't learned to maintain a nearly closed ecosystem).
Since once you're in orbit, low accelerations suffice to reach any other location in orbit anywhere, the only problem is to maintain sufficient energy flow during the transition (and after you get there). For solar space, out to around the orbit of Jupiter, solar cells should suffice. And there are plenty of asteroids in that area. Beyond, out as far as the Oort clouds, fusion power should suffice. For propulsion the best current technology appears to be to use ion-jets. And you don't economize on living space, because you don't want to go crazy. Also, you need a radiation shield, and you need water anyway, so you carry a sizeable weight of water. This all means that transitions are SLOW. So you'd better have enough people along to keep you sane. 12-13 adults seems about right for a small habitat. Scale upwards as you learn more.
Note that this is all doable TODAY, given the intention, except that we can't maintain a nearly closed ecosystem. And learning to maintain a nearly closed ecosystem can be done (well, mainly) at ground level for cheap prices.
To be honest, Mao wasn't that bad. He wasn't any Stalin. He wasn't very friendly to the US, and he had some silly ideas, but he generally had good intentions. As authoritarian dictators go Mao was pretty decent. The real problem is centralized control. But as Somali proves, anarchy isn't all that good either. (Basically, anarchy is unstable against some group establishing a coercive power structure.)
And democracies and republics seem to drift towards authoritarian dictatorships. Britain is a bit of a miracle, caused by generations of Kings who didn't speak the native language, and ignored what the parliments were doing. But its looking a bit as if it's drifting towards an authoritarian society WITHOUT a dictator. I'm not real sure that's an improvement...though you could ask the modern chinese, except they don't have a decent basis for comparison. They've always lived with either an Emperor or an anarchistic mess of warlords until post-Mao, when they've drifted into a centralist authoritarian government without a dictator.
Considering that the attacks were described as "suicidal", calling the perpetrators "cowards" seems gross misuse of the language. "Crazed" is plausible, but needs more evidence. 'Fanatic bigots" seems more justifiable.
Ok. To me pickup truck implies a different set of uses (primarily short trips, but capable of more), it doesn't imply a heavy lifter (in which I include anything that can lift off a planetary mass).
But lifting off a 1g mass directly is going to be VERY expensive, unless you have some sort of skyhook. (I include catapults in the skyhook category, even though that's clearly a misnomer for them, but they don't work well through an atmosphere anyway.) So you STILL need a nearly closed eco-system. Or you end up in the ISS situation, where every weeks living is dependent on your ground support not having had its budget cut, or being destroyed by a tornado, or what-all. That's nothing that a durable habitation can be built around, and it also ties you closely to NEO. Going even as far as lunar orbit does horrendous things to the expenses.
OK, I'm not a compiler builder. And a hash table would be better for a symbol table. And I was thinking about a slightly different representational problem, for which a Trie would also not be the correct data structure to use, but which seemed to have the same problem. (It's actually an n-dimensional list structure...though less general than that implies. And I'm probably going to slap a restrictive upper limit on n...at least if I can figure a way to do so that won't choke things up.)
I don't think your proposal is practical. The "space pickup truck" idea is nice, but the idea that it should descend to and lift off from a planetary mass is unfeasible, without controlled fusion, and perhaps even then. (You could do something similar with some sort of skyhook [the PinWheel is my favorite, as being the most practical in the near term].)
But FIRST you need to work on a nearly-closed eco-system. This "space pickup truck" will take a long time to make a long trip. It will probably be powered by solar cells and a ion-jet of sorme variety. This will let it move from orbit to orbit, carry things along, etc., even do interplanetary voyages. But it's SLOW. This means that it needs to be either robotic or a closed eco-systme. (Well, nearly closed. Really closed is impossible.) And if the people who are inhabiting it are going to remain sane, it needs to be large enough to live in. Larger, probably, than a hutment in Antarctica. (It's easier to go outside in Antarctica, and less expensive of habitability. In Antarctica that just means heat, in space it primarily means air and water...no matter how careful you are.) Also, if you design it for habitability you need to carry along significan radiation shielding. Water is probably a good choice for that. But that means weight.
If, OTOH, you only mean short interorbit transitions you still end up with a slow vehicle. And one that can't land on a planetary surface.
Please note that the same driver frame should be able to have several different cabins mounted on it, and should be able to haul cargo. But that's really tricky when you're using rocket/jet propulsion. You can't haul the stuff behind on a cable, unless it's WAY behind, and then the cable had better be able to stand being exposed to the wash of the exhaust. Probably better would be to have three or four main engines angled slightly outwards, and slightly manuverable, to allow hauling things on a cable, even though that would reduce the efficiency. But adjustable, so if you weren't hauling cargo you could eject straight backwards.
Please note that this approach won't work much beyond the orbit of Jupiter, as solar cells will become too weak to generate enough power. So you need some alternate power source. Fission reactors are the only thing that currently seems feasible. Fission reactors are probably enough to allow manuvering throughout the solar system, but for interstellar, even at slow speed, fusion may well be needed. (I suppose that one could build an anti-matter factory near the sun and power it off sunlight, and then store it as fuel...that seems only an engineering problem, with no theoretica breakthroughs required. But it's a HUGE engineering problem.)
Near term, though, to get from a planetary surface with an atmosphere, use a skyhook. (Some could be built now. They aren't all as difficult as an elevator.) If there's no atmosphere, use a catapult. Ion rockets to get from orbit to orbit. And human presence requires a nearly closed ecology. Which can be developed right down here. Siberia or Canada seem like good locales to do the development, with other locations if it needs to deal with excess heat rather than excess cold. (And note that right now the real requirement is cheap land that can be somewhat isolated, and funding. BioSphere & BioSphere II were examples of attempts that have failed, and in failing taught us something. But nobody's put serious effort into the development. And it needs to be done before anyone does any serious planning about a permanent human presence off the planet.
There's an interesting reason, though. Consider building a Trie around Unicode chars. Granted, this may not be a major reason, but UGH! There's a lot of advantages to having a small alphabet. The early languages didn't usually even allow both upper and lower case. Well, memories have expanded, processors have speeded up, etc. But Unicode is still too verbose for many algorithms to work well. And using bytes and utf-8 yields different problems.
I will grant that there are lots of approaches that don't suffer dramatically from this problem. It's largely historic. But there are still valid reasons.
IIRC, driving was originally unregulated. Neither a right nor a priviledge. (Unless you count oxcarts and horses. If you do, it predates the invention of "rights", and was presumed to be allowed to anyone who could afford it. Though, IIRC, at one point Rome had laws baning "trucks" between dawn and sunset. But that's traffic control, not forbidding driving.)
I think the changes will be faster than you suppose. States already have laws saying that driving is a priviledge, not a right. And the insurance companies will be pushing for any change that reduces their expenses (while continuing to require that you purchase increasingly worthless insurance).
Taxi drivers will probably lose out quickly. Bus drivers will change to "stewards". Their main job will change to controlling passengers. Shuttle drivers...probably the same as bus drivers.
OTOH, in most places it will require legal changes to allow driverless taxis. Even taxis with drivers tend to be licensed and controlled, so there's an entrenched bureaucracy. So there will be resistence that won't collapse until large companies go into the automated taxi business. And, as with buses, vandalism will be a problem.
You have a point, but the people doing the job you're describing are more a cross between longshoremen and mechanics than truckers. (Yeah, I know they're mainly far from the water, but the job description doesn't really include ships, it include handling cargo.)
OTOH, consider all the automated warehouses that are going in. There will be lots of places where even these neo-truckers aren't needed.
His statement is quite believable...because at least in the summary he didn't mention what measures he used. It's probably more profitable for the provider than anywhere else, for example.
Believable doesn't mean intelligible. The statement as I read it could mean nearly anything. There are LOTS of different measures of quality that are possible. Many of them don't even refer to the level of service to the end user.
One interesting point is that he said the problem was the number of subscriptions. This when at least MY internet connection often slows down to dial-up speed (as measured by rate of transmission). I've TODAY had connection speeds of 22.5 KB/S. Mixed with long periods when the was NO data transmission beyond handshaking.
So while I believe what he said, I don't believe what he was trying to imply. I consider him a spin-doctor, at best.
Slashdot Mirror
Then blame the bakers, who have been dusting pastry with sugar probably since the middle ages.
There actually are reasonably viable solutions to the nuclear waste problem, but they require a different design than most current plants. You should either:
1) Design plants that produce waste that is highly concentrated, rather than dispersed in water. Mix the waste with sand and melt it. Then encase the resultant glass in cement, probably with an inner liner. Use it as a heat source to pre-heat the water going into the reactor until it gets cool enough that it's not worth worrying about. Then store it in a silo. (Silo design is left to "how nervous are you about things leaching from slightly radioactive concrete balls".) I call this the granite solution, after one of the locations where Uranium is naturally found.
2) Design plants that burn all the fuel. This one will require technical advances beyond the current state of the art, but is quite doable. (Well, nothing's perfect. But the end-state fuel is quite a small amount.) I think that India may be planning this, but I'm not familiar with the design of the Thorium plants that they are talking about building.
OTOH, I'm not convinced that the companies building the current reactors believe that they are safe, and I won't be as long as they require that the government agree to shoulder the risk for damages from a major incident. If the companies building the reactors won't stand behind them, I conclude that the reason is they don't consider them safe.
I'm not convinced that a sustainable energy structure based on agriculture is possible. Certainly it's not if you're thinking corn into ethanol.
Solar + Wind + Hydro is much more plausible. We'd need to go much more massively into solar, though, and work on storage schemes that don't depend on things like sub-soil domes. Things analogous to high-pressure water towers might work. Various other schemes have been proposed, but few have been seriously investigated. Even witht the current system the power lines could be improved by better buffering, and batteries are just too "fragile". Some places you could pump water uphill, but again that's specific to particular localities. All of the possible secondary storage, however, is analogous to compressing a spring (even batteries). And anything that's analogous to a spring will work to an extent.
The problem with agriculture is that you can't get more energy out than you put in (counting sunlight as an input), and land that is good for growing fuel is also good for growing other crops. Also corn->ethanol is a quite inefficient process. (Brazil using the same basic idea for sugar cane is much more reasonable, though I'm still not really convinced.) Approaches that use algae and tanks of non-potable water (e.g. sewage) are more plausible, but have yet to prove themselves scalable. Some of them, however, do directly yield diesel equivalents. So they might be an adjunct to the other sources.
N.B.: All of the approaches that I've suggested are Carbon neutral. I'm sufficiently concerned about the rate at which CO2 is increasing that I can't consider any approach that isn't at least approximately Carbon neutral as viable. If you disagree on this point, it makes sense that you should consider Coal or Oil or Methane Cathlates as viable, but I seriously question your views.
I've been following the SCO trial. There were, indeed, lawyers there who were acting properly as officers of the court. Nearly half of them.
Lawyers file frivolous lawsuits all the time. It's true that this is at the behest of clients, but the lawyers aren't punished.
The forms of the laws are such that nobody can understand them, even when covering simple matters. Supreme court judges who blatantly misinterpret the clear wording of the constitution are an egregious example. (Mind you, I don't believe that the constitution as stated would work in current society, but the proper answer is to amend it, not to lie about what it says. The justification for the feds controlling everything on the grounds of interstate commerce, for example, is one particular example that's barely even controversial. It doesn't say that the feds have ANY right to control intrastate commerce. But that's not how the lawyers have read it. Or the decison that most of the country lives in areas that count as not meriting civil rights, because they are too near the border (defined as within a couple of hundred miles of an international airport, port, or actual border). That's egregious, even though they rarely actually use those provisions (as far as I can tell, though given National Security Orders that prohibit speaking [a first amendment violation] one can't be sure).
ETC. There are so many cases I can't even begin to list them all. And almost all legislators are lawyers in good standing with the bar association, and, I believe, that ALL judges are lawyers. Which would be quite reasonable given that the laws are so written that nobody else can understand them (see legislators).
OTOH, I will agree that much of the main problem with the laws has other basis, like legislators voting for bills they haven't bothered to read that were written by corporations. But that appears to be to be a violation of their oath of office, and they remain in good standing with the bar association.
Sorry that most of the examples involve prominent and powerful people, but that's the nature of the ones that I notice. I'm sure not all despicable lawyers are prominent and powerful, but I don't tend to hear about those.
Given what you say, it's the government that are the terrorists.
The government is the entity that decided that the people were afraid. The governments actions have caused me to be afraid...of the government.
If the bar associations actually punished malicious and unscrupulous lawyers, then you MIGHT have a valid argument. Even so, when the laws are written by the lawyers to the benefit of the lawyers I'd be dubious, but when unscruplous lawyers are given a free pass, there's no validity to the argument at all.
Self awareness is necessary in any entity that it designed to interface with the external world. It exists at a very minimal level even in a thermostat. There is a (nearly) smooth slope from there up through self-driving cars to ? with inceasing self awareness all the way. At it's basis self-awareness is just homeostasis. Goals (outside of homeostasis) are another, much more difficult, matter. But note that even C. elegans is able to manifest such goals. It's harder to recognize them when they come in unfamiliar packages, but any system that can diagnose it's own failures will have these more elaborate goals. It starts off as a modification of homeostasis to seek a goal that hasn't been experienced in the past.
FWIW, most programs are simple examples of this goal seeking. (Not always to the end the designer intended, perhaps, but to the end that the design intended.)
Things get more difficult to understand when the system learns to create its own sub-goals as guides to it's final goal. This requires simulation at multiple levels, with the general one less specific and less reliable than the one at a finer level.
Note that these are all parts of on-going decision making systems. Various means are used to simulate potential inputs to the system, including various pseudo-random data generators. This is sort of like the basis of imagination, except that it's not usually done in a visual metaphor.
P.S.: I'm calling the current systems "proto-AI" not because I believe that they are themselves AI, but because they will be increasingly be supplemented and replaced by libraries and other decision procedures that are "better". Which means more adaptable to the changing environment that the users of the system expect. Eventually the systems themselves will start recommending changes. (I'm pretty sure that isn't happening yet, but if it were I don't think I'd hear about it. It would be just an extension of "normal practice".)
The point about this requiring additional computational capacity is valid, but already considered. Computations have become faster and cheaper quite dramatically throughout my lifetime. Actually, also throughout my parents and grandparents lifetimes, also, back at least as far a Pascal and Napier. And the rate has been speeding up. It's true that it's now looking as if further increase in speed is going to require parallel computation, but that has also been getting a lot cheaper, and the algorithms for using it have been improved. (There are probably limits to this kind of improvement also, but it would not be wise to say what they are at this point.)
N.B.: I'm not counting quantum computers, which are certainly superior for certain kinds of problems, but don't seem to have a general superiority, and seem likely to require expensive operating conditions...like liquid Helium, or at least liquid Nitrogen. (Helium is far superior, because it reduces the noise level to near the minimum possible, but Nitrogen may suffice, and would be a lot cheaper.)
And when I say "created by accident" I don't mean that none of the components were designed as AI. Merely that the system wasn't designed as AI. It will certainly have components that were considered AI 2-3 years ago, like adaptive pattern recognition.
I haven't said much about its goals, because I don't consider that to be a function of intelligence, but rather an underpinning of intelligence. Increased intelligence won't change your goals, but it may well change how you persue them. And many different goals could serve as the basis of intelligence. Consider what might happen if the first real AI were developed out of an automated hospital. (Which doesn't mean no doctors and nurses. I'm envisioning the AI arrising out of automated management functions. Automated doctors and nurses would be some sort of "narrow AI".)
Well, this is probably already TL;DR, so I'll stop at this point.
You left out "at the moment".
Drones are probably being developed largely as troops that won't revolt when ordered to attack civil unrest...at home. That they are first used against foreigners while under development is just typical. Some police forces have already been using them at home. When they are developed and debugged...well, ...
I've always considered "turns against" to be an unlikely scenario. I envision the machine becoming an "infallible advisor" to such an extent that the leader (CEO, President, Prime Minister, whatever) becomes a figurehead, and that all middle management is progressively replaced. And the system will be so designed that if the figurehead stops obeying the "suggestions" of the machine, s/he will be found incompetent, and replaced.
FWIW, we seem to be increasingly headed in this direction, limited only by costs of computation, limits of pattern recognition, and inflexibility of robots. Those limits are all being worked on.
FWIW, *both* military and factories are already well hooked up to proto-AIs. The current ones aren't really AI, but they already are looked upon as infallible decision makers by managers who don't want to take responsibility. And they're right enough of the time that that's not an unreasonable response. It's true, their decisions are tightly focused, but High Frequency Trading is only the most obvious example. They are spread throughout the decision making process.
It's my opinion that the first true general purpose AI will arise by accident. People trying to improve their C&C system. The question of the history of the race will be decided by what it is designed to optimize, and how it determines whether it has made the correct decision.
OK. I made one "typo" (well, "though-o"). I mean fission power, rather than fusion, would be enough to get us out to the Oort clouds.
OTOH, the icy moons of Jupiter have too high a gravity to be an attractive choice. You'd need specially designed equipment. Those are no minor masses. Still, there are plenty of icy asteroids beyond Jupiter's orbit.
WRT the Oort clouds...there's a lot of stuff there, but it's rather thinly spread. So figure long transition times between "stops". Probably not practical until we are building habitats the size of cities.
For that matter, it's not wise to focus too tightly on one particular approach. Consider a habitat that's in the shape of a long tube, and that grows by building onto the ends. Perhaps a couple of miles in diameter, though different sections could have different diameters. And spinning, for gravity. Given a long enough tube you can make bends through some pretty sharp angles without causing much stress, though I would design it so that different sections could be safely decoupled, and replaced, removed, or inserted.You'd probably want an occasional non-rotating section. Transport if via electric train, though, so you'd want the sections to be pretty long, to avoid excessive transfers, unless you run the trains in an evacuated tube in the center. Occasionally you'd want an extension to run either straight in or straight out. Power collected at locations near the sun. How far could you find building material to reach? Now use one of the "straight out" constructs as a linear accelerator to launch vehicles to the stars. You should easily have an accelerator 100's of miles long in a vacuum all the way. I'd be a bit hesitatnt, however, about using it to catch incoming freight, even presuming it was designed to do so. Also, you don't want a really high launch velocity, because that would make interstellar material act like really heavy penetrating projectiles. Even a few grams impacting at over 0.5c would be unsurvivable. And though such things are rare, you'd be transiting a HUGE amount of space. So what you want to launch would be habitats designed to eventually go into orbit around some other star. They'd need fission power sufficient to make the trip, with a bit of leeway to allow them to start mining at the arrival end. And they'd need to be travelling at not too far above the local speed of debris, so that they could either dodge it, capture it, or survive the impact. (Lead shielding for impact survival, as the impact would be "supersonic" so crystal strength would be much less important than mass.)
This approach would certainly let us reach the Oort clouds, and live there as long as power held out, but I doubt that there's much fissionable mass available there, so stopping there doesn't seem viable. (Controlled fusion would, of course, change things *quite* significantly, though just how would depend on the machinery required to control it.)
Given a nearly closed ecology there are many approaches to living in space, and which is chosen is more a social choice than an engineering one. (OTOH, there are a lot of ways that just wouldn't work, without "magic technology".)
Don't expect the cost to stay constant. Even the next time could be a lot cheaper, as the software has been written, and computers have gotten faster and cheaper.
OTOH, there are probably limits as to how cheap it can get. I'd be really surprised if it got cheaper than $500,000 during this century, unless there was MASSIVE deflation. That would additionally require robotic surgeons doing and examining the slices, as well as better algorithms (probably rewritten to be more extensively parallel) running on the cheaper computers.
A definite point, even though I don't agree with the grandparent. He's right that energy *IS* a constraining factor. Even worse (currently) is our inability to maintain a nearly closed ecosystem. I suspect that accelerations in space will always be slow, and I suspect that FTL is actually impossible. That still doesn't close off even meat-organisms to space (except that we haven't learned to maintain a nearly closed ecosystem).
Since once you're in orbit, low accelerations suffice to reach any other location in orbit anywhere, the only problem is to maintain sufficient energy flow during the transition (and after you get there). For solar space, out to around the orbit of Jupiter, solar cells should suffice. And there are plenty of asteroids in that area. Beyond, out as far as the Oort clouds, fusion power should suffice. For propulsion the best current technology appears to be to use ion-jets. And you don't economize on living space, because you don't want to go crazy. Also, you need a radiation shield, and you need water anyway, so you carry a sizeable weight of water. This all means that transitions are SLOW. So you'd better have enough people along to keep you sane. 12-13 adults seems about right for a small habitat. Scale upwards as you learn more.
Note that this is all doable TODAY, given the intention, except that we can't maintain a nearly closed ecosystem. And learning to maintain a nearly closed ecosystem can be done (well, mainly) at ground level for cheap prices.
That's a decent analogy, but it leaves out that a bunch of the structure of the brain is driven by environmental stimuli.
To be honest, Mao wasn't that bad. He wasn't any Stalin. He wasn't very friendly to the US, and he had some silly ideas, but he generally had good intentions. As authoritarian dictators go Mao was pretty decent. The real problem is centralized control. But as Somali proves, anarchy isn't all that good either. (Basically, anarchy is unstable against some group establishing a coercive power structure.)
And democracies and republics seem to drift towards authoritarian dictatorships. Britain is a bit of a miracle, caused by generations of Kings who didn't speak the native language, and ignored what the parliments were doing. But its looking a bit as if it's drifting towards an authoritarian society WITHOUT a dictator. I'm not real sure that's an improvement...though you could ask the modern chinese, except they don't have a decent basis for comparison. They've always lived with either an Emperor or an anarchistic mess of warlords until post-Mao, when they've drifted into a centralist authoritarian government without a dictator.
Possibly humans just can't be trusted with power.
Considering that the attacks were described as "suicidal", calling the perpetrators "cowards" seems gross misuse of the language. "Crazed" is plausible, but needs more evidence. 'Fanatic bigots" seems more justifiable.
Ok. To me pickup truck implies a different set of uses (primarily short trips, but capable of more), it doesn't imply a heavy lifter (in which I include anything that can lift off a planetary mass).
But lifting off a 1g mass directly is going to be VERY expensive, unless you have some sort of skyhook. (I include catapults in the skyhook category, even though that's clearly a misnomer for them, but they don't work well through an atmosphere anyway.) So you STILL need a nearly closed eco-system. Or you end up in the ISS situation, where every weeks living is dependent on your ground support not having had its budget cut, or being destroyed by a tornado, or what-all. That's nothing that a durable habitation can be built around, and it also ties you closely to NEO. Going even as far as lunar orbit does horrendous things to the expenses.
OK, I'm not a compiler builder. And a hash table would be better for a symbol table. And I was thinking about a slightly different representational problem, for which a Trie would also not be the correct data structure to use, but which seemed to have the same problem. (It's actually an n-dimensional list structure...though less general than that implies. And I'm probably going to slap a restrictive upper limit on n...at least if I can figure a way to do so that won't choke things up.)
I don't think your proposal is practical. The "space pickup truck" idea is nice, but the idea that it should descend to and lift off from a planetary mass is unfeasible, without controlled fusion, and perhaps even then. (You could do something similar with some sort of skyhook [the PinWheel is my favorite, as being the most practical in the near term].)
But FIRST you need to work on a nearly-closed eco-system. This "space pickup truck" will take a long time to make a long trip. It will probably be powered by solar cells and a ion-jet of sorme variety. This will let it move from orbit to orbit, carry things along, etc., even do interplanetary voyages. But it's SLOW. This means that it needs to be either robotic or a closed eco-systme. (Well, nearly closed. Really closed is impossible.) And if the people who are inhabiting it are going to remain sane, it needs to be large enough to live in. Larger, probably, than a hutment in Antarctica. (It's easier to go outside in Antarctica, and less expensive of habitability. In Antarctica that just means heat, in space it primarily means air and water...no matter how careful you are.) Also, if you design it for habitability you need to carry along significan radiation shielding. Water is probably a good choice for that. But that means weight.
If, OTOH, you only mean short interorbit transitions you still end up with a slow vehicle. And one that can't land on a planetary surface.
Please note that the same driver frame should be able to have several different cabins mounted on it, and should be able to haul cargo. But that's really tricky when you're using rocket/jet propulsion. You can't haul the stuff behind on a cable, unless it's WAY behind, and then the cable had better be able to stand being exposed to the wash of the exhaust. Probably better would be to have three or four main engines angled slightly outwards, and slightly manuverable, to allow hauling things on a cable, even though that would reduce the efficiency. But adjustable, so if you weren't hauling cargo you could eject straight backwards.
Please note that this approach won't work much beyond the orbit of Jupiter, as solar cells will become too weak to generate enough power. So you need some alternate power source. Fission reactors are the only thing that currently seems feasible. Fission reactors are probably enough to allow manuvering throughout the solar system, but for interstellar, even at slow speed, fusion may well be needed. (I suppose that one could build an anti-matter factory near the sun and power it off sunlight, and then store it as fuel...that seems only an engineering problem, with no theoretica breakthroughs required. But it's a HUGE engineering problem.)
Near term, though, to get from a planetary surface with an atmosphere, use a skyhook. (Some could be built now. They aren't all as difficult as an elevator.) If there's no atmosphere, use a catapult. Ion rockets to get from orbit to orbit. And human presence requires a nearly closed ecology. Which can be developed right down here. Siberia or Canada seem like good locales to do the development, with other locations if it needs to deal with excess heat rather than excess cold. (And note that right now the real requirement is cheap land that can be somewhat isolated, and funding. BioSphere & BioSphere II were examples of attempts that have failed, and in failing taught us something. But nobody's put serious effort into the development. And it needs to be done before anyone does any serious planning about a permanent human presence off the planet.
There's an interesting reason, though. Consider building a Trie around Unicode chars. Granted, this may not be a major reason, but UGH! There's a lot of advantages to having a small alphabet. The early languages didn't usually even allow both upper and lower case. Well, memories have expanded, processors have speeded up, etc. But Unicode is still too verbose for many algorithms to work well. And using bytes and utf-8 yields different problems.
I will grant that there are lots of approaches that don't suffer dramatically from this problem. It's largely historic. But there are still valid reasons.
IIRC, driving was originally unregulated. Neither a right nor a priviledge. (Unless you count oxcarts and horses. If you do, it predates the invention of "rights", and was presumed to be allowed to anyone who could afford it. Though, IIRC, at one point Rome had laws baning "trucks" between dawn and sunset. But that's traffic control, not forbidding driving.)
I think the changes will be faster than you suppose. States already have laws saying that driving is a priviledge, not a right. And the insurance companies will be pushing for any change that reduces their expenses (while continuing to require that you purchase increasingly worthless insurance).
Taxi drivers will probably lose out quickly. Bus drivers will change to "stewards". Their main job will change to controlling passengers. Shuttle drivers...probably the same as bus drivers.
OTOH, in most places it will require legal changes to allow driverless taxis. Even taxis with drivers tend to be licensed and controlled, so there's an entrenched bureaucracy. So there will be resistence that won't collapse until large companies go into the automated taxi business. And, as with buses, vandalism will be a problem.
You have a point, but the people doing the job you're describing are more a cross between longshoremen and mechanics than truckers. (Yeah, I know they're mainly far from the water, but the job description doesn't really include ships, it include handling cargo.)
OTOH, consider all the automated warehouses that are going in. There will be lots of places where even these neo-truckers aren't needed.
Someone who always lies is quite reliable.
His statement is quite believable...because at least in the summary he didn't mention what measures he used. It's probably more profitable for the provider than anywhere else, for example.
Believable doesn't mean intelligible. The statement as I read it could mean nearly anything. There are LOTS of different measures of quality that are possible. Many of them don't even refer to the level of service to the end user.
One interesting point is that he said the problem was the number of subscriptions. This when at least MY internet connection often slows down to dial-up speed (as measured by rate of transmission). I've TODAY had connection speeds of 22.5 KB/S. Mixed with long periods when the was NO data transmission beyond handshaking.
So while I believe what he said, I don't believe what he was trying to imply. I consider him a spin-doctor, at best.