If you don't see people being displaced yet, it's because you aren't looking in the right place. The first thing you need to realize is the the figures for unemployment are highly fictitious... well, not *exactly* fictitious, but rather carefully redefined to give a politically acceptable result. And that the ways of calculating them are re-defined as needed. Similarly for the statistics on money supply, etc. You can't easily do historical comparisons, because the accessible data has been corrupted.
OTOH, it was worse at the start of the industrial revolution, at least so far. Then there was a high immediate fatality rate which wasn't tracked officially at all. "The sheep are eating the men" wasn't hyperbole, merely metaphor.
One can hope that your analogy with flying is correct. There are still many things that birds do better than planes. Even so I consider that a conservative projection when given without a time-line.
AI's are already improving their own code. That's what "learning" means. It's just that it's more difficult to apply that technique when the satisfaction constraints are so vague. That's why the early AIs specialized in games. It's easy to specify what satisfactory performance is. Driving is more difficult, because you don't want to *ever* experience an unsatisfactory outcome, but at least you can tell when one happened. Other skills are more difficult.
Then there's the "world complexity" where the conditions of the world are an orthogonal measure of the difficulty of the problem. AIs had already handled simple abstract problems in the early 1970's. An AI came up with quite original geometric proofs. Simple world, easy determination of success. These days mathematical proof checkers are pretty much regular tools in standard proofs. (The well validated ones, however, haven't been adapted to uncommon tools.) This isn't what we normally call AI, it's validation of logic, but it's an important *part* of a good AI...in certain fields. And programming is one of those fields.
The reason that an AI programmer has been slow to show up is because the problems are usually posed in a manner that's, at best, sloppy, and the satisfaction criteria are not easily judged. But AIs are making continual in-roads on handling both of those constraints.
I don't know what the right answer is, but it's not unions. Unions exist to protect jobs and employment. The Pacific Longshoremen's Union during the 1960's&70's was an aberration in the the union bosses didn't primarily look after maintaining their own power via maintaining a large number of jobs, but rather opted into profit sharing, protecting the current workers at the expense of future power. Usually a union can be depended upon to fight automation, rather than to seek maximization of public good.
Basic Income has lots of things going for it, but it doesn't feel like the exactly right answer. I'm more in favor of a linear income tax (with no writeoffs or exemptions) where the intercept is set to equal what the basic income would otherwise have been. And this tax should include *ALL* sources of income and replace all other taxes.
The problem with basic income style solutions is that it centralizes control which strengthens a central point of failure. It's true the problem already exists (money is only worth something because you need to pay your taxes in money, so everyone needs it...and money is printed by the central government), but I'm not sure strengthening the centralization doesn't make the problem worse.
But what better answers are there? It would be nice to at least notice their existence before we hit the narrow part of the pass.
What you're ignoring is that the military is becoming steadily more mechanized also. There won't be many jobs there, either. Robots are more reliable and less likely to side with the protesters.
Wouldn't tell him how and wanted to charge $320...not exactly "refused", but certainly far from assisted.
If that happened to me I might well characterize them as having refused to help me. A fuller explanation would be more accurate, but would also be so long most people wouldn't listen.
I think you've got the wrong problem. ISTM that the problem (which *may* have been fixed) is that people who know the subject matter keep getting their edits reverted by those who don't have a clue, but who have an investment in making a lot of edits. I've heard many complain that they were never going to bother editing a Wikipedia page again, because it was like writing on the wind. Nothing they wrote would be preserved, so why bother.
Unfortunately Wikipedia now has such a bad reputation that there are many experts who will not only never contribute to them again, they will, if asked, strongly recommend against anyone else so wasting their time.
I don't think you understand international politics. The US has actually been rather benign as dominant international powers go. Admittedly it's been quite unpleasant, and as time goes on it has been getting worse. If historical patterns continue it will continue to get worse until it finally realizes that it lost it's international dominance quite awhile ago. (That's the normal pattern for the decline of a dominant international power.) It *may* recover afterwards, though that's uncommon.
FWIW, it looks as if China may already be the actual dominant international power. I think the election of Trump was partially caused by the implicit realization that the US has already lost it's dominant position. I'd been hoping that it would be Japan, but that doesn't look like happening, though India is still an outside chance. (But pretty far outside.)
It's not clear that Trump is a racist, though certainly many of his friends and appointees are, but it *is* clear that he's quite willing to take unjust advantage of anyone in a weak position, such as one caused by other people being racist. And it's quite clear that he's willing to break contracts unjustly, when he finds it convenient.
That a bit long-winded, so most people just say "Trump's a racist" which isn't exactly clearly true. In fact I rather doubt it, as I think he finds it quite convenient to have a bunch of people who are powerless to defend themselves around.
FWIW, I believe that if/when Britain withdraws from the EU it will automatically cease to be a member of the WTO. This only affects half of your argument, however.
You've got the wrong country. Of course, the problems you raise are related to the article being commented upon, but it's a second- or third-hand relationship.
Still, since they're all happening at the same time, they will interact in patterns of reinforcement and cancellation. But trying to model everything at once is too complicated to do. It's bad enough trying to model *just* BREXIT, or *just* Trump.
There's lots of fake news that isn't explicitly political. This *could* be fake news, but I see no reason to believe it is. But just being economic doesn't make it fake news.
(FWIW, I think it probably *is* incorrect news, but not for the reasons indicated, but because it's being too rosy-optimistic. And possibly not intentionally so, but only because they can't bring themselves to look at the general world picture and imagine how bad it's likely to get.)
You're right, my acquaintance with COBOL was always at a distance. I did see some code, but it was in printout, and I think I once wrote a 5 line routine, but someone else keypunched it. But I would have expected to know if it was picky about line columns. Perhaps it was a matter of which implementation was being used, or compiler options or some such. (Certainly modern Fortrans aren't picky about columns used, unless you pick certain compiler options.)
Well, at the time COBOL was common, the only editor was a *I* had was a keypunch machine. But I never knew that COBOL was sensitive to indentation. Fortran was, as the first 5 columns were reserved for numeric labels, and the 6th column was reserved for a continuation marker, and columns 73-80 were reserved for non-code uses (often sequence numbers), but I never heard any rule like that for COBOL.
So back then I couldn't, and didn't, use tabs. (I *did* use special formatting cards in the keypunch...but I no longer remember the details, just as I no longer remember how to punch carriage control tapes for the printer.)
That said, the tabs vs. spaces argument is silly. I prefer tabs, but I can automate conversion of them to spaces when I ship the code to someone who prefers spaces. The only problem comes when you mix them...which can happen when converting spaces to tabs, but not in the tabs-to-spaces direction. (Yeah, I *could* tell my editor to always replace tabs by spaces, but I won't. I see no advantage and a few disadvantages. [The disadvantages are only annoyances, but they *are* annoyances.])
IIUC, there is a rule against mixing tabs and spaces at the beginning of lines. I think that's a syntax error. This was a problem with Python 2.?.
That said, as I always use tabs at the beginning of lines I've never encountered the error, at least not within the last couple of years. I prefer tabs over spaces because if I start getting deeply indented lines it's easy to reduce the amount of horizontal indentation without editing the lines. (Just change how the editor displays tabs.)
OTOH, I follow this rule whatever language I using. It's not something specific to Python. I feel that a consistent pattern of indentation is valuable in making code intelligible whether or not braces are used...or for that matter begin...end pairs. (Ada does/did insist on the function name appended to the end token, but that doesn't really change things, and I often do that even in Python. e.g.:
def fun (args): ... some code here...
#def fun (args)
Sorry about the spacing, but/. doesn't seem to preserve it.)
Actually, if you assume that the AI is a good implementation of his vision, that's not an unreasonable way to do the management. Unfortunately(?) the environment isn't static, and once his vision is fixed in code, it *is*. Perhaps only at a rather abstract level, if it's a really smart AI (which, I believe, is beyond the current state of the art), but fixed. So when the environment changes in an unexpected way, it will fail...or at least react in an unpredictable way.
Well, there was this book called "A Random Walk Down Wall Street", which pretty much made the same claim, but I think that the actual job is a bit more skilled than that. Of course, this doesn't mean that it wouldn't be well within the capabilities of a good AI.
Please note, this *ISN'T* the level of automation that people have been predicting "at least since the first decades of the 20th century". This is a relatively simple thing. The automated cars, imperfect as they are, are much more impressive. But this *is* a step along the way, and it *is* the kind of thing that will be done increasingly. And many of them will be "for the benefit of the owners/stockholders", which knocks on the head that "for the benefit of the living" argument.
No, what's innovative about this is that it's automating away management. If it's successful, this is going to be wildly unpopular with many powerful people, and equally popular with other powerful people. We may here have the first signs of the successor to "international communism" as the demon of the century.
Unfortunately, "moron" is more likely. I've run into many people who just can't imagine how bad some other people's lives are. And "moron" is in quotes because many of these are people who are quite competent in areas in which they are experienced.
It actually *is* mentally easy *IF* you have an infallible memory AND you can work basic logic AND your logic runs quite quickly.
Admittedly, that would only qualify you as a very bottom level "grand master", but it would suffice to beat most masters. The games would be uninspired, but technically sound, and the defense could be essentially unbeatable. Just about all your games against master level players would be draws.
You are probably incorrect to conflate "eagle-works" was it? with NASA, but several reasonably respectable groups have reported small positive results. And I'm including the "eagle-works" report in the "reasonably respectable groups". So it's likely to be correct, and it's also likely to fall within the bounds of current physics, even if nobody can really explain why just yet. Physics has a lot of stuff in it with complex interactions, so it's quite reasonable that it should include things that nobody has thought of putting together in quite that way yet which produce results that aren't what anyone expected. This is because complex systems often have unexpected interactions. If you go back 200 years nobody expected that it could explain thought. And maybe it doesn't, and we still need some more basic explanations, but that's not the way to bet.
The page I read yesterday claimed that this was entirely consistent with the conventional laws of physics, and was based on radiation pressure in a resonant cavity of an appropriate shape.
This would seem to put rather strong limits on the amount of power that could be generated by an EM drive, putting in as only slightly better than ion rockets as far as impulse (and perhaps not as good), but significantly simpler in design and only requiring electricity as fuel.
In short, if the paper was correct this is probably an improvement on current technology for certain applications, but nothing dramatic, and only useful for certain applications. If you're thinking of using a light sail (without a launch laser) or an ion rocket, you should consider this as an option.
Slashdot Mirror
If you don't see people being displaced yet, it's because you aren't looking in the right place. The first thing you need to realize is the the figures for unemployment are highly fictitious ... well, not *exactly* fictitious, but rather carefully redefined to give a politically acceptable result. And that the ways of calculating them are re-defined as needed. Similarly for the statistics on money supply, etc. You can't easily do historical comparisons, because the accessible data has been corrupted.
OTOH, it was worse at the start of the industrial revolution, at least so far. Then there was a high immediate fatality rate which wasn't tracked officially at all. "The sheep are eating the men" wasn't hyperbole, merely metaphor.
One can hope that your analogy with flying is correct. There are still many things that birds do better than planes. Even so I consider that a conservative projection when given without a time-line.
AI's are already improving their own code. That's what "learning" means. It's just that it's more difficult to apply that technique when the satisfaction constraints are so vague. That's why the early AIs specialized in games. It's easy to specify what satisfactory performance is. Driving is more difficult, because you don't want to *ever* experience an unsatisfactory outcome, but at least you can tell when one happened. Other skills are more difficult.
Then there's the "world complexity" where the conditions of the world are an orthogonal measure of the difficulty of the problem. AIs had already handled simple abstract problems in the early 1970's. An AI came up with quite original geometric proofs. Simple world, easy determination of success. These days mathematical proof checkers are pretty much regular tools in standard proofs. (The well validated ones, however, haven't been adapted to uncommon tools.) This isn't what we normally call AI, it's validation of logic, but it's an important *part* of a good AI...in certain fields. And programming is one of those fields.
The reason that an AI programmer has been slow to show up is because the problems are usually posed in a manner that's, at best, sloppy, and the satisfaction criteria are not easily judged. But AIs are making continual in-roads on handling both of those constraints.
I don't know what the right answer is, but it's not unions. Unions exist to protect jobs and employment. The Pacific Longshoremen's Union during the 1960's&70's was an aberration in the the union bosses didn't primarily look after maintaining their own power via maintaining a large number of jobs, but rather opted into profit sharing, protecting the current workers at the expense of future power. Usually a union can be depended upon to fight automation, rather than to seek maximization of public good.
Basic Income has lots of things going for it, but it doesn't feel like the exactly right answer. I'm more in favor of a linear income tax (with no writeoffs or exemptions) where the intercept is set to equal what the basic income would otherwise have been. And this tax should include *ALL* sources of income and replace all other taxes.
The problem with basic income style solutions is that it centralizes control which strengthens a central point of failure. It's true the problem already exists (money is only worth something because you need to pay your taxes in money, so everyone needs it...and money is printed by the central government), but I'm not sure strengthening the centralization doesn't make the problem worse.
But what better answers are there? It would be nice to at least notice their existence before we hit the narrow part of the pass.
What you're ignoring is that the military is becoming steadily more mechanized also. There won't be many jobs there, either. Robots are more reliable and less likely to side with the protesters.
Wouldn't tell him how and wanted to charge $320...not exactly "refused", but certainly far from assisted.
If that happened to me I might well characterize them as having refused to help me. A fuller explanation would be more accurate, but would also be so long most people wouldn't listen.
I think you've got the wrong problem. ISTM that the problem (which *may* have been fixed) is that people who know the subject matter keep getting their edits reverted by those who don't have a clue, but who have an investment in making a lot of edits. I've heard many complain that they were never going to bother editing a Wikipedia page again, because it was like writing on the wind. Nothing they wrote would be preserved, so why bother.
Unfortunately Wikipedia now has such a bad reputation that there are many experts who will not only never contribute to them again, they will, if asked, strongly recommend against anyone else so wasting their time.
I don't think you understand international politics. The US has actually been rather benign as dominant international powers go. Admittedly it's been quite unpleasant, and as time goes on it has been getting worse. If historical patterns continue it will continue to get worse until it finally realizes that it lost it's international dominance quite awhile ago. (That's the normal pattern for the decline of a dominant international power.) It *may* recover afterwards, though that's uncommon.
FWIW, it looks as if China may already be the actual dominant international power. I think the election of Trump was partially caused by the implicit realization that the US has already lost it's dominant position. I'd been hoping that it would be Japan, but that doesn't look like happening, though India is still an outside chance. (But pretty far outside.)
It's not clear that Trump is a racist, though certainly many of his friends and appointees are, but it *is* clear that he's quite willing to take unjust advantage of anyone in a weak position, such as one caused by other people being racist. And it's quite clear that he's willing to break contracts unjustly, when he finds it convenient.
That a bit long-winded, so most people just say "Trump's a racist" which isn't exactly clearly true. In fact I rather doubt it, as I think he finds it quite convenient to have a bunch of people who are powerless to defend themselves around.
FWIW, I believe that if/when Britain withdraws from the EU it will automatically cease to be a member of the WTO. This only affects half of your argument, however.
You've got the wrong country. Of course, the problems you raise are related to the article being commented upon, but it's a second- or third-hand relationship.
Still, since they're all happening at the same time, they will interact in patterns of reinforcement and cancellation. But trying to model everything at once is too complicated to do. It's bad enough trying to model *just* BREXIT, or *just* Trump.
There's lots of fake news that isn't explicitly political. This *could* be fake news, but I see no reason to believe it is. But just being economic doesn't make it fake news.
(FWIW, I think it probably *is* incorrect news, but not for the reasons indicated, but because it's being too rosy-optimistic. And possibly not intentionally so, but only because they can't bring themselves to look at the general world picture and imagine how bad it's likely to get.)
You're right, my acquaintance with COBOL was always at a distance. I did see some code, but it was in printout, and I think I once wrote a 5 line routine, but someone else keypunched it. But I would have expected to know if it was picky about line columns. Perhaps it was a matter of which implementation was being used, or compiler options or some such. (Certainly modern Fortrans aren't picky about columns used, unless you pick certain compiler options.)
Well, at the time COBOL was common, the only editor was a *I* had was a keypunch machine. But I never knew that COBOL was sensitive to indentation. Fortran was, as the first 5 columns were reserved for numeric labels, and the 6th column was reserved for a continuation marker, and columns 73-80 were reserved for non-code uses (often sequence numbers), but I never heard any rule like that for COBOL.
So back then I couldn't, and didn't, use tabs. (I *did* use special formatting cards in the keypunch...but I no longer remember the details, just as I no longer remember how to punch carriage control tapes for the printer.)
That said, the tabs vs. spaces argument is silly. I prefer tabs, but I can automate conversion of them to spaces when I ship the code to someone who prefers spaces. The only problem comes when you mix them...which can happen when converting spaces to tabs, but not in the tabs-to-spaces direction. (Yeah, I *could* tell my editor to always replace tabs by spaces, but I won't. I see no advantage and a few disadvantages. [The disadvantages are only annoyances, but they *are* annoyances.])
IIUC, there is a rule against mixing tabs and spaces at the beginning of lines. I think that's a syntax error. This was a problem with Python 2.?.
That said, as I always use tabs at the beginning of lines I've never encountered the error, at least not within the last couple of years. I prefer tabs over spaces because if I start getting deeply indented lines it's easy to reduce the amount of horizontal indentation without editing the lines. (Just change how the editor displays tabs.)
OTOH, I follow this rule whatever language I using. It's not something specific to Python. I feel that a consistent pattern of indentation is valuable in making code intelligible whether or not braces are used...or for that matter begin...end pairs. (Ada does/did insist on the function name appended to the end token, but that doesn't really change things, and I often do that even in Python. e.g.:
... some code here ...
/. doesn't seem to preserve it.)
def fun (args):
#def fun (args)
Sorry about the spacing, but
Actually, if you assume that the AI is a good implementation of his vision, that's not an unreasonable way to do the management. Unfortunately(?) the environment isn't static, and once his vision is fixed in code, it *is*. Perhaps only at a rather abstract level, if it's a really smart AI (which, I believe, is beyond the current state of the art), but fixed. So when the environment changes in an unexpected way, it will fail...or at least react in an unpredictable way.
Well, there was this book called "A Random Walk Down Wall Street", which pretty much made the same claim, but I think that the actual job is a bit more skilled than that. Of course, this doesn't mean that it wouldn't be well within the capabilities of a good AI.
Please note, this *ISN'T* the level of automation that people have been predicting "at least since the first decades of the 20th century". This is a relatively simple thing. The automated cars, imperfect as they are, are much more impressive. But this *is* a step along the way, and it *is* the kind of thing that will be done increasingly. And many of them will be "for the benefit of the owners/stockholders", which knocks on the head that "for the benefit of the living" argument.
No, what's innovative about this is that it's automating away management. If it's successful, this is going to be wildly unpopular with many powerful people, and equally popular with other powerful people. We may here have the first signs of the successor to "international communism" as the demon of the century.
Unfortunately, "moron" is more likely. I've run into many people who just can't imagine how bad some other people's lives are. And "moron" is in quotes because many of these are people who are quite competent in areas in which they are experienced.
It actually *is* mentally easy *IF* you have an infallible memory AND you can work basic logic AND your logic runs quite quickly.
Admittedly, that would only qualify you as a very bottom level "grand master", but it would suffice to beat most masters. The games would be uninspired, but technically sound, and the defense could be essentially unbeatable. Just about all your games against master level players would be draws.
Sorry, but history repeatedly shows otherwise.
You can be certain that they *ARE* being influenced. This doesn't mean he's lying, but there are certainly things he isn't saying.
But this isn't fake, it's just "What kind of idiot didn't know he would be forced into contact with Russian Intelligence if he wanted to stay there?".
I'll agree, however, that this is published because "It's far too useful to some folks when they need to sway public opinion on something."
You are probably incorrect to conflate "eagle-works" was it? with NASA, but several reasonably respectable groups have reported small positive results. And I'm including the "eagle-works" report in the "reasonably respectable groups". So it's likely to be correct, and it's also likely to fall within the bounds of current physics, even if nobody can really explain why just yet. Physics has a lot of stuff in it with complex interactions, so it's quite reasonable that it should include things that nobody has thought of putting together in quite that way yet which produce results that aren't what anyone expected. This is because complex systems often have unexpected interactions. If you go back 200 years nobody expected that it could explain thought. And maybe it doesn't, and we still need some more basic explanations, but that's not the way to bet.
The page I read yesterday claimed that this was entirely consistent with the conventional laws of physics, and was based on radiation pressure in a resonant cavity of an appropriate shape.
This would seem to put rather strong limits on the amount of power that could be generated by an EM drive, putting in as only slightly better than ion rockets as far as impulse (and perhaps not as good), but significantly simpler in design and only requiring electricity as fuel.
In short, if the paper was correct this is probably an improvement on current technology for certain applications, but nothing dramatic, and only useful for certain applications. If you're thinking of using a light sail (without a launch laser) or an ion rocket, you should consider this as an option.