I only interacted with him a small amount, but he seemed like a nice guy. He was clearly very dedicated to all sorts of open-source projects and his dedication was combined with a cleverness and willingness to listen to those around him. He will be missed.
I'm not originally from Massachusetts, and I've lived in a whole bunch of states (Alabama, Connecticut, Maine and now Iowa), and I can see why people from Mass have this attitude about environmental issues. It is very clear that on environmental issues both large and small, not only is Mass better than they in terms of regulations but also in terms of people simply being willing to do minor things in their day-to-day lives like reusing things rather than throwing them out, or keeping their heating and air conditioning at temperatures that reduce use, etc.
Bill Clinton, George W. Bush, and Barack Obama all had serious transparency issues and trouble paying attention to public feedback. However, neither Bush nor Obama was by and large that bad, and the current administration is so bad it makes the previous ones look like paragons of transparency and responsiveness.
Is there any evidence that anyone has attempted or intends to do this? As far as I can tell, this is essentially a conservative urban legend. And when RAND did a study on the costs of including transgender individuals they seem to have concluded that this was not at all likely https://www.rand.org/news/press/2016/06/30.html.
The old Slashdot slogan used to be "News for nerds, stuff that matters." Most Slashdot articles fall into both categories. It is rare to see an article which more so unambiguously falls into one category and not the other. Still, a very impressive feat.
So, I agree that there are specific concerns in that context, and they do require substantial examination to see how much they matter. Luckily, we also have data from other countries. For example, the IDF allows transgender individuals and has had no problems aside from some very minor accommodations they had to make. There are many medical circumstances where we don't put people into front-line combat while they are undergoing it, but that doesn't mean we don't put them in the military wholesale.
Whether we label something as a mental health problem is less relevant to whether or not it is a mental health problem that interferes with their ability to fight. Many minor mental health problems are allowed for people in the military or in security contexts as long as they don't impair ability. That's something that should be made on a judgment based on that sort of concern.
The idea isn't that people are consciously putting sexism in front of profits but rather that they are unconsciously allowing sexism to cause them to underestimate their potential profits.
The purpose of the military, is to break things and kill people. It needs to be a tight cohesive bunch, not with a bunch of "my gender of the week" types.
I don't know why you think someone having a different gender identity than their assigned at birth identity somehow must make them "weak." Heck, for that matter, even if you did have someone whose gender identity changed every week as you suggest, I don't see why that would them have trouble killing people or being part of a cohesive fighting force.
The military should be EXEMPT from all of this political correctness garbage, which, is set up on purpose, to reduce morality, and destroy the best fighting force in the world.
Or this is about people who genuinely want the right to fight for their country. You appear to be jumping from the fact that you disagree with other people to concluding that they must have nefarious goals. Incidentally, it is worth noting that the exact same arguments you are making now have now been made twice before. They were first made when we desegregated the military to have black and white people in the same units. This argument was again made only a few years ago when we let gays and lesbians openly serve. Why is it different this time?
No, the concern is that we don't know when it will happen, but we do know that it will happen eventually, and that specific aspects of what intelligent entities can do cause concern. The argument is that we need to get this right before we get the actual technologies.
We also for that matter don't know how complicated long-term planning is and how much planning is essentially sophisticated pattern recognition with a few other algorithms grafted on. Everyone agrees that current systems *cannot* duplicate human cognition. That isn't the concern.
I am not talking about how much we remember (application layer) or how much we think we remember, but about the raw storage capacity of our brain (physical layer) here the capacity is enormous. Consider: to our current understanding, information is not stored in neurons but in circuits of neurons. We have 100 billion neurons and each can link to up to 10000 other neurons. I have not found any indicator as to up to how many neurons can form one circuit, but suffice to say that the possible number of circuits is very, very high.
The mistake here is going from that we could conceive of a brain using that much information storage in the most efficient form possible to thinking that it does. We know that in practice, brains engage in massive amounts of compression of stored information and reconstruct memories when they are retrieved. That's a big part of why human memory is so often wrong or distorted http://www.newyorker.com/science/maria-konnikova/idea-happened-memory-recollection.
As for planes, I agree that by many metrics they are less efficient than birds, but by some metrics (maxmimum speed for example) they are substantially better. But it also isn't intended as a perfect analogy anyways; the point of both the bird example and the fire example is that one doesn't need to understand well a natural thing in order to make something similar to it.
If the last AI winter has taught us anything then that we should not expect the AI development to move very fast.
Part of the AI winter was a collapse in funding more than anything else. And in so far as the AI winter happened, it doesn't mean a similar winter will happen. It isn't infrequent for a technology to emerge at first, seem to not be progressing, and then years later to burst on the scene with little warning. Smartphones, electric cars, and solar panels are all examples.
There are massive ranges in how much your brain stores, depending a lot on how much detail of episodic memory we keep. We also don't know how efficient our processing arrangement is, and the fact that many bird species are highly intelligent with very small brains suggests that it isn't optimized that much.
Planes were never meant to fly "like birds", or how many wpm does the new Airbus 380 do? (wingflaps per minute).
That's the exact point though. We didn't need to duplicate exactly how a bird flies to get flying objects. But you are apparently very certain that we'll need to duplicate how a human brain thinks to get a general AI.
That you find something ridiculous or label it magical doesn't make it less of a concern.
If you don't understand the human brain, what makes you so certain that changes to existing machine learning absolutely cannot result in a strong AI with a few small additional insights? For that matter, how do you possibly justify that we are "infinite far away" when we have neural networks and have developed understanding of specific brain aspects. Moreover, and most importantly, we made airplanes before we understood how birds fly. We made fire for ourselves before we understood how lightning made fire in the forests. Etc. The ability to have a perfect understanding of something is not a prerequisite for making a technology which functionally duplicates something else.
A general AI doesn't need immediate and total control to quickly wipe out humans. Consider for example that we're fast approaching the point where you can order specific proteins synthesized for you over the internet. An AI could make a deadly pathogen would be one option.
And a general AI doesn't even need to act suddenly in such a directly and obviously malicious fashion. It is highly plausible that an AI could slowly establish itself until it had functional control. http://slatestarcodex.com/2015/04/07/no-physical-substrate-no-problem/ is a decent essay which discusses examples of this.
Musk may be a visionary, but he's also a bit loony on some topics. Don't forget he believes it's a near certainty that we're all living inside a massive computer simulation.
So, this amounts to essentially an irrelevant ad hominem attack. At no point in my post did I say that Musk's expertise was important here. Moreover, there are a lot of people who take these sorts of issues seriously who aren't Musk and, if it is at all relevant, don't assign a high probability of us being in a simulation.
Frankly, I think that Musk and people like him probably overestimate the risks of artificial intelligence, but the reasons for that are substantially more subtle than these sort of very weak responses that don't grapple with their concerns and arguments. And even if one thinks as I do that they are overestimating the probability of a problem, it doesn't take a high probability for a substantial existential risk to be something that should be treated seriously. It makes more sense in this context to criticize the proposal of regulation as being a useful solution than anything else.
Right, in general, we went through a very long period where the success and speed of AI research was wildly overestimated. However, in the last few years there have been many successes which came *faster* than most people anticipated. The success of AlphaGo is an excellent example: many AI people thought that it would take much longer for a Go-playing AI to beat the top players. There's also been a large amount of use of AI systems to do clever additional research on their own. One recent example is http://recursed.blogspot.com/2017/07/using-decision-method-to-prove-new.html and this is just becoming more not less common .
It is true that we don't have human-level AI. However, we also don't in general know how close we are to human-level AI and it isn't implausible that some highly clever tweak to deep learning will have a very large impact. Moreover, an AI does not need to be human-level in all skills to pose a threat. An AI that doesn't understand poetry can still create real problems.
Moreover, and this is really important, people like Musk who are concerned about general AI don't think it is likely that it will show up tomorrow or the day after. But when we do get it, if were not ready, then we might be facing an extinction level threat. The argument goes that we need to be thinking about AI safety issues *now* before the AI arises when we may not then have the time to get it right then.
We already have anti-trust laws. The primary point of them is to break up companies that are too big, or to prevent the formation of companies that are too big. The solution is to enforce those laws seriously not to add more laws on top.
On the one hand is a lazy, incompetent and mendacious corporation and on the other hand is a lazy, incompetent and mendacious government agency. Whom should I trust?
I sympathize with your viewpoint a lot. At the same time, if everyone Blue and Red States decide to not do anything involving money going to the other we'll be in pretty bad shape. On the other hand, it looks almost like the Republicans are trying to make a tax system that disproportionately hits Blue States http://www.latimes.com/opinion/readersreact/la-ol-le-republican-tax-deductions-plan-20170619-story.html so maybe we're already at that point. Here's my suggestion: if you would have donated to this but won't because of the state, instead donate the same amount to a solar charity. The Solar Electric Light Fund http://self.org/ and Everybody Solar http://www.everybodysolar.org/ are both good options.
Airports are very public locations so there's no reasonable presumption of privacy under the 4th Amendment. This is not an unreasonable search. Now, for general policy reasons I by and large *don't want the government doing this* for what amount to privacy concerns as well as concerns about too much data being gathered with little oversight, but that doesn't make it an "unreasonable search" in any legal sense. It is possible for something to be a bad idea without it being unconstitutional.
So, I'm currently a faculty member of a major university. Prior to my most recent position I was at Birmingham Southern College (a small school in Alabama) and before that I was at the University of Maine. It is pretty clear that left-wing student protests and threats of student protest are having a real chilling effect on what schools do and what sort of speakers they invite; there's also a clear chilling effect from the right, albeit smaller. However, none of this is impacting regular education much at all. The vast majority of classes are not impacted; it is a serious mistake to think that because we have a problem with some student groups trying to push for control and censorship now that somehow there's a problem with colleges and universities as a whole. Colleges remain the primary and best way to get serious education on almost any topic; as technology and science become more advanced and more relevant to serious issues we face as a society, the importance of colleges and universities if anything has grown. Universities also remain the hotbeds of basic research, a vital aspect of a long-term healthy economy.
It is unfortunate that people are using the genuine but minor problem of student activism as an excuse to have a generally anti-intellectual position against colleges as a whole. Moreover, if one is concerned about the influence of such groups, the last thing one should want to do is to give up the colleges and universities to them wholesale. If they become purely "liberal" or "left-wing" institutions, we all lose.
Slashdot Mirror
I only interacted with him a small amount, but he seemed like a nice guy. He was clearly very dedicated to all sorts of open-source projects and his dedication was combined with a cleverness and willingness to listen to those around him. He will be missed.
I'm not originally from Massachusetts, and I've lived in a whole bunch of states (Alabama, Connecticut, Maine and now Iowa), and I can see why people from Mass have this attitude about environmental issues. It is very clear that on environmental issues both large and small, not only is Mass better than they in terms of regulations but also in terms of people simply being willing to do minor things in their day-to-day lives like reusing things rather than throwing them out, or keeping their heating and air conditioning at temperatures that reduce use, etc.
Trump tweeting whatever crap is on his mind is not transparency when government agencies are not responding to basic FOIA requests and where all sorts of basic government reports have been taken off-line or deleted; http://news.nationalgeographic.com/2017/02/wildlife-watch-usda-animal-welfare-trump-records/ is one example.
Bill Clinton, George W. Bush, and Barack Obama all had serious transparency issues and trouble paying attention to public feedback. However, neither Bush nor Obama was by and large that bad, and the current administration is so bad it makes the previous ones look like paragons of transparency and responsiveness.
Is there any evidence that anyone has attempted or intends to do this? As far as I can tell, this is essentially a conservative urban legend. And when RAND did a study on the costs of including transgender individuals they seem to have concluded that this was not at all likely https://www.rand.org/news/press/2016/06/30.html.
The old Slashdot slogan used to be "News for nerds, stuff that matters." Most Slashdot articles fall into both categories. It is rare to see an article which more so unambiguously falls into one category and not the other. Still, a very impressive feat.
So, I agree that there are specific concerns in that context, and they do require substantial examination to see how much they matter. Luckily, we also have data from other countries. For example, the IDF allows transgender individuals and has had no problems aside from some very minor accommodations they had to make. There are many medical circumstances where we don't put people into front-line combat while they are undergoing it, but that doesn't mean we don't put them in the military wholesale.
Whether we label something as a mental health problem is less relevant to whether or not it is a mental health problem that interferes with their ability to fight. Many minor mental health problems are allowed for people in the military or in security contexts as long as they don't impair ability. That's something that should be made on a judgment based on that sort of concern.
The idea isn't that people are consciously putting sexism in front of profits but rather that they are unconsciously allowing sexism to cause them to underestimate their potential profits.
The purpose of the military, is to break things and kill people. It needs to be a tight cohesive bunch, not with a bunch of "my gender of the week" types.
I don't know why you think someone having a different gender identity than their assigned at birth identity somehow must make them "weak." Heck, for that matter, even if you did have someone whose gender identity changed every week as you suggest, I don't see why that would them have trouble killing people or being part of a cohesive fighting force.
The military should be EXEMPT from all of this political correctness garbage, which, is set up on purpose, to reduce morality, and destroy the best fighting force in the world.
Or this is about people who genuinely want the right to fight for their country. You appear to be jumping from the fact that you disagree with other people to concluding that they must have nefarious goals. Incidentally, it is worth noting that the exact same arguments you are making now have now been made twice before. They were first made when we desegregated the military to have black and white people in the same units. This argument was again made only a few years ago when we let gays and lesbians openly serve. Why is it different this time?
No, the concern is that we don't know when it will happen, but we do know that it will happen eventually, and that specific aspects of what intelligent entities can do cause concern. The argument is that we need to get this right before we get the actual technologies. We also for that matter don't know how complicated long-term planning is and how much planning is essentially sophisticated pattern recognition with a few other algorithms grafted on. Everyone agrees that current systems *cannot* duplicate human cognition. That isn't the concern.
I am not talking about how much we remember (application layer) or how much we think we remember, but about the raw storage capacity of our brain (physical layer) here the capacity is enormous. Consider: to our current understanding, information is not stored in neurons but in circuits of neurons. We have 100 billion neurons and each can link to up to 10000 other neurons. I have not found any indicator as to up to how many neurons can form one circuit, but suffice to say that the possible number of circuits is very, very high.
The mistake here is going from that we could conceive of a brain using that much information storage in the most efficient form possible to thinking that it does. We know that in practice, brains engage in massive amounts of compression of stored information and reconstruct memories when they are retrieved. That's a big part of why human memory is so often wrong or distorted http://www.newyorker.com/science/maria-konnikova/idea-happened-memory-recollection. As for planes, I agree that by many metrics they are less efficient than birds, but by some metrics (maxmimum speed for example) they are substantially better. But it also isn't intended as a perfect analogy anyways; the point of both the bird example and the fire example is that one doesn't need to understand well a natural thing in order to make something similar to it.
If the last AI winter has taught us anything then that we should not expect the AI development to move very fast.
Part of the AI winter was a collapse in funding more than anything else. And in so far as the AI winter happened, it doesn't mean a similar winter will happen. It isn't infrequent for a technology to emerge at first, seem to not be progressing, and then years later to burst on the scene with little warning. Smartphones, electric cars, and solar panels are all examples.
Planes were never meant to fly "like birds", or how many wpm does the new Airbus 380 do? (wingflaps per minute).
That's the exact point though. We didn't need to duplicate exactly how a bird flies to get flying objects. But you are apparently very certain that we'll need to duplicate how a human brain thinks to get a general AI. That you find something ridiculous or label it magical doesn't make it less of a concern.
If you don't understand the human brain, what makes you so certain that changes to existing machine learning absolutely cannot result in a strong AI with a few small additional insights? For that matter, how do you possibly justify that we are "infinite far away" when we have neural networks and have developed understanding of specific brain aspects. Moreover, and most importantly, we made airplanes before we understood how birds fly. We made fire for ourselves before we understood how lightning made fire in the forests. Etc. The ability to have a perfect understanding of something is not a prerequisite for making a technology which functionally duplicates something else.
A general AI doesn't need immediate and total control to quickly wipe out humans. Consider for example that we're fast approaching the point where you can order specific proteins synthesized for you over the internet. An AI could make a deadly pathogen would be one option.
And a general AI doesn't even need to act suddenly in such a directly and obviously malicious fashion. It is highly plausible that an AI could slowly establish itself until it had functional control. http://slatestarcodex.com/2015/04/07/no-physical-substrate-no-problem/ is a decent essay which discusses examples of this.
Musk may be a visionary, but he's also a bit loony on some topics. Don't forget he believes it's a near certainty that we're all living inside a massive computer simulation.
So, this amounts to essentially an irrelevant ad hominem attack. At no point in my post did I say that Musk's expertise was important here. Moreover, there are a lot of people who take these sorts of issues seriously who aren't Musk and, if it is at all relevant, don't assign a high probability of us being in a simulation.
Frankly, I think that Musk and people like him probably overestimate the risks of artificial intelligence, but the reasons for that are substantially more subtle than these sort of very weak responses that don't grapple with their concerns and arguments. And even if one thinks as I do that they are overestimating the probability of a problem, it doesn't take a high probability for a substantial existential risk to be something that should be treated seriously. It makes more sense in this context to criticize the proposal of regulation as being a useful solution than anything else.
Right, in general, we went through a very long period where the success and speed of AI research was wildly overestimated. However, in the last few years there have been many successes which came *faster* than most people anticipated. The success of AlphaGo is an excellent example: many AI people thought that it would take much longer for a Go-playing AI to beat the top players. There's also been a large amount of use of AI systems to do clever additional research on their own. One recent example is http://recursed.blogspot.com/2017/07/using-decision-method-to-prove-new.html and this is just becoming more not less common .
It is true that we don't have human-level AI. However, we also don't in general know how close we are to human-level AI and it isn't implausible that some highly clever tweak to deep learning will have a very large impact. Moreover, an AI does not need to be human-level in all skills to pose a threat. An AI that doesn't understand poetry can still create real problems.
Moreover, and this is really important, people like Musk who are concerned about general AI don't think it is likely that it will show up tomorrow or the day after. But when we do get it, if were not ready, then we might be facing an extinction level threat. The argument goes that we need to be thinking about AI safety issues *now* before the AI arises when we may not then have the time to get it right then.
We already have anti-trust laws. The primary point of them is to break up companies that are too big, or to prevent the formation of companies that are too big. The solution is to enforce those laws seriously not to add more laws on top.
On the one hand is a lazy, incompetent and mendacious corporation and on the other hand is a lazy, incompetent and mendacious government agency. Whom should I trust?
I sympathize with your viewpoint a lot. At the same time, if everyone Blue and Red States decide to not do anything involving money going to the other we'll be in pretty bad shape. On the other hand, it looks almost like the Republicans are trying to make a tax system that disproportionately hits Blue States http://www.latimes.com/opinion/readersreact/la-ol-le-republican-tax-deductions-plan-20170619-story.html so maybe we're already at that point. Here's my suggestion: if you would have donated to this but won't because of the state, instead donate the same amount to a solar charity. The Solar Electric Light Fund http://self.org/ and Everybody Solar http://www.everybodysolar.org/ are both good options.
If you imagine it is being read by William Shatner it makes complete sense.
Airports are very public locations so there's no reasonable presumption of privacy under the 4th Amendment. This is not an unreasonable search. Now, for general policy reasons I by and large *don't want the government doing this* for what amount to privacy concerns as well as concerns about too much data being gathered with little oversight, but that doesn't make it an "unreasonable search" in any legal sense. It is possible for something to be a bad idea without it being unconstitutional.
Sigh. Telling me to go and read something is not very helpful. What specific detail are you objecting to?
Is there a specific factual issue with the summary you object to?
So, I'm currently a faculty member of a major university. Prior to my most recent position I was at Birmingham Southern College (a small school in Alabama) and before that I was at the University of Maine. It is pretty clear that left-wing student protests and threats of student protest are having a real chilling effect on what schools do and what sort of speakers they invite; there's also a clear chilling effect from the right, albeit smaller. However, none of this is impacting regular education much at all. The vast majority of classes are not impacted; it is a serious mistake to think that because we have a problem with some student groups trying to push for control and censorship now that somehow there's a problem with colleges and universities as a whole. Colleges remain the primary and best way to get serious education on almost any topic; as technology and science become more advanced and more relevant to serious issues we face as a society, the importance of colleges and universities if anything has grown. Universities also remain the hotbeds of basic research, a vital aspect of a long-term healthy economy.
It is unfortunate that people are using the genuine but minor problem of student activism as an excuse to have a generally anti-intellectual position against colleges as a whole. Moreover, if one is concerned about the influence of such groups, the last thing one should want to do is to give up the colleges and universities to them wholesale. If they become purely "liberal" or "left-wing" institutions, we all lose.