Variants of this problem have been along for a very long time. It was popular as a recreational game in the 1930s and is now used more with Middle School students as a way of getting them more familiar with different operations.
If one uses instead of 4 uses 1 and has a restricted operation set or the like then you can get some actually non-trivial math by asking how many 1s you need. For example, one can define the integer complexity of a number n, denoted by ||n|| as the minimum number of 1s needed to write n as a product or sum of 1s with any number of parentheses. Thus for example, 6=(1+1)(1+1+1) shows that ||n|| 1 one has 3log_3 n - a better result is actually known that what is in that post, and I'm writing it up now. The other person mentioned there is Harry Altman who probably has thought more about different notions of complexity of numbers than anyone else at this point (his dissertation was on the subject). We had a joint paper https://arxiv.org/abs/1207.4841 that is mostly readable.
Another neat variant of this is again looking at 1s and allowing just addition, but allow that once you have made a number you can use it again, and now you count how many operations you have used. So for example, in this framework, you can use 3 additions to get 6 because 3=1+1+1 and 6=3+3. This is the addition chain complexity https://en.wikipedia.org/wiki/Addition_chain and is closely related to how to quickly exponentiate objects (such as matrices, or specific numbers mod another group) https://en.wikipedia.org/wiki/Addition-chain_exponentiation which is very important for doing a lot of practical algorithms efficiently (such as RSA and Diffie-Hellman).
The problem in the original post is essentially silly but it connects to a lot of neat, serious math. (Also, Numberphile is in general great.)
SpaceX has had this sort of launch rate as a goal for years now. Repeated delays and two rocket failures, CRS-7 https://en.wikipedia.org/wiki/SpaceX_CRS-7 and last year's on the ground explosion of a Falcon 9 have both slowed things down a lot. They are right now cheaper and more cutting edge than pretty much every other medium payload launcher, but the pressure of that sort of launch schedule is going to be very tough. And we're already seeing the expected slippage- CRS-10 was originally scheduled for late January, then got rescheduled to Feb 14 and is now on Feb 17. Given that they only have two launch facilities and they share personnel, delays for any given launch can easily start eating into prep time for later launches, bumping them even further.
We know at least one form of general intelligence that works under the laws of physics: humans. Why are you then certain that other general intelligences cannot be constructed?
Jumping to insults doesn't really help this conversation and makes it seem unlikely to be productive. I agree that current AI do not determine their own goals; that's not the concern. When AI can do so is when we'll have a problem, and a big part of the worry is that that might happen very suddenly. I recommend reading Nick Bostrom's book "Superintelligence" which discusses this and related topics in detail. Insults do not make a problem go away.
There's been a lot written on this. I recommend Bostrom's book "Superintelligence" for a start. The short answer is that a sufficiently smart AI with a goal set that doesn't include human existence or happy humans is a problem. The classic thought example is an AI programmed to maximize the number of paperclips which responds by taking over the planet and turning everything into paperclips.
Missing the point: yes, these are weak AIs. But if almost all tasks that humans can do, a weak AI can do better, than it is all the more reason to expect that the first general AIs when they arise will quickly be so much smarter than humans that their goals will be all that effectively matters.
Nothing in my comment said anything about whether recent improvements were due more to hardware improvements or algorithms. But note that if anything, you should find that hardware improvements doing just this to be scarier; that means we might not even need any big breakthroughs. Simple, steady improvement can create real problems.
While I do know this already, I don't think this is by itself bad. Estimating exactly which knowledge should be taken for granted can be difficult to estimate; for example, a while ago there was an article here about P != NP (the biggest open problem in theoretical computer science) and there were a whole host of comments that essentially amounted to people declaring that they had no idea what the article was talking about.
Poker has for a long time been a game that was considered very difficult for AIs to do. We're now in a situation where very rapidly many things that we think of as hard problems for AI (playing poker, playing Go, image recognition, translation) are having AI close to equal or surpass humans. That should be concerning at multiple levels: first, this will have large-scale economic impacts. Second, and potentially more disturbingly, it means that we are closer to the point where AI may pose an existential risk to humans, and that tipping point could occur with very little warning.
On the one hand, Trump is so terribly awful that any sort of support of him is problematic and dangerous. It risks becoming swept up in his complete disregard for facts or decency. On the other hand, having a few people involved like Musk are very sane and wealthy enough that Trump will listen to them (wealth being apparently one of the very few things he cares about), and might be positive moderating influences.
Essentially in agreement; my point was primarily just about the claim that because things are transitioning for economic reasons that we can avoid serious work. That's clearly false.
No; that's misleading at a fundamental level: there are in many circumstances market advantages. That doesn't mean that the changes will happen quickly enough to avoid serious damage. The timing here is extremely critical, and keeping running older fossil fuel plants and putting out massive amounts of CO2 will screw us over even if the slow market progression would make us carbon neutral in 2050.
Luckily for them, alternative energy sources and low CO2 energy sources like natural gas are much cheaper now than they were previously. Solar is replacing almost all the retiring coal in Texas and this is for primarily economic rather than environmental reasons http://breakingenergy.com/2016/06/06/solar-will-replace-nearly-all-retiring-coal-in-texas/. Moreover, the people who are unhappy about paying more at the pump and on their monthly electric bill don't realize that they are really paying a lot more for coal and oil in terms of pollution caused and other issues that aren't directly in the price. Getting them to understand that last bit though seems hard.
No one is suggesting that we run superconducting lines everywhere. They work best in specific types of locations where the total amount of electric power going through a very tiny area is very small; that's why they choose to use superconducting lines for Tres Amigas. Holbrook itself was more experimental but it works fine. Note also that the refrigeration equipment actually scales pretty well, so the amount of helium you need scales at a much less than linear rate, so even if one did want to make continent wide superconducting lines (which we're not suggesting anyone should plan on doing with the current technology) it would be substantially easier than your numbers suggest.
In general, we use much less power during night-time., but some solar systems (in particular, solar-thermal systems) provide power a few hours into the night. We're also getting better with storage and transmission also which helps, because one can then not only store solar power for use when the sun is not out, but also move power from areas where it is still out to where it is. High-voltage DC is really great for this, and we're also starting to have superconducting transmission lines like the Holbrook line https://en.wikipedia.org/wiki/Holbrook_Superconductor_Project and the planned Tres Amigas Superstation which will link the three major US grids (East, West and Texas) https://en.wikipedia.org/wiki/Tres_Amigas_SuperStation. Honestly, the fact that we have superconducting power lines falls into the how-is-this-not-a-bigger-deal-we-live-in-the-freaking-future. Any long-term energy plan will of course not use just solar, but likely solar, with some storage and some amount of wind, nuclear, geothermal, hydroelectric, with possibly some natural gas for quick spin-up during high load periods or when there's an unexpected drop-off in the power level. But it does look plausible at this point that a grid where the largest power source is solar is doable and may happen for primarily direct economic reasons even without the environmental considerations.
On the contrary, this is exactly why I trust serious journalistic sources like the Washington Post. They sometimes get things wrong, but when they do get it seriously wrong, they own up to it. This is what makes so-called mainstream media so great: they aren't perfect but they own up to their mistakes.
In fairness to NIST, there's no evidence that had any reason to think that the elliptic curve encryption supplied by the NSA had a backdoor. In the past, the NSA had been highly helpful without pulling that sort of junk. You are correct to note that things might change under a Trump administration.
As I replied to arth1, this seems to be more of a language issue in what one means by differentiate then. Radiation works by damaging lots of cells and using that the cancer cells have a tougher time repairing themselves. So the differentiating is effectively done by the difference in repair ability.
It seems like this is an issue of interpretation of English rather than a substantial issue then. Damaging lots of cells and relying on the fact that the cancer cells are bad at repair seems to me to be effectively differentiating between cancer cells and healthy cells.
How do you intend to use this to fight cancer? Yes, you might be able to use it as a delivery mechanism, but there's no easy way to differentiate cancer cells from regular cells. Nothing here jumps out as doing something different that's relevant. Most things we use on cancer (radiation, and chemo) work by differentiating between regular cells and cancer cells, generally using the fact that cancer cells are always reproducing. No aspect of this process has anything to do with cell reproduction. This isn't even in the petri dish category https://xkcd.com/1217/.
Even space travel they have focused on trying to do it cleanly. A big part of why their next generation of engines, the Raptor, uses methane as a fuel is that in the long-run one can synthesize methane directly and a straightforward way https://en.wikipedia.org/wiki/Sabatier_reaction. This has both an advantage in terms of Mars (can make more fuel on Mars) and also in terms of eventually making clean fuel on Earth.
This isn't really how the history went The Democratic-Republican party wasn't really connected to the Republican party at all. The Constitution was written before any political parties existed at all, and they didn't originally intend for their to be political parties. And in the pre Civil-War era, the Republicans were primarily in the North, which was the area which had less proportional strength from the electoral college.
Slashdot Mirror
Yes, they meant positive integer. And yes, one can. Watch the video. In fact one only needs logs, division and square root operations to do it.
Variants of this problem have been along for a very long time. It was popular as a recreational game in the 1930s and is now used more with Middle School students as a way of getting them more familiar with different operations.
If one uses instead of 4 uses 1 and has a restricted operation set or the like then you can get some actually non-trivial math by asking how many 1s you need. For example, one can define the integer complexity of a number n, denoted by ||n|| as the minimum number of 1s needed to write n as a product or sum of 1s with any number of parentheses. Thus for example, 6=(1+1)(1+1+1) shows that ||n|| 1 one has 3log_3 n - a better result is actually known that what is in that post, and I'm writing it up now. The other person mentioned there is Harry Altman who probably has thought more about different notions of complexity of numbers than anyone else at this point (his dissertation was on the subject). We had a joint paper https://arxiv.org/abs/1207.4841 that is mostly readable.
Another neat variant of this is again looking at 1s and allowing just addition, but allow that once you have made a number you can use it again, and now you count how many operations you have used. So for example, in this framework, you can use 3 additions to get 6 because 3=1+1+1 and 6=3+3. This is the addition chain complexity https://en.wikipedia.org/wiki/Addition_chain and is closely related to how to quickly exponentiate objects (such as matrices, or specific numbers mod another group) https://en.wikipedia.org/wiki/Addition-chain_exponentiation which is very important for doing a lot of practical algorithms efficiently (such as RSA and Diffie-Hellman).
The problem in the original post is essentially silly but it connects to a lot of neat, serious math. (Also, Numberphile is in general great.)
SpaceX has had this sort of launch rate as a goal for years now. Repeated delays and two rocket failures, CRS-7 https://en.wikipedia.org/wiki/SpaceX_CRS-7 and last year's on the ground explosion of a Falcon 9 have both slowed things down a lot. They are right now cheaper and more cutting edge than pretty much every other medium payload launcher, but the pressure of that sort of launch schedule is going to be very tough. And we're already seeing the expected slippage- CRS-10 was originally scheduled for late January, then got rescheduled to Feb 14 and is now on Feb 17. Given that they only have two launch facilities and they share personnel, delays for any given launch can easily start eating into prep time for later launches, bumping them even further.
We know at least one form of general intelligence that works under the laws of physics: humans. Why are you then certain that other general intelligences cannot be constructed?
Jumping to insults doesn't really help this conversation and makes it seem unlikely to be productive. I agree that current AI do not determine their own goals; that's not the concern. When AI can do so is when we'll have a problem, and a big part of the worry is that that might happen very suddenly. I recommend reading Nick Bostrom's book "Superintelligence" which discusses this and related topics in detail. Insults do not make a problem go away.
There's been a lot written on this. I recommend Bostrom's book "Superintelligence" for a start. The short answer is that a sufficiently smart AI with a goal set that doesn't include human existence or happy humans is a problem. The classic thought example is an AI programmed to maximize the number of paperclips which responds by taking over the planet and turning everything into paperclips.
Missing the point: yes, these are weak AIs. But if almost all tasks that humans can do, a weak AI can do better, than it is all the more reason to expect that the first general AIs when they arise will quickly be so much smarter than humans that their goals will be all that effectively matters.
Nothing in my comment said anything about whether recent improvements were due more to hardware improvements or algorithms. But note that if anything, you should find that hardware improvements doing just this to be scarier; that means we might not even need any big breakthroughs. Simple, steady improvement can create real problems.
While I do know this already, I don't think this is by itself bad. Estimating exactly which knowledge should be taken for granted can be difficult to estimate; for example, a while ago there was an article here about P != NP (the biggest open problem in theoretical computer science) and there were a whole host of comments that essentially amounted to people declaring that they had no idea what the article was talking about.
Poker has for a long time been a game that was considered very difficult for AIs to do. We're now in a situation where very rapidly many things that we think of as hard problems for AI (playing poker, playing Go, image recognition, translation) are having AI close to equal or surpass humans. That should be concerning at multiple levels: first, this will have large-scale economic impacts. Second, and potentially more disturbingly, it means that we are closer to the point where AI may pose an existential risk to humans, and that tipping point could occur with very little warning.
On the one hand, Trump is so terribly awful that any sort of support of him is problematic and dangerous. It risks becoming swept up in his complete disregard for facts or decency. On the other hand, having a few people involved like Musk are very sane and wealthy enough that Trump will listen to them (wealth being apparently one of the very few things he cares about), and might be positive moderating influences.
Essentially in agreement; my point was primarily just about the claim that because things are transitioning for economic reasons that we can avoid serious work. That's clearly false.
No; that's misleading at a fundamental level: there are in many circumstances market advantages. That doesn't mean that the changes will happen quickly enough to avoid serious damage. The timing here is extremely critical, and keeping running older fossil fuel plants and putting out massive amounts of CO2 will screw us over even if the slow market progression would make us carbon neutral in 2050.
Luckily for them, alternative energy sources and low CO2 energy sources like natural gas are much cheaper now than they were previously. Solar is replacing almost all the retiring coal in Texas and this is for primarily economic rather than environmental reasons http://breakingenergy.com/2016/06/06/solar-will-replace-nearly-all-retiring-coal-in-texas/. Moreover, the people who are unhappy about paying more at the pump and on their monthly electric bill don't realize that they are really paying a lot more for coal and oil in terms of pollution caused and other issues that aren't directly in the price. Getting them to understand that last bit though seems hard.
Contrast this with the incoming administration which wants to favor fossil fuels above pretty much all else https://www.nytimes.com/2016/05/27/us/politics/donald-trump-global-warming-energy-policy.html, http://www.nature.com/news/trump-s-next-move-scientists-struggle-with-foggy-future-1.21339.
No one is suggesting that we run superconducting lines everywhere. They work best in specific types of locations where the total amount of electric power going through a very tiny area is very small; that's why they choose to use superconducting lines for Tres Amigas. Holbrook itself was more experimental but it works fine. Note also that the refrigeration equipment actually scales pretty well, so the amount of helium you need scales at a much less than linear rate, so even if one did want to make continent wide superconducting lines (which we're not suggesting anyone should plan on doing with the current technology) it would be substantially easier than your numbers suggest.
In general, we use much less power during night-time., but some solar systems (in particular, solar-thermal systems) provide power a few hours into the night. We're also getting better with storage and transmission also which helps, because one can then not only store solar power for use when the sun is not out, but also move power from areas where it is still out to where it is. High-voltage DC is really great for this, and we're also starting to have superconducting transmission lines like the Holbrook line https://en.wikipedia.org/wiki/Holbrook_Superconductor_Project and the planned Tres Amigas Superstation which will link the three major US grids (East, West and Texas) https://en.wikipedia.org/wiki/Tres_Amigas_SuperStation. Honestly, the fact that we have superconducting power lines falls into the how-is-this-not-a-bigger-deal-we-live-in-the-freaking-future. Any long-term energy plan will of course not use just solar, but likely solar, with some storage and some amount of wind, nuclear, geothermal, hydroelectric, with possibly some natural gas for quick spin-up during high load periods or when there's an unexpected drop-off in the power level. But it does look plausible at this point that a grid where the largest power source is solar is doable and may happen for primarily direct economic reasons even without the environmental considerations.
On the contrary, this is exactly why I trust serious journalistic sources like the Washington Post. They sometimes get things wrong, but when they do get it seriously wrong, they own up to it. This is what makes so-called mainstream media so great: they aren't perfect but they own up to their mistakes.
In fairness to NIST, there's no evidence that had any reason to think that the elliptic curve encryption supplied by the NSA had a backdoor. In the past, the NSA had been highly helpful without pulling that sort of junk. You are correct to note that things might change under a Trump administration.
As I replied to arth1, this seems to be more of a language issue in what one means by differentiate then. Radiation works by damaging lots of cells and using that the cancer cells have a tougher time repairing themselves. So the differentiating is effectively done by the difference in repair ability.
It seems like this is an issue of interpretation of English rather than a substantial issue then. Damaging lots of cells and relying on the fact that the cancer cells are bad at repair seems to me to be effectively differentiating between cancer cells and healthy cells.
That's a good expansion of how it works; how does it disagree with anything I said?
How do you intend to use this to fight cancer? Yes, you might be able to use it as a delivery mechanism, but there's no easy way to differentiate cancer cells from regular cells. Nothing here jumps out as doing something different that's relevant. Most things we use on cancer (radiation, and chemo) work by differentiating between regular cells and cancer cells, generally using the fact that cancer cells are always reproducing. No aspect of this process has anything to do with cell reproduction. This isn't even in the petri dish category https://xkcd.com/1217/.
Even space travel they have focused on trying to do it cleanly. A big part of why their next generation of engines, the Raptor, uses methane as a fuel is that in the long-run one can synthesize methane directly and a straightforward way https://en.wikipedia.org/wiki/Sabatier_reaction. This has both an advantage in terms of Mars (can make more fuel on Mars) and also in terms of eventually making clean fuel on Earth.
This isn't really how the history went The Democratic-Republican party wasn't really connected to the Republican party at all. The Constitution was written before any political parties existed at all, and they didn't originally intend for their to be political parties. And in the pre Civil-War era, the Republicans were primarily in the North, which was the area which had less proportional strength from the electoral college.