there is no way for an algorithm to evaluate progress inside the game when feedback comes only from finishing
Why not? It should be possible to train NNs with many different games to recognize a "game", to identify game controls, to recognize characteristics of a possible game objective, and to recognize signs of success. You then connect these to a NN with "memory" to attempt those possible goals until one triggers recognizable success. With repetition, that NN becomes the one that understands that game's play. It could feedback its knowledge of the game to the others so that that game is recognized in the future and doesn't require the training of a new NN to play.
Also, NNs should lean toward using outputs of existing subordinate NNs when possible instead of relearning an existing skill. This makes it easier to learn new games that are derivatives of old ones. I think that the way our brain is largely constrained to thinking in terms of patterns we can physically accomplish is key to segmenting our knowledge to make this kind of sharing of NNs more practical.
True AI will only be reached when we create a standard framework where supervisory NNs can recognize the need for a new skill, create subordinate NNs to learn a new skill, and identify valuable interconnections between subordinate NNs and make or erase them. We'll also have to design in the equivalent of wants or desires as well as dislikes to drive it to learn. At that point, it will just be a matter of time and resources before the intelligence appears.
Good question. I see microphones in widespread use for entering records and device control, but the device isn't talking back yet. And when it does, perhaps it could be routed to the communication earpieces they've started using now.
I would think the combination of an antibacterial finish, a sealed keyboard, and surface speakers (no grille required) would be vastly more effective.
Perhaps these have no speaker grills or other crevices that germs can settle in? Otherwise, the ability to use wipes on it is actually creating a danger - the danger that people may become complacent in a false belief that the device is sterile just because they constantly wipe it.
and finds that remote work will keep increasing. That's big news. It's not like surveys are so easily manipulated by the questions asked that not manipulating them is the really hard part.
Surveys performed at the request of people or organizations with a stake in the results are rarely brokered by a third party to hide the customer and are thus rarely worth the paper they are written on. Yes, I know this is electronic.
While going to work on an interstate in slushy weather, I was driving 65 in the fast lane which was separated from the oncoming lanes by a concrete barrier above headlight level. A truck in the oncoming lane hit a slush puddle and sent a wall of black slush over the barrier. It fully covered my windshield and side window. I couldn't see anything other than the fact that the car behind me got plastered too. So he couldn't see me either. I couldn't put my brakes on. I had to pray the car in front of me didn't stop, turn wipers and wash on, and wait to see again. It was a long few seconds. These things will happen.
A sensor could avoid being blinded in the same situation if it had something like eyelids with similar reflexes and speed.
The sensor processing should have a parallel path that does nothing but detect things coming at the sensor and send a signal to close the lid at just the right moment to block most of the debris. That combined with new technology to make the sensor windows hydrophobic should go a long way to keep them clean.
Redundancy is also important. I personally think they should explore audio sensing to augment the visual. It is cheap and can warn about threats that are out of sight. I prefer driving with my windows down in city environments because the noises give me a good 3D map of the traffic without taking my eyes from the road in front.
No. Civil wars are the nastiest ones. The next real civil war in an advanced nation will be fought with weapons vastly more horrible than guns. The question will be whether anyone will be left to pry the gun out of cold, dead hands.
If we stick to ships, the Prelude is six times larger than the Allure of the Seas in displacement at 600,000 tons and covers an area around 5 times that of the International Space Station at 450 tons.
This mismatch in weight vs size is the somewhat lame point the article's premise is based on.
I could change the equation dramatically by talking about the size of a blimp versus the ISS. The Hindenburg had a footprint larger than the ISS, weighed half as much, and enclosed more than 200 times the volume.
Also, certainly it is easy to argue that the ultimate size of machines in space is larger than is possible on Earth... or is it? What if we were to start completely covering Earth in shells of very strong engineered material, gradually transferring material from under the shells and building both downwards into the evacuated space and upwards as we do so, until we consume all of the Earth's material to the center of the Earth. We'd essentially be turning our planet into a vastly larger in volume, much lower density, satellite. We could throw in the moon's materials while we are at it. At that point, we suddenly see Earth as the very large machine in space it already is today.
The ratio of ads to text was vastly lower in newspapers when they were king than in web viewed news today. Also, the ads were much more tasteful because the reader connected them with the newspaper's reputation. A large portion of newspaper revenue was from the subscription fee.
I understand that I am no longer paying a subscription fee and thus there needs to be an increase in advertisements. A ratio of one ad to one article would be vastly greater than what used to be in the newspapers and is acceptable. Beyond that is absolutely ridiculous.
I'd also like to get back to the sites being responsible for the ads, but don't know how that happens.
When I came out of school, I had theory and the experience of writing (on my own) a compiler for a largish subset of Ada 83, a cross-assembler, a couple of device drivers, a few minor device firmwares, and various assundry other useless classroom exercises. That was enough to have put me on my school's team for the ACM regionals. As a Computer Engineer, I had also designed and constructed devices and could quickly make designs without having pre-made boards like Raspberry Pis in existence. But the reality of the difficulty of accomplishing a design that also met reliability, manufacturability, and other real-world goals had not yet hit.
Within days of starting work as an Associate Engineer in Avionics, I understood that I had gone to school from ages 4 to 21 to get the learner's permit that would allow me to start learning how to design and program real-world systems. It took about three more years of 90 hour weeks with unbelievable financing and toys (the lab I was in probably had over a billion dollars of hardware) to play with to reach a level that I would consider a competent systems engineer.
In that day, virtually everyone spent their first years in corporate environments that gave you a year or so as an Associate Engineer contributing nearly nothing and two or three years as an Engineer barely breaking even in the contribution to project versus cost of mistakes equation before becoming a truly useful Senior Engineer. If we've lost that level of corporate training to season the college output, we might as well quit.
So, with the Feds budgeting $200 million a year for K-12 CS at the behest of U.S. tech leaders, can't the tech giants at least put a BASIC on every phone/tablet/laptop for kids?
$200 million is a token amount to pretend that action is being taken. It is less than $4 a year for each student. There are about 56 million K-12 students in the U.S. Multiply that by at least 50 to get near $200 per year and we might start seeing "programming" take baby steps towards becoming a basic skill as it should.
"engaging, interactive, and actionable email experiences."
All of those words have to do with marketing. What user really wants their spam to be more "engaging, interactive, and actionable"?
"Actionable" especially. That is Google-speak for "the user can initiate a purchase directly from the page". This change has absolutely nothing to do with providing a feature to users.
(actually, an ex did once say in response to a stated wish for ads to be illegal with the question "but how would we know what to buy?", but that's one reason why she's an ex.)
What sounds like hell to you sounds like choice to me though I am much older than the millennials. I think it will be nice not to have to choose between the atmosphere I like or the kitchen I like. Automated delivery will likely be able to get everyone's food there at the perfect time, all preordered / scheduled before they go. And I believe with competition between eating areas no longer based on the kitchen's capabilities, the eating areas will become much nicer.
If profits are down, the walk-in traffic must be down - presumably because some are taking delivery now. This is to be expected. You get into the delivery business precisely because, if you don't, someone else will. Then you'd just be left with the declining restaurant revenue that I suspect is the real problem here.
The delivery business doesn't require the seating area, wait staff, parking lot, etc. All of that and the storefront can go away. If you want to understand your business expenses, you need to separate the books and don't put anything on the delivery business' costs that it doesn't need. That 20-40% revenue that goes to the delivery costs is its equivalent of the storefront and wait staff. It shouldn't have to shoulder both.
Eventually, I think the seating area disappears and all you have is a delivery business.
This comes full circle when someone creates restaurant seating areas with drink service but no kitchens and facilitates easy ordering from any nearby delivery kitchens. This will allow groups a place to meet and dine while allowing the individuals to order from anywhere they want. The product of the dining rooms will be space, atmosphere, wait staff for drinks and cleanup, etc. There won't be as many seats in this system because most will prefer to have their food delivered to work, home, whatever park they are sitting in, etc.
The thing is, I'd rather not ever input with anything other than voice beyond taps. And if I won't a display, it needs to be AR with some sort of wireless connection to my device, not some bulky thing I carry around and have to pull out to look at. I hate the smartphone form factor and hope it dies soon. Honestly, if I picked a feature of those I listed to eliminate altogether, the display would be it. I text, make queries, make calls, etc. using the assistant even on my smartphone.
I think people don't care about the smartwatch because it is still a glorified watch instead of what it should be. A leap in tech is needed to make it what it should be though Apple is nearing the ballpark. If Android Wear matched Apple tech, we'd be within a generation or two of the critical tech mass for smartwatches.
I'd like for it to have full-time EKG as opposed to HR, SPO2, body temperature sensor, blood sugar from sweat for the diabetics out there, a display at least as large as the ionic, Google Assistant, different vibration patterns for different reminders, LTE, WiFi, android apps, accurate GPS augmented by WiFi and accelerometers to get very accurate locations, speaker, mic, bluetooth, etc.
I'd then have everything I need in one place and could eliminate the bulky smartphone. I hate having things in pockets or on my belt. This is why it isn't being pushed. Ultimately, it could replace the more lucrative smartphone.
Frankly, I'm not sure I need the watch function though the computer has to have it.
Only issue will be that the Chromecast will only play what it understands how to play, which likely means no sound from anything in a MKV container if my experience trying to cast the screen to get around this previously is any indication, but that's not VLC's fault. Will be happy to be able to cast from it regardless.
From the article:
VideoLAN designed VLC’s Chromecast support as a workaround for any media formats Chromecast can’t natively support. VLC can re-encode video on the fly to make it work with Chromecast, including casting DVDs from your Windows machine.
I gather from the article's statement that the greatest value of the Chromecast support is precisely that any video format VLC supports can be reencoded on the fly to play on the Chromecast. So VLC is ahead of you on this score.
I am disappointed that VLC doesn't support receiving a Chromecast. It would be very nice to be able to Chromecast from my phone and other devices to a Window on one of my PC displays. This would also enable the assistant on my Google Home to audio and video to my PC in response to voice commands. Google has a tool to receive a Chromecast, but only offers it to schools. They apparently don't want to do anything to make PCs easier to use. Otherwise,
Google Assistant itself would already be a feature of Chrome on PCs.
As this is not a generalized neural net simulation but a specific one whose connections both between neurons and from neurons to sensors and actuators mirror the real world one, it is likely that they had to achieve weights and biases very close to those in the real world to get it to work. The exact same neurons would have to be actuating things in response to the same neurons sensing things as in the real world.
I see this as a validation that there are few remaining surprises in how these neurons work.
That said, I hope that their next step is to simulate the real-world learning process too and demonstrate it by starting the simulation at the single cell level and adding cells and connections in the same pattern as a real worm while mirroring its functionality throughout the process without any manual adjustments to the weights and biases during the process.
You cannot conclude by looking at current salaries vs. age that the average person younger than the peak will peak at that point! The comparison across age is an apples and oranges one. The salary environment at the time you entered engineering has a long-term effect on your salary progress.
With that in mind, it is possible that the peak indicates older engineers took a bigger hit in 2009 and that starting salaries and early career raises have been growing faster than the salaries of experienced engineers in the post-2009 recovery due to rampant age and experience discrimination in the industry. This would result in a peak that is moving up in age (recovering to the normal early-50s range) though likely not at a year for year rate.
In other words, that peak might be 46 in two more years and 47 in four as the younger beneficiaries of the salary increases age.
Some large corporations experience this as a matter of policy because they limit max pay raises of employees to 1.5x the average pay raise but have to pay the market price for new engineers. This creates a situation where it is not unheard of for an employee to receive maximum raises for years and be responsible for managing new hires who make almost as much or more than they do because the market rate has been increasing by more than 1.5x the corporation's average salary increases.
I'm actually encouraged by the slowdown in "real computing" speed increases. The answer is in different computing, not faster computing. Until the current path gets difficult, we won't go off the path and find true breakthroughs.
I once read a wonderful short story written in a future where the MPAA had won all of their battles, celebrities owned their likeness, and laws had been passed making it criminal to replicate them. The punch line was that there was no exception for dopplegangers. If I remember correctly, a violator had to either be confined to prison or have cosmetic surgery to alter their appearance so that they no longer looked like the celebrity (or themself).
Maybe someone here can supply the title. It was a nice exploration of where this path takes us.
I expect that 20 to 30 years from now we will be able to sit down at our home entertainment system and have any kind of movie or show we'd like to see custom created for us on the fly with whatever characters and storyline that it senses we want. It will monitor our reactions and do things like always surprising us at the exact moment we least expect it because it knew we had relaxed. It will be in AR or VR and nearly indistinguishable from real life.
Having to make characters that don't look like any one of the billions of people on this planet would really kill the value of these systems. People need to get over it. Imagination augmentation should be no more controlled than imagination itself.
I guess alien porn will become a leading industry - at least until the aliens complain.
Algorithms don't obey speed laws. They may base their routing on them.
Exactly. What I meant is that they don't guide you to straight and clear 30mph roads where people using their app are doing 50mph when a legal 40mph route is available. Google's algorithms do have access to the actual speed limits (and the app displays them in some areas) as opposed to just the data collected from the speed of users running their apps.
There are increasing numbers of stories about neighborhoods that are being inundated by rush-hour traffic being routed off the throughways and through their residential areas because the routing algorithm sees that as a faster path.
I've never had that happen with Google. Now, if you're using something like Waze, that overly aggressive routing is their trademark. I don't use Waze because of that. I have no desire to chance hitting a child playing in the street to save a few minutes of driving. This isn't a problem in the technology. It is a choice made by a company for profit. Many drivers make the same bad choices.
I use Google's app everywhere I go even if it is a route I travel almost every day just because I don't know what accidents may have occurred. I live in a metro area with complex roads (not well planned) and have never had it take me via a non-mainstream route with the exception of mapping me around a few accidents. It is also nice to know my ETA. I find the initial ETA of a 30-minute city ride is accurate to about plus or minus a minute almost all of the time. I'm not sure if it has adjusted to me or if I'm just that average in the way I drive.
I have noted that it often optimizes my route for right-hand turns (no waiting at the light) and sometimes routes me in a fashion that avoids a heavy traffic spotlight, especially if a jam is forecast due to the time of day. Of course, the capabilities of these apps are heavily related to how many people are using the app at the moment - as every user is also a traffic sensor.
As I understand it, other apps such as Waze have the type of routing you've described (going through neighborhoods to save a minute) as a feature. I would never use that. It is too aggressive. I have no desire to be responsible for killing some kid playing in the road in front of his home.
Slashdot Mirror
there is no way for an algorithm to evaluate progress inside the game when feedback comes only from finishing
Why not? It should be possible to train NNs with many different games to recognize a "game", to identify game controls, to recognize characteristics of a possible game objective, and to recognize signs of success. You then connect these to a NN with "memory" to attempt those possible goals until one triggers recognizable success. With repetition, that NN becomes the one that understands that game's play. It could feedback its knowledge of the game to the others so that that game is recognized in the future and doesn't require the training of a new NN to play.
Also, NNs should lean toward using outputs of existing subordinate NNs when possible instead of relearning an existing skill. This makes it easier to learn new games that are derivatives of old ones. I think that the way our brain is largely constrained to thinking in terms of patterns we can physically accomplish is key to segmenting our knowledge to make this kind of sharing of NNs more practical.
True AI will only be reached when we create a standard framework where supervisory NNs can recognize the need for a new skill, create subordinate NNs to learn a new skill, and identify valuable interconnections between subordinate NNs and make or erase them. We'll also have to design in the equivalent of wants or desires as well as dislikes to drive it to learn. At that point, it will just be a matter of time and resources before the intelligence appears.
Good question. I see microphones in widespread use for entering records and device control, but the device isn't talking back yet. And when it does, perhaps it could be routed to the communication earpieces they've started using now.
I would think the combination of an antibacterial finish, a sealed keyboard, and surface speakers (no grille required) would be vastly more effective.
Perhaps these have no speaker grills or other crevices that germs can settle in? Otherwise, the ability to use wipes on it is actually creating a danger - the danger that people may become complacent in a false belief that the device is sterile just because they constantly wipe it.
and finds that remote work will keep increasing. That's big news. It's not like surveys are so easily manipulated by the questions asked that not manipulating them is the really hard part.
Surveys performed at the request of people or organizations with a stake in the results are rarely brokered by a third party to hide the customer and are thus rarely worth the paper they are written on. Yes, I know this is electronic.
Found a very brief statement.
While going to work on an interstate in slushy weather, I was driving 65 in the fast lane which was separated from the oncoming lanes by a concrete barrier above headlight level. A truck in the oncoming lane hit a slush puddle and sent a wall of black slush over the barrier. It fully covered my windshield and side window. I couldn't see anything other than the fact that the car behind me got plastered too. So he couldn't see me either. I couldn't put my brakes on. I had to pray the car in front of me didn't stop, turn wipers and wash on, and wait to see again. It was a long few seconds. These things will happen.
A sensor could avoid being blinded in the same situation if it had something like eyelids with similar reflexes and speed.
The sensor processing should have a parallel path that does nothing but detect things coming at the sensor and send a signal to close the lid at just the right moment to block most of the debris. That combined with new technology to make the sensor windows hydrophobic should go a long way to keep them clean.
Redundancy is also important. I personally think they should explore audio sensing to augment the visual. It is cheap and can warn about threats that are out of sight. I prefer driving with my windows down in city environments because the noises give me a good 3D map of the traffic without taking my eyes from the road in front.
No. Civil wars are the nastiest ones. The next real civil war in an advanced nation will be fought with weapons vastly more horrible than guns. The question will be whether anyone will be left to pry the gun out of cold, dead hands.
If we stick to ships, the Prelude is six times larger than the Allure of the Seas in displacement at 600,000 tons and covers an area around 5 times that of the International Space Station at 450 tons.
This mismatch in weight vs size is the somewhat lame point the article's premise is based on.
I could change the equation dramatically by talking about the size of a blimp versus the ISS. The Hindenburg had a footprint larger than the ISS, weighed half as much, and enclosed more than 200 times the volume.
Also, certainly it is easy to argue that the ultimate size of machines in space is larger than is possible on Earth... or is it? What if we were to start completely covering Earth in shells of very strong engineered material, gradually transferring material from under the shells and building both downwards into the evacuated space and upwards as we do so, until we consume all of the Earth's material to the center of the Earth. We'd essentially be turning our planet into a vastly larger in volume, much lower density, satellite. We could throw in the moon's materials while we are at it. At that point, we suddenly see Earth as the very large machine in space it already is today.
The ratio of ads to text was vastly lower in newspapers when they were king than in web viewed news today. Also, the ads were much more tasteful because the reader connected them with the newspaper's reputation. A large portion of newspaper revenue was from the subscription fee.
I understand that I am no longer paying a subscription fee and thus there needs to be an increase in advertisements. A ratio of one ad to one article would be vastly greater than what used to be in the newspapers and is acceptable. Beyond that is absolutely ridiculous.
I'd also like to get back to the sites being responsible for the ads, but don't know how that happens.
When I came out of school, I had theory and the experience of writing (on my own) a compiler for a largish subset of Ada 83, a cross-assembler, a couple of device drivers, a few minor device firmwares, and various assundry other useless classroom exercises. That was enough to have put me on my school's team for the ACM regionals. As a Computer Engineer, I had also designed and constructed devices and could quickly make designs without having pre-made boards like Raspberry Pis in existence. But the reality of the difficulty of accomplishing a design that also met reliability, manufacturability, and other real-world goals had not yet hit.
Within days of starting work as an Associate Engineer in Avionics, I understood that I had gone to school from ages 4 to 21 to get the learner's permit that would allow me to start learning how to design and program real-world systems. It took about three more years of 90 hour weeks with unbelievable financing and toys (the lab I was in probably had over a billion dollars of hardware) to play with to reach a level that I would consider a competent systems engineer.
In that day, virtually everyone spent their first years in corporate environments that gave you a year or so as an Associate Engineer contributing nearly nothing and two or three years as an Engineer barely breaking even in the contribution to project versus cost of mistakes equation before becoming a truly useful Senior Engineer. If we've lost that level of corporate training to season the college output, we might as well quit.
So, with the Feds budgeting $200 million a year for K-12 CS at the behest of U.S. tech leaders, can't the tech giants at least put a BASIC on every phone/tablet/laptop for kids?
$200 million is a token amount to pretend that action is being taken. It is less than $4 a year for each student. There are about 56 million K-12 students in the U.S. Multiply that by at least 50 to get near $200 per year and we might start seeing "programming" take baby steps towards becoming a basic skill as it should.
but the computer is apparently perfectly capable of being the fall guy.
"engaging, interactive, and actionable email experiences."
All of those words have to do with marketing. What user really wants their spam to be more "engaging, interactive, and actionable"? "Actionable" especially. That is Google-speak for "the user can initiate a purchase directly from the page". This change has absolutely nothing to do with providing a feature to users.
(actually, an ex did once say in response to a stated wish for ads to be illegal with the question "but how would we know what to buy?", but that's one reason why she's an ex.)
What sounds like hell to you sounds like choice to me though I am much older than the millennials. I think it will be nice not to have to choose between the atmosphere I like or the kitchen I like. Automated delivery will likely be able to get everyone's food there at the perfect time, all preordered / scheduled before they go. And I believe with competition between eating areas no longer based on the kitchen's capabilities, the eating areas will become much nicer.
If profits are down, the walk-in traffic must be down - presumably because some are taking delivery now. This is to be expected. You get into the delivery business precisely because, if you don't, someone else will. Then you'd just be left with the declining restaurant revenue that I suspect is the real problem here.
The delivery business doesn't require the seating area, wait staff, parking lot, etc. All of that and the storefront can go away. If you want to understand your business expenses, you need to separate the books and don't put anything on the delivery business' costs that it doesn't need. That 20-40% revenue that goes to the delivery costs is its equivalent of the storefront and wait staff. It shouldn't have to shoulder both.
Eventually, I think the seating area disappears and all you have is a delivery business.
This comes full circle when someone creates restaurant seating areas with drink service but no kitchens and facilitates easy ordering from any nearby delivery kitchens. This will allow groups a place to meet and dine while allowing the individuals to order from anywhere they want. The product of the dining rooms will be space, atmosphere, wait staff for drinks and cleanup, etc. There won't be as many seats in this system because most will prefer to have their food delivered to work, home, whatever park they are sitting in, etc.
The thing is, I'd rather not ever input with anything other than voice beyond taps. And if I won't a display, it needs to be AR with some sort of wireless connection to my device, not some bulky thing I carry around and have to pull out to look at. I hate the smartphone form factor and hope it dies soon. Honestly, if I picked a feature of those I listed to eliminate altogether, the display would be it. I text, make queries, make calls, etc. using the assistant even on my smartphone.
I think people don't care about the smartwatch because it is still a glorified watch instead of what it should be. A leap in tech is needed to make it what it should be though Apple is nearing the ballpark. If Android Wear matched Apple tech, we'd be within a generation or two of the critical tech mass for smartwatches.
I'd like for it to have full-time EKG as opposed to HR, SPO2, body temperature sensor, blood sugar from sweat for the diabetics out there, a display at least as large as the ionic, Google Assistant, different vibration patterns for different reminders, LTE, WiFi, android apps, accurate GPS augmented by WiFi and accelerometers to get very accurate locations, speaker, mic, bluetooth, etc.
I'd then have everything I need in one place and could eliminate the bulky smartphone. I hate having things in pockets or on my belt. This is why it isn't being pushed. Ultimately, it could replace the more lucrative smartphone.
Frankly, I'm not sure I need the watch function though the computer has to have it.
Only issue will be that the Chromecast will only play what it understands how to play, which likely means no sound from anything in a MKV container if my experience trying to cast the screen to get around this previously is any indication, but that's not VLC's fault. Will be happy to be able to cast from it regardless.
From the article:
VideoLAN designed VLC’s Chromecast support as a workaround for any media formats Chromecast can’t natively support. VLC can re-encode video on the fly to make it work with Chromecast, including casting DVDs from your Windows machine.
I gather from the article's statement that the greatest value of the Chromecast support is precisely that any video format VLC supports can be reencoded on the fly to play on the Chromecast. So VLC is ahead of you on this score.
I am disappointed that VLC doesn't support receiving a Chromecast. It would be very nice to be able to Chromecast from my phone and other devices to a Window on one of my PC displays. This would also enable the assistant on my Google Home to audio and video to my PC in response to voice commands. Google has a tool to receive a Chromecast, but only offers it to schools. They apparently don't want to do anything to make PCs easier to use. Otherwise, Google Assistant itself would already be a feature of Chrome on PCs.
As this is not a generalized neural net simulation but a specific one whose connections both between neurons and from neurons to sensors and actuators mirror the real world one, it is likely that they had to achieve weights and biases very close to those in the real world to get it to work. The exact same neurons would have to be actuating things in response to the same neurons sensing things as in the real world.
I see this as a validation that there are few remaining surprises in how these neurons work.
That said, I hope that their next step is to simulate the real-world learning process too and demonstrate it by starting the simulation at the single cell level and adding cells and connections in the same pattern as a real worm while mirroring its functionality throughout the process without any manual adjustments to the weights and biases during the process.
You cannot conclude by looking at current salaries vs. age that the average person younger than the peak will peak at that point! The comparison across age is an apples and oranges one. The salary environment at the time you entered engineering has a long-term effect on your salary progress.
With that in mind, it is possible that the peak indicates older engineers took a bigger hit in 2009 and that starting salaries and early career raises have been growing faster than the salaries of experienced engineers in the post-2009 recovery due to rampant age and experience discrimination in the industry. This would result in a peak that is moving up in age (recovering to the normal early-50s range) though likely not at a year for year rate.
In other words, that peak might be 46 in two more years and 47 in four as the younger beneficiaries of the salary increases age.
Some large corporations experience this as a matter of policy because they limit max pay raises of employees to 1.5x the average pay raise but have to pay the market price for new engineers. This creates a situation where it is not unheard of for an employee to receive maximum raises for years and be responsible for managing new hires who make almost as much or more than they do because the market rate has been increasing by more than 1.5x the corporation's average salary increases.
I'm actually encouraged by the slowdown in "real computing" speed increases. The answer is in different computing, not faster computing. Until the current path gets difficult, we won't go off the path and find true breakthroughs.
I once read a wonderful short story written in a future where the MPAA had won all of their battles, celebrities owned their likeness, and laws had been passed making it criminal to replicate them. The punch line was that there was no exception for dopplegangers. If I remember correctly, a violator had to either be confined to prison or have cosmetic surgery to alter their appearance so that they no longer looked like the celebrity (or themself).
Maybe someone here can supply the title. It was a nice exploration of where this path takes us.
I expect that 20 to 30 years from now we will be able to sit down at our home entertainment system and have any kind of movie or show we'd like to see custom created for us on the fly with whatever characters and storyline that it senses we want. It will monitor our reactions and do things like always surprising us at the exact moment we least expect it because it knew we had relaxed. It will be in AR or VR and nearly indistinguishable from real life.
Having to make characters that don't look like any one of the billions of people on this planet would really kill the value of these systems. People need to get over it. Imagination augmentation should be no more controlled than imagination itself.
I guess alien porn will become a leading industry - at least until the aliens complain.
Algorithms don't obey speed laws. They may base their routing on them.
Exactly. What I meant is that they don't guide you to straight and clear 30mph roads where people using their app are doing 50mph when a legal 40mph route is available. Google's algorithms do have access to the actual speed limits (and the app displays them in some areas) as opposed to just the data collected from the speed of users running their apps.
There are increasing numbers of stories about neighborhoods that are being inundated by rush-hour traffic being routed off the throughways and through their residential areas because the routing algorithm sees that as a faster path.
I've never had that happen with Google. Now, if you're using something like Waze, that overly aggressive routing is their trademark. I don't use Waze because of that. I have no desire to chance hitting a child playing in the street to save a few minutes of driving. This isn't a problem in the technology. It is a choice made by a company for profit. Many drivers make the same bad choices.
I use Google's app everywhere I go even if it is a route I travel almost every day just because I don't know what accidents may have occurred. I live in a metro area with complex roads (not well planned) and have never had it take me via a non-mainstream route with the exception of mapping me around a few accidents. It is also nice to know my ETA. I find the initial ETA of a 30-minute city ride is accurate to about plus or minus a minute almost all of the time. I'm not sure if it has adjusted to me or if I'm just that average in the way I drive.
I have noted that it often optimizes my route for right-hand turns (no waiting at the light) and sometimes routes me in a fashion that avoids a heavy traffic spotlight, especially if a jam is forecast due to the time of day. Of course, the capabilities of these apps are heavily related to how many people are using the app at the moment - as every user is also a traffic sensor.
As I understand it, other apps such as Waze have the type of routing you've described (going through neighborhoods to save a minute) as a feature. I would never use that. It is too aggressive. I have no desire to be responsible for killing some kid playing in the road in front of his home.