Some "journalist" wanted to write a story about hacking, but, you know, different. They then located some guy who builds sensors who was willing to wax poetic about systems engineering, and voila.
Python is one of the few current languages that you can learn as a purely procedural language, but also has a sane standard object system you can pick up later, without having to completely switch to a new language. As an extra benefit it's widely used.
The OOP extensions of Pascal were also like that and, not coincidentally, were the standard learning languages for a long time.
When Swift was first announced it sounded like it was supposed to be Apple's version of Python. I was disappointed when I first saw it. It's basically C with some sane object oriented extensions (i.e. not C++) and some random unnecessary stuff like those let keywords.
I had a similar encounter with an old database designed by amateurs in FileMaker. I had to acquire a copy of an old version to load the file, then upgrade it to a later version that had SQL export.
It's an interesting question how much complexity in the elements (synapses and neurons) versus complexity in the network, is required. For the synapses, humans have fairly complicated ones, especially if you consider every type of neuron, but there are other animals that are considerably simpler. Some use only a single neurotransmitter.
We hear a lot about the simple elements/complex network approaches at the moment because they're making a lot of progress. People working from the other side haven't been as successful in terms of solving actual problems, although some of the models have shown interesting hints of self-organizing and emergent behaviour.
Personally I think that a lot of the biological complexity is imposed by biological limitations and the good enough nature of evolution, but it's quite possible there are more basic structural secrets to discover.
That was a pretty good idea in the 80s when Penrose proposed it, but it didn't really pan out. Other variations (and honestly, Penrose's work too) really sound pretty hokey, and many rely on assuming that wavefunction collapse (a) happens and (b) happens only under conscious observation as justification.
There are studies that show complicated Excel spreadsheets are so buggy (and hard to debug) that they're not really any better than random number generators.
You're correct: Excel has too many features, and it makes things deceptively easy.
If it scales up well, you could make a chip with much greater capacity than you get with current deep learning techniques. Simulating this kind of thing on a digital chip isn't really very efficient, in storage, circuits, or speed.
I'm curious, what money would you make on Mars that you couldn't make on the moon?
The moon is close enough and small enough that you might be able to mine it and return the result to Earth economically if you could find anything valuable there, or could potentially be a practical location for dirty industries if environmental costs on Earth got high enough.
It's certainly where you'd want to supply any kind of Earth orbital industry from, and has lots of the kinds of materials you'd need for that: oxygen, silicates, probably water. It would be the source of raw materials for building large space infrastructure: habitats, orbital rings, solar plants, interplanetary/interstellar ships, launch lasers, etc. With low gravity and vacuum it's also a good place to build really big telescopes, particle accelerators, launch loops, space elevators and other such things.
Mars seems primarily interesting for the possibility of extinct life, but I'm not sure it has any real advantages over the moon or asteroids for resources, or living space.
You're quite a good troll. Not only are you pretty entertaining, but what you say is usually technically correct, just superficial, and calculated to grate on the Slashdot groupthink. I expect you probably post quite a bit of insightful stuff under a different name.
"'Deep' learning is 'deeper' than shallow learning". Hard to argue with that, except you convinced a bunch of random Slashdotters to try.
There's a baby in a carriage, sleeping. The mother or nanny is on her phone. That's not terribly different than the past, except for the phone.
A couple of kids in the 2-6 range, one of which is playing on a phone, the other on a tablet. Mom/nanny is on her phone.
A minivan just pulled up outside. I can't see how many kids are in the back, but I can see that it has a ceiling mounted DVD player.
I commend you on the way you're raising your kids. If I have kids I intend to do my best to do the same. But that doesn't seem to be the reality for most people. Television, educational toys, apps to help with homework, cell phones for entertaining and keeping tabs on kids, tracking bracelets, child rearing hasn't been overlooked by tech. And all those things are still pretty crude. It's easy to see how that might go further in the future: the Japanese are madly trying to develop elder care robots to get them out of their demographic crisis, but the same technology has long been envisioned by science fiction for raising children.
I suspect eventually we will come to our senses, spend a small amount of our time working and devote much of the rest to pursuits that we enjoy. One of the most popular may be raising kids. In the meantime, I suspect that people will continue to spend less time raising their kids and make up the increasing shortfall using technology.
I'd put my bet on artist, musician, poet, or similar. The society we live in is a system adapted to the exponential growth and political conditions existing around the allocation of capital that prevailed in the twentieth century. It's going to change, and I strongly suspect that questions like "what fields should we reskill workers into?" are going to sound ridiculous a generation or two down the line.
Someone I know posted a diagram of the feudal system on Facebook today. It shows food flowing down to the peasants from the aristocracy. That's exactly what the aristocracy likes the peasants to think... never mind that the only source of food is the farms that the peasants work.
I have another friend who's a lawyer. She was complaining about working long hours. I asked if her firm was in trouble and couldn't afford to hire more lawyers. Nah, we're swimming in money. Shortage of lawyers? Nah, lots of good unemployed ones. So why? The culture says that if you don't work ridiculous hours you're not a good lawyer, so everyone does it.
You're right, some people need to work so people can eat, have shelter, etc. At one time, when humanity existed on the edge, that number was equal to (sometimes exceeded) the population. It has been decreasing for a long time, and is currently surprisingly small. It looks to decrease dramatically in the future, as more automation takes over many of the few remaining critical jobs.
The vast majority of us in western nations do not work so we (or anybody else) can eat or have shelter. We work doing various things, a surprising number of which are completely unnecessary, in order to convince our lords to give us food and money.
The Russians don't actually have any 100 MT warheads. They made one and blew it up. IIRC correctly they had a second in some state of assembly, and they had a few extra casings around. Everything was disassembled after the test. There's a casing or two on display in museums. The things weren't practical, and the Soviets knew it at the time.
I suspect the reason it's salted with cobalt is because "intercontinental torpedo with 100 MT warhead that could destroy New York" isn't as scary as "... salted with cobalt to cause lots of radioactive death!" The only thing salting a bomb would do is make sure that any allies you might have had who would stick with you through a nuclear first strike would definitely decide the world was better without you.
As you pointed out, the whole thing is ludicrous in many ways. It sounds like American or Russian propaganda (it's hard to tell the difference between the two).
They're called neural nets because their basic structure was roughly inspired by actual neural networks. The individual elements are certainly extremely simplified, but it's an interesting question how much of that extra complexity is actually necessary.
The "deep" is indeed about how many layers there are, but that is very much involved in how much power to do things they have. There's a proof that deeper architectures can be exponentially more efficient than shallow ones.
Deep learning neural networks can be statistical classifiers, but most of the ones used today aren't really statistical, and many of them aren't classifiers.
You certainly can do unsupervised learning with deep neural networks, as well as related techniques like reinforcement learning. Unsupervised learning is often paired with a bit of supervised learning, but it is in humans too (someone taught you English). Reinforcement learning lets deep learning networks learn to perform tasks based only on things like good/not good feedback, which can be provided by the environment.
AI is also construction equipment that can level a site and build a foundation autonomously (that exists now), then build a skyscraper on it (will exist soon).
AI is a class of technology that will let automation push into job categories that were previously thought safe.
What we have now is a very productive economy with an output allocation problem. Our current solution to that problem is to invent busywork that we can "pay" people to do.
Automation will certainly kill jobs, but it will also wipe away the delusion that many jobs are anything but make work. Then we'll actually have to solve the allocation problem.
"I wonder about child-rearing. I don't think automation will replace that for at least 100 years."
Looking around, I think technology has already mostly replaced child rearing. The last step is replacing the adult who's mostly there to take the responsibility if anything bad happens. And frankly, it's not hard to see a day soon when a robot is more reliable in that capacity than today's typical daycare worker.
Slashdot Mirror
If legally mandated that would outlaw a lot of devices with batteries. Nearly every phone and tablet. Most notebook computers. Cars.
Some "journalist" wanted to write a story about hacking, but, you know, different. They then located some guy who builds sensors who was willing to wax poetic about systems engineering, and voila.
I can do an iPhone in about 10 minutes. I'm sure someone who does it regularly could do it in 5. Seems easy enough.
Why is learning Python first ridiculous?
Python is one of the few current languages that you can learn as a purely procedural language, but also has a sane standard object system you can pick up later, without having to completely switch to a new language. As an extra benefit it's widely used.
The OOP extensions of Pascal were also like that and, not coincidentally, were the standard learning languages for a long time.
Hm. I enjoying the ability to use greek letters as variables in Python 3. Makes writing math-related code very nice.
However, I do have a few projects for which the ability to name variables *pile of poo* is very interesting....
When Swift was first announced it sounded like it was supposed to be Apple's version of Python. I was disappointed when I first saw it. It's basically C with some sane object oriented extensions (i.e. not C++) and some random unnecessary stuff like those let keywords.
It's true, isn't it? Most languages look like C. Except C++, which looks like C that was rescued from a corrupted hard disk.
I had a similar encounter with an old database designed by amateurs in FileMaker. I had to acquire a copy of an old version to load the file, then upgrade it to a later version that had SQL export.
And spaceships are just space stations that go somewhere.
It's an interesting question how much complexity in the elements (synapses and neurons) versus complexity in the network, is required. For the synapses, humans have fairly complicated ones, especially if you consider every type of neuron, but there are other animals that are considerably simpler. Some use only a single neurotransmitter.
We hear a lot about the simple elements/complex network approaches at the moment because they're making a lot of progress. People working from the other side haven't been as successful in terms of solving actual problems, although some of the models have shown interesting hints of self-organizing and emergent behaviour.
Personally I think that a lot of the biological complexity is imposed by biological limitations and the good enough nature of evolution, but it's quite possible there are more basic structural secrets to discover.
That was a pretty good idea in the 80s when Penrose proposed it, but it didn't really pan out. Other variations (and honestly, Penrose's work too) really sound pretty hokey, and many rely on assuming that wavefunction collapse (a) happens and (b) happens only under conscious observation as justification.
https://en.wikipedia.org/wiki/...
There are studies that show complicated Excel spreadsheets are so buggy (and hard to debug) that they're not really any better than random number generators.
You're correct: Excel has too many features, and it makes things deceptively easy.
If it scales up well, you could make a chip with much greater capacity than you get with current deep learning techniques. Simulating this kind of thing on a digital chip isn't really very efficient, in storage, circuits, or speed.
I'm curious, what money would you make on Mars that you couldn't make on the moon?
The moon is close enough and small enough that you might be able to mine it and return the result to Earth economically if you could find anything valuable there, or could potentially be a practical location for dirty industries if environmental costs on Earth got high enough.
It's certainly where you'd want to supply any kind of Earth orbital industry from, and has lots of the kinds of materials you'd need for that: oxygen, silicates, probably water. It would be the source of raw materials for building large space infrastructure: habitats, orbital rings, solar plants, interplanetary/interstellar ships, launch lasers, etc. With low gravity and vacuum it's also a good place to build really big telescopes, particle accelerators, launch loops, space elevators and other such things.
Mars seems primarily interesting for the possibility of extinct life, but I'm not sure it has any real advantages over the moon or asteroids for resources, or living space.
You're quite a good troll. Not only are you pretty entertaining, but what you say is usually technically correct, just superficial, and calculated to grate on the Slashdot groupthink. I expect you probably post quite a bit of insightful stuff under a different name.
"'Deep' learning is 'deeper' than shallow learning". Hard to argue with that, except you convinced a bunch of random Slashdotters to try.
Well done.
I'm currently sitting in a coffee shop.
There's a baby in a carriage, sleeping. The mother or nanny is on her phone. That's not terribly different than the past, except for the phone.
A couple of kids in the 2-6 range, one of which is playing on a phone, the other on a tablet. Mom/nanny is on her phone.
A minivan just pulled up outside. I can't see how many kids are in the back, but I can see that it has a ceiling mounted DVD player.
I commend you on the way you're raising your kids. If I have kids I intend to do my best to do the same. But that doesn't seem to be the reality for most people. Television, educational toys, apps to help with homework, cell phones for entertaining and keeping tabs on kids, tracking bracelets, child rearing hasn't been overlooked by tech. And all those things are still pretty crude. It's easy to see how that might go further in the future: the Japanese are madly trying to develop elder care robots to get them out of their demographic crisis, but the same technology has long been envisioned by science fiction for raising children.
I suspect eventually we will come to our senses, spend a small amount of our time working and devote much of the rest to pursuits that we enjoy. One of the most popular may be raising kids. In the meantime, I suspect that people will continue to spend less time raising their kids and make up the increasing shortfall using technology.
I'd put my bet on artist, musician, poet, or similar. The society we live in is a system adapted to the exponential growth and political conditions existing around the allocation of capital that prevailed in the twentieth century. It's going to change, and I strongly suspect that questions like "what fields should we reskill workers into?" are going to sound ridiculous a generation or two down the line.
Someone I know posted a diagram of the feudal system on Facebook today. It shows food flowing down to the peasants from the aristocracy. That's exactly what the aristocracy likes the peasants to think... never mind that the only source of food is the farms that the peasants work.
I have another friend who's a lawyer. She was complaining about working long hours. I asked if her firm was in trouble and couldn't afford to hire more lawyers. Nah, we're swimming in money. Shortage of lawyers? Nah, lots of good unemployed ones. So why? The culture says that if you don't work ridiculous hours you're not a good lawyer, so everyone does it.
You're right, some people need to work so people can eat, have shelter, etc. At one time, when humanity existed on the edge, that number was equal to (sometimes exceeded) the population. It has been decreasing for a long time, and is currently surprisingly small. It looks to decrease dramatically in the future, as more automation takes over many of the few remaining critical jobs.
The vast majority of us in western nations do not work so we (or anybody else) can eat or have shelter. We work doing various things, a surprising number of which are completely unnecessary, in order to convince our lords to give us food and money.
Forgot:
The Russians don't actually have any 100 MT warheads. They made one and blew it up. IIRC correctly they had a second in some state of assembly, and they had a few extra casings around. Everything was disassembled after the test. There's a casing or two on display in museums. The things weren't practical, and the Soviets knew it at the time.
I suspect the reason it's salted with cobalt is because "intercontinental torpedo with 100 MT warhead that could destroy New York" isn't as scary as "... salted with cobalt to cause lots of radioactive death!" The only thing salting a bomb would do is make sure that any allies you might have had who would stick with you through a nuclear first strike would definitely decide the world was better without you.
As you pointed out, the whole thing is ludicrous in many ways. It sounds like American or Russian propaganda (it's hard to tell the difference between the two).
A) you're arguing with a troll.
B) you're not really correct:
They're called neural nets because their basic structure was roughly inspired by actual neural networks. The individual elements are certainly extremely simplified, but it's an interesting question how much of that extra complexity is actually necessary.
The "deep" is indeed about how many layers there are, but that is very much involved in how much power to do things they have. There's a proof that deeper architectures can be exponentially more efficient than shallow ones.
Deep learning neural networks can be statistical classifiers, but most of the ones used today aren't really statistical, and many of them aren't classifiers.
You certainly can do unsupervised learning with deep neural networks, as well as related techniques like reinforcement learning. Unsupervised learning is often paired with a bit of supervised learning, but it is in humans too (someone taught you English). Reinforcement learning lets deep learning networks learn to perform tasks based only on things like good/not good feedback, which can be provided by the environment.
AI is also construction equipment that can level a site and build a foundation autonomously (that exists now), then build a skyscraper on it (will exist soon).
AI is a class of technology that will let automation push into job categories that were previously thought safe.
What we have now is a very productive economy with an output allocation problem. Our current solution to that problem is to invent busywork that we can "pay" people to do.
Automation will certainly kill jobs, but it will also wipe away the delusion that many jobs are anything but make work. Then we'll actually have to solve the allocation problem.
"I wonder about child-rearing. I don't think automation will replace that for at least 100 years."
Looking around, I think technology has already mostly replaced child rearing. The last step is replacing the adult who's mostly there to take the responsibility if anything bad happens. And frankly, it's not hard to see a day soon when a robot is more reliable in that capacity than today's typical daycare worker.
"we're fucked"
I suspect they're fucked. As in glorious revolution fucked.