It's not Team Capitalism against the rest of the world. Attacking critics of a concept with fallacies such as ad hominems is simply irrational.
Capitalism has its merits and its flaws. In a lot of markets it simply doesn't produce what we want as a society, but in others it absolutely does. Arguments encouraging thinking critically about capitalism as a way to organize a society are much more fruitful than defending capitalism to the death, with all the rhetorical devices you can think of.
I assume you mean artificial neuron, because we don't really know what a neuron does in this context.
We really do have a rough idea, though. Also, I explicitly included 'to me', so the context was my definition of pattern matching with regard to neurons. Read it again.
It's probably useful for a range of applications, but I'm not sure it does much for general AI.
My point was that they clearly capture essential aspects ('patterns') of objective reality, such as gender, age (of an image of a face in the case of the Nvidia video), etc. Capturing essential patterns of objective reality in turn is what is required for general AI.
Older ANNs are much more like black boxes in the way they captured patterns in the data. Being able to instruct the network to create a more female-like image with a simple slider is a huge step towards understanding and mastering the insides of ANNs.
We don't really understand any of this so it's hard to make any solid conclusions.
That is a gross overstatement. We really do understand some of it.
I assume most people would agree adding convolution was an important step for image classification with deep learning. We are probably going to have to add a lot more structure.
CNNs have existed for quite a while before the 'deep learning revolution'. It does show how much potency can be added with implementing more complex topology with in essence identical components. Of course we're going to need more complex topology (or 'structure'). That is so obvious that attacking the opposite there would be attacking a straw man. See next point.
I suppose you could claim that anything that has a neural structure is some type of pattern matching, but I would say that's a pretty non-standard definition.
Yes, that is exactly what I did when I started my post. Lately, Slashdot AI articles have been steeped in people implying that there is something special about organic brains, something that ANNs don't have yet. It is trivially true that ANNs aren't at the level of intelligence of humans yet and it is also trivially true that their topologies are not even close to being as complex or as large as those of human brains (or even insects). To then say that the lack of the existence of AGI or anything close to it, or that current systems 'only do pattern matching' is proof that they are missing something fundamental is fallacious (as what else could it refer to than the components it is made up of?). It reeks of metaphysicism.
In defense of my definition: yes, neurons do pattern matching. It fires for a certain subset, i.e. a pattern of its input space. Take the original claim I was responding to and apply it to the simplest insect brain you can think of. Does the system as a whole really do much more than pattern matching?
I guess it depends on how you define pattern matching.
That is a good point. To me, what a single neuron or collection of neurons does is pattern matching, in the sense that the output of a neuron or collection thereof can be regarded as identifying a certain (abstracted, meaningful) pattern within the universe.
Generative Adversarial Networks (GANs) to me are definitely showing signs of that. Just look at Nvidia's latest stuff: https://www.youtube.com/watch?...
A more sophisticated intelligent system is probably going to need more structure for both the problem and the solution.
Will it, though?
The difference between us and apes in intelligence is huge, but in terms of biological evolution and makeup we're not that far apart. Finding the differences between the brain of a bonobo and a human is pretty hard. If there is a 'special sauce' to intelligence, it must be a fairly small (yet very significant) variation on the theme of the chimp biological neural network.
I'm not sure where I'd put current artificial neural networks exactly on an analogous evolutionary scale, but I'd say we are past invertebrate and into insect levels of potency (although perhaps not complexity just yet). Consider advances in science like this: - https://www.sciencedaily.com/r... - https://www.nationalgeographic... Moving ANNs up the evolutionary scale is going to be far from trivial, but I'd say there is a pretty clear path forward.
Citing an incompetent inference engine as proof that organic brains do more than pattern matching is not really evidence at all. In essence, you have to prove that the core elements of organic brains perform more than pattern matching.
As an analogy: that initial computers weren't capable of predicting tomorrow's weather to any reasonable degree of accuracy doesn't mean that there is a requirement for any 'special sauce' beyond 'just symbol manipulation' to do so. Of course there are all kinds of complexities to predicting tomorrow's weather and in that sense the 'symbol manipulation' part of it is as much of a red herring as 'pattern matching'. Current artificial neural networks are woefully simplistic in both size and topology compared to the human brain and to imply that the pattern matching nature (of its components) will forever be inadequate to produce intelligence is at the very least premature.
My experience is that many people just really, really want a 'special sauce' to exist, simply to protect their ego.
They only recently reneged on their keyboard Moto Mod and instead announced that they were switching to developing a complete device: https://www.indiegogo.com/proj... It could still turn out to be a disappointment, but I am nevertheless cautiously optimistic. At least someone seems to be seriously developing a (landscape) slider, because Blackberry sure as fuck isn't anymore (I bought and use the Priv).
Wrong. I did not make this claim: "Whatever it is, it's more than just matching patterns." The burden of proof is on the one making the claim.
Furthermore, you are implying that if no existing pattern matching system (rudimentary as they currently are) has 'remotely emulated human capabilities' (which these rudimentary systems actually have), then no pattern matching system will ever be able to emulate human capabilities. This is an obvious logical fallacy.
I agree there, but I'd like to add that it does so in the same way and at the same level as pretty much all other ad hominems and a slew of other fallacies. The internet is in great need of a commenting/microposting-system that effectively cultivates well-reasoned debate. To this day, Slashdot is still the best I've encountered.
toxic because it feeds the mob mentality that leads to doxing and threats of violence that cannot simply be ignored.
I'm not sure I can get behind (acting on) this. Has there ever been a case of a Slashdotter being doxed? There certainly are plenty of harsh (to say the least) ad hominems going around here.
I think the problem with doxing and threats of violence is that the targets serve as the personification of the arch enemy of whichever tribe the perpetrators have joined. I'm sure that seeing others pile hatred on the target du jour has an unwanted impact on them, but the general level of opposition against 'the other team' seems much more influential to me. As an example, I'd say that "The Clintons are running a child sex ring in restaurant Comet Ping Pong" is much more toxic (in the sense you described) than "Hillary is a fat cunt". Obviously "Hillary is evil and must die!" would be much more toxic than both of the above.
What I'm getting at is that comments should be judged on their insidiousness and dishonesty, not on the number of 'offensive words' in them.
In a way, it is. - The financial crisis in 2008 was definitely made possible by deregulation of Wallstreet and the push for laissez faire in general. The ripple effect it had is still felt in countries where nationalism is on the rise. Specifically, it rallied people around the idea of national austerity in economically hard times which led to more income inequality (with most of those responsible going unpunished or even largely unimpacted). So there is also plenty of blame to go around for those (rightwing) governments that have facilitated said income inequality.
- War-mongering in the Middle East has been a mostly Republican affair and it has certainly cultivated religious hatred worldwide (back and forth) as well as created many, many refugees. Don't get me wrong, I'm definitely not implying that the Middle East would have been some peaceful oasis without Western 'interventions', but in hindsight I believe we can be fairly sure that we haven't really helped.
The problem is our instincts are not the best course of action for the problem. It is like an opossum playing dead on a busy street. These are problems that require work, planning, sacrifice and thought to solve. Not a gut feeling.
To me that describes the rightwing approach to governing pretty well. Driven by instincts and gut feelings to a large degree and representing the world in an overly simplified fashion.
Well, to be fair, the addition of 'women' does imply that gender is somehow significant in that. Titles such as 'Every 30 seconds a Jew is abused on Twitter' would definitely lead to thoughts and questions about antisemitism.
Looking at some of their examples, the gender part is really quite insignificant: - "hope you need the emergency services and they fucking refuse to help you. Sour faced twat." - "B list politician. Grade A cunt." - "Another thin skinned whiny little bitch." ( 3/8 of the examples they give in the slider viewer here: https://decoders.amnesty.org/p... ) Especially in Britain these are very generic and genderless insults.
I'd like to add that as far as internet comments go, these are also pretty mild. I'd wager that there isn't a Slashdotter around that would actually be offended by these and wouldn't just brush them off.
It's more likely just some shitty oblivious newfag who only recently discovered the concept of online multiplayer gaming and didn't bother to look beyond yesterday to see that pretty much any popular online multiplayer game can be described in exactly the same way. I mean World of fucking Warcraft, anyone?
What a stupid question. 1. What evidence do you have that it isn't applied for practical uses? Because you are unaware of something, doesn't mean it does not exist. 2. Why wouldn't you apply it to pretty much everything you can think of? Machine learning is a very, very potent technology that everybody and their dog wants to get a piece of. Governments will apply it to governing, companies to making money and hobbyists will apply it as a pastime. Problem?
To be fair, we have PGP and other forms of encrypted communication. And setting up your own SMTP and POP server is still doable. It's not that it can't be done, just that it isn't popular to do. Installing Whatsapp is easier, especially for the rest of your friends.
This leads me to something I always think when reading shit like this "to stop terrorists and criminals who rely on secure messaging apps to communicate". I'd assume that proper terrorist cells are technically capable enough not to rely on Whatsapp or something similar. In fact, a security professional has indicated to me that because it's fairly easy to find open source encrypted communication code, these guys just wrote their own apps and distributed them outside of regular channels. I'd guess that even significant drug dealers would be incentivized enough to secure their communications that they'd be fairly able to avoid the compromised communication channels.
Don't get me wrong, fuck Amazon, but a story about how a huge company slowly degrades the quality of some product/service of theirs? I'm shocked. Shocked, I tell you.
Then a second person would come along with an old tire, and figuring that two tires really aren't any worse than one, he'd dump his there too. The process would snowball at an extraordinary rate, and soon you'd have hundreds, thousands, or even millions of tires.
So the last million people came along with their old tires and climbed up the mountain of tires to put theirs somewhere at the top? Or did these guys all bring their tire flinging catapults?
While I'll likely get flamed for this opinion but I don't see much writing (online, at least) that indicates an "over-reliance on teaching grammar". IMHO, it's not being emphasized enough.
I agree with the former, but I believe that it is mostly due to people being unwilling to revisit grammar (and spelling) at an adult age. My approach in any (i.e. primary or secondary) language is to lookup words or grammar rules whenever I am unsure about them. A simple and effective strategy. We all forget things or have simply never learnt them well enough.
Other people seem unwilling to do this, perhaps out of shame of engaging in an activity that seems infantile (as learning grammar and spelling is something associated with primary school). Perhaps this is also why people feel offended when you correct their language (which I will always welcome) with something along the lines of "you knew what I meant!" or "yeah, whatever, grammar nazi". A coping mechanism for something that feels as if they weren't able to calculate 7*8 and if having to look that up would be admitting utter failure.
I didn't say 'waste'. I said 'shit'. I meant actual shit and vats of cells or 3D-printed steaks do not produce that.
In fact, actual shit contains colonies of bacteria and their waste products. Those are very often responsible for the contamination of food and thus to food borne illnesses. Enjoy your botulism!
The fact of the matter is that cultured meat will always be produced in a much more controlled environment than naturally grown meat. Growing animals, keeping them alive and reasonably healthy, dealing with their shit, slicing them up, and packaging the results is much more involved than 3D-printing some cells or growing them in bioreactors.
Also, people regularly get food poisoning to the point of death ( https://www.foodsafetynews.com... ), so the current situation is clearly far from ideal, even with the huge amounts of antibiotics used in natural meat production.
Finally, culturing meat is pretty high-tech stuff and especially in the beginning it will be an expensive product aimed at wealthy well-informed early adopters, and will be subject to much higher levels of regulation than the existing meat products. If anything, the level of cutting corners will increase as the market matures and the race to produce the cheapest cultured meat takes off.
Slashdot Mirror
Stop it with the tribalism already.
It's not Team Capitalism against the rest of the world. Attacking critics of a concept with fallacies such as ad hominems is simply irrational.
Capitalism has its merits and its flaws. In a lot of markets it simply doesn't produce what we want as a society, but in others it absolutely does. Arguments encouraging thinking critically about capitalism as a way to organize a society are much more fruitful than defending capitalism to the death, with all the rhetorical devices you can think of.
I assume you mean artificial neuron, because we don't really know what a neuron does in this context.
We really do have a rough idea, though. Also, I explicitly included 'to me', so the context was my definition of pattern matching with regard to neurons. Read it again.
It's probably useful for a range of applications, but I'm not sure it does much for general AI.
My point was that they clearly capture essential aspects ('patterns') of objective reality, such as gender, age (of an image of a face in the case of the Nvidia video), etc. Capturing essential patterns of objective reality in turn is what is required for general AI.
Older ANNs are much more like black boxes in the way they captured patterns in the data. Being able to instruct the network to create a more female-like image with a simple slider is a huge step towards understanding and mastering the insides of ANNs.
We don't really understand any of this so it's hard to make any solid conclusions.
That is a gross overstatement. We really do understand some of it.
I assume most people would agree adding convolution was an important step for image classification with deep learning. We are probably going to have to add a lot more structure.
CNNs have existed for quite a while before the 'deep learning revolution'. It does show how much potency can be added with implementing more complex topology with in essence identical components. Of course we're going to need more complex topology (or 'structure'). That is so obvious that attacking the opposite there would be attacking a straw man. See next point.
I suppose you could claim that anything that has a neural structure is some type of pattern matching, but I would say that's a pretty non-standard definition.
Yes, that is exactly what I did when I started my post. Lately, Slashdot AI articles have been steeped in people implying that there is something special about organic brains, something that ANNs don't have yet. It is trivially true that ANNs aren't at the level of intelligence of humans yet and it is also trivially true that their topologies are not even close to being as complex or as large as those of human brains (or even insects). To then say that the lack of the existence of AGI or anything close to it, or that current systems 'only do pattern matching' is proof that they are missing something fundamental is fallacious (as what else could it refer to than the components it is made up of?). It reeks of metaphysicism.
In defense of my definition: yes, neurons do pattern matching. It fires for a certain subset, i.e. a pattern of its input space. Take the original claim I was responding to and apply it to the simplest insect brain you can think of. Does the system as a whole really do much more than pattern matching?
I guess it depends on how you define pattern matching.
That is a good point. To me, what a single neuron or collection of neurons does is pattern matching, in the sense that the output of a neuron or collection thereof can be regarded as identifying a certain (abstracted, meaningful) pattern within the universe.
Generative Adversarial Networks (GANs) to me are definitely showing signs of that. Just look at Nvidia's latest stuff:
https://www.youtube.com/watch?...
A more sophisticated intelligent system is probably going to need more structure for both the problem and the solution.
Will it, though?
The difference between us and apes in intelligence is huge, but in terms of biological evolution and makeup we're not that far apart. Finding the differences between the brain of a bonobo and a human is pretty hard. If there is a 'special sauce' to intelligence, it must be a fairly small (yet very significant) variation on the theme of the chimp biological neural network.
I'm not sure where I'd put current artificial neural networks exactly on an analogous evolutionary scale, but I'd say we are past invertebrate and into insect levels of potency (although perhaps not complexity just yet). Consider advances in science like this:
- https://www.sciencedaily.com/r...
- https://www.nationalgeographic...
Moving ANNs up the evolutionary scale is going to be far from trivial, but I'd say there is a pretty clear path forward.
Citing an incompetent inference engine as proof that organic brains do more than pattern matching is not really evidence at all. In essence, you have to prove that the core elements of organic brains perform more than pattern matching.
As an analogy: that initial computers weren't capable of predicting tomorrow's weather to any reasonable degree of accuracy doesn't mean that there is a requirement for any 'special sauce' beyond 'just symbol manipulation' to do so. Of course there are all kinds of complexities to predicting tomorrow's weather and in that sense the 'symbol manipulation' part of it is as much of a red herring as 'pattern matching'. Current artificial neural networks are woefully simplistic in both size and topology compared to the human brain and to imply that the pattern matching nature (of its components) will forever be inadequate to produce intelligence is at the very least premature.
My experience is that many people just really, really want a 'special sauce' to exist, simply to protect their ego.
I got tired of your dodging and ad hominems and decided to return the favor.
Horseshit.
Unless you define 'good' as 'unusably thin'.
Good news, though! The market may actually finally provide what we desire:
https://www.readyfx.com/
https://twitter.com/livermoriu...
They only recently reneged on their keyboard Moto Mod and instead announced that they were switching to developing a complete device: https://www.indiegogo.com/proj...
It could still turn out to be a disappointment, but I am nevertheless cautiously optimistic. At least someone seems to be seriously developing a (landscape) slider, because Blackberry sure as fuck isn't anymore (I bought and use the Priv).
Answer the question, fuckface.
"How do you know [that organic brains do more than just matching patterns]?"
Wrong. I did not make this claim: "Whatever it is, it's more than just matching patterns."
The burden of proof is on the one making the claim.
Furthermore, you are implying that if no existing pattern matching system (rudimentary as they currently are) has 'remotely emulated human capabilities' (which these rudimentary systems actually have), then no pattern matching system will ever be able to emulate human capabilities. This is an obvious logical fallacy.
Whatever it is, it's more than just matching patterns.
How do you know that?
because it stifles debate
I agree there, but I'd like to add that it does so in the same way and at the same level as pretty much all other ad hominems and a slew of other fallacies.
The internet is in great need of a commenting/microposting-system that effectively cultivates well-reasoned debate. To this day, Slashdot is still the best I've encountered.
toxic because it feeds the mob mentality that leads to doxing and threats of violence that cannot simply be ignored.
I'm not sure I can get behind (acting on) this. Has there ever been a case of a Slashdotter being doxed? There certainly are plenty of harsh (to say the least) ad hominems going around here.
I think the problem with doxing and threats of violence is that the targets serve as the personification of the arch enemy of whichever tribe the perpetrators have joined. I'm sure that seeing others pile hatred on the target du jour has an unwanted impact on them, but the general level of opposition against 'the other team' seems much more influential to me. As an example, I'd say that "The Clintons are running a child sex ring in restaurant Comet Ping Pong" is much more toxic (in the sense you described) than "Hillary is a fat cunt". Obviously "Hillary is evil and must die!" would be much more toxic than both of the above.
What I'm getting at is that comments should be judged on their insidiousness and dishonesty, not on the number of 'offensive words' in them.
Well it isn't the Republicans Fault
In a way, it is.
- The financial crisis in 2008 was definitely made possible by deregulation of Wallstreet and the push for laissez faire in general. The ripple effect it had is still felt in countries where nationalism is on the rise. Specifically, it rallied people around the idea of national austerity in economically hard times which led to more income inequality (with most of those responsible going unpunished or even largely unimpacted). So there is also plenty of blame to go around for those (rightwing) governments that have facilitated said income inequality.
- War-mongering in the Middle East has been a mostly Republican affair and it has certainly cultivated religious hatred worldwide (back and forth) as well as created many, many refugees. Don't get me wrong, I'm definitely not implying that the Middle East would have been some peaceful oasis without Western 'interventions', but in hindsight I believe we can be fairly sure that we haven't really helped.
The problem is our instincts are not the best course of action for the problem. It is like an opossum playing dead on a busy street. These are problems that require work, planning, sacrifice and thought to solve. Not a gut feeling.
To me that describes the rightwing approach to governing pretty well. Driven by instincts and gut feelings to a large degree and representing the world in an overly simplified fashion.
Well, to be fair, the addition of 'women' does imply that gender is somehow significant in that. Titles such as 'Every 30 seconds a Jew is abused on Twitter' would definitely lead to thoughts and questions about antisemitism.
Looking at some of their examples, the gender part is really quite insignificant:
- "hope you need the emergency services and they fucking refuse to help you. Sour faced twat."
- "B list politician. Grade A cunt."
- "Another thin skinned whiny little bitch."
( 3/8 of the examples they give in the slider viewer here: https://decoders.amnesty.org/p... )
Especially in Britain these are very generic and genderless insults.
I'd like to add that as far as internet comments go, these are also pretty mild. I'd wager that there isn't a Slashdotter around that would actually be offended by these and wouldn't just brush them off.
It's more likely just some shitty oblivious newfag who only recently discovered the concept of online multiplayer gaming and didn't bother to look beyond yesterday to see that pretty much any popular online multiplayer game can be described in exactly the same way.
I mean World of fucking Warcraft, anyone?
What a stupid question.
1. What evidence do you have that it isn't applied for practical uses? Because you are unaware of something, doesn't mean it does not exist.
2. Why wouldn't you apply it to pretty much everything you can think of? Machine learning is a very, very potent technology that everybody and their dog wants to get a piece of. Governments will apply it to governing, companies to making money and hobbyists will apply it as a pastime.
Problem?
To be fair, we have PGP and other forms of encrypted communication. And setting up your own SMTP and POP server is still doable. It's not that it can't be done, just that it isn't popular to do. Installing Whatsapp is easier, especially for the rest of your friends.
This leads me to something I always think when reading shit like this "to stop terrorists and criminals who rely on secure messaging apps to communicate". I'd assume that proper terrorist cells are technically capable enough not to rely on Whatsapp or something similar. In fact, a security professional has indicated to me that because it's fairly easy to find open source encrypted communication code, these guys just wrote their own apps and distributed them outside of regular channels.
I'd guess that even significant drug dealers would be incentivized enough to secure their communications that they'd be fairly able to avoid the compromised communication channels.
First world problem if ever I've seen one.
Don't get me wrong, fuck Amazon, but a story about how a huge company slowly degrades the quality of some product/service of theirs? I'm shocked. Shocked, I tell you.
The largest were called "tire mountains"
(my emphasis)
Then a second person would come along with an old tire, and figuring that two tires really aren't any worse than one, he'd dump his there too. The process would snowball at an extraordinary rate, and soon you'd have hundreds, thousands, or even millions of tires.
So the last million people came along with their old tires and climbed up the mountain of tires to put theirs somewhere at the top? Or did these guys all bring their tire flinging catapults?
Technically, in journalism, it is called the 'lead' (or 'lede') and it should be the first paragraph:
https://en.wikipedia.org/wiki/...
https://en.wikipedia.org/wiki/...
Related and relevant:
https://en.wikipedia.org/wiki/...
Also good form when writing a lengthy email or comment.
So if the final paragraph summaries the preceding stuff, that is all a reader needs to bother with.
Spoken like a true Slashdotter. TFS is all you need, eh?
While I'll likely get flamed for this opinion but I don't see much writing (online, at least) that indicates an "over-reliance on teaching grammar". IMHO, it's not being emphasized enough.
I agree with the former, but I believe that it is mostly due to people being unwilling to revisit grammar (and spelling) at an adult age. My approach in any (i.e. primary or secondary) language is to lookup words or grammar rules whenever I am unsure about them. A simple and effective strategy. We all forget things or have simply never learnt them well enough.
Other people seem unwilling to do this, perhaps out of shame of engaging in an activity that seems infantile (as learning grammar and spelling is something associated with primary school). Perhaps this is also why people feel offended when you correct their language (which I will always welcome) with something along the lines of "you knew what I meant!" or "yeah, whatever, grammar nazi". A coping mechanism for something that feels as if they weren't able to calculate 7*8 and if having to look that up would be admitting utter failure.
Nice straw men. Good day.
Fetal bovine serum is on its way out:
https://www.wired.co.uk/articl...
If you think you can keep it out you are living in an imaginary world. âBetter living through chemicals!â
That's the spirit! Give up before you've even started! Better living through inaction and Luddism!
I didn't say 'waste'. I said 'shit'. I meant actual shit and vats of cells or 3D-printed steaks do not produce that.
In fact, actual shit contains colonies of bacteria and their waste products. Those are very often responsible for the contamination of food and thus to food borne illnesses. Enjoy your botulism!
The fact of the matter is that cultured meat will always be produced in a much more controlled environment than naturally grown meat. Growing animals, keeping them alive and reasonably healthy, dealing with their shit, slicing them up, and packaging the results is much more involved than 3D-printing some cells or growing them in bioreactors.
Also, people regularly get food poisoning to the point of death ( https://www.foodsafetynews.com... ), so the current situation is clearly far from ideal, even with the huge amounts of antibiotics used in natural meat production.
Finally, culturing meat is pretty high-tech stuff and especially in the beginning it will be an expensive product aimed at wealthy well-informed early adopters, and will be subject to much higher levels of regulation than the existing meat products. If anything, the level of cutting corners will increase as the market matures and the race to produce the cheapest cultured meat takes off.