The benefit is more social than economic for the politicians themselves usually, since they're mostly rich before they got into office. We, the people, control that social reward. We can withhold our attention from them when they are greedy, selfish, short-sighted, patronizing, hypocritical...
Jobs should not be the ultimate goal. We must challenge the idea that jobs are the only way to contribute. Let us free people from the necessity of making a living by doing what a boss tells them, and let them instead pursue their own creative interests. Give everyone the option of a basic income, and have lots of challenges by business and government to stimulate the natural curiosity and scientific spirit that most of us are born with. Knowledge and technology will advance, which is what confers survival fitness by better enabling us to predict and adapt to sudden catastrophic change.
The best option (that I can think of, at least) is to give everyone a basic income (an idea that goes back to Founding Father Thomas Paine in his 1795 Agrarian Justice), and stimulate innovation and technological progress with challenges from both biz and govt (X Prize, DARPA challenges, Google bug bounties, Netflix prize, etc.). The resulting increase in knowledge advancement will raise our survival fitness fastest because knowledge empowers us to better predict and adapt to sudden catastrophic changes.
We start by challenging the fundamental assumptions of popular economics, one of which is that government can only spend what it takes in. This assumption has been violated by the history of the United States, which has had a national debt since its very founding. Lincoln printed some $480 million greenbacks to raise money without increasing taxes or borrowing it. Japan runs a 230% debt-to-gdp ratio and has a currency they keep trying to devalue. Dick Cheney was right: Reagan proved that deficits don't matter.
"how are you going to continually assess students to make sure the class is learning, and not just following patterns or playing a game."
What do you know about learning? "Those who can't, teach." Teachers try to validate themselves by requiring students to pay attention to them, or else!
I prefer Socrates's method: teach for free, and don't give exams. If you end up in a state of aporia, that's okay. As Confucius says in The Analects, Book II Chapter XVII: "Yu, shall I teach you what knowledge is? When you know a thing, to hold that you know it; and when you do not know a thing, to allow that you do not know it;-- this is knowledge."
Instead of obsessing over whether a student is learning or not, and spending time trying to evaluate others, just concentrate on transferring knowledge; if you want to give assignments, ask the students to figure out something you don't know how to do yet. Work with the students to further knowledge, instead of acting as their adversary and withholding knowledge "with the closed fist of the teacher who keeps some things back" (Maha-parinibbana Sutta: Last Days of the Buddha, Part 2, Verse 32).
"It's widely accepted that the first hidden layer of an ANN serves as a feature detector (possibly sub-symoblic features that you can't put a name to), and each successive layer serves as a detector for higher-order features."
A couple weeks ago I was blocked from posting comments on slashdot. When I tried, i got the accusatory message: "You are not allowed to use this resource.", as if I was trying to do something I shouldn't be doing. I felt unfairly targeted.
It may have been some kind of ban on my mobile broadband ip address. The problem hasn't recurred. (I get a new IP every time I log on.)
If that's what it was, it points out how unfair ip-based blocks are.
So if Alan Turing posted about homosexuality which was illegal in his time, he should never have been hired for anything?
Anyway a better answer is a guaranteed basic income. Then people can post freely and employers can discriminate against them, and each individual can still contribute to society by working on their own projects, and/or towards challenges held by govt and biz (bug bounties, netflix prize, darpa challenges, etc.).
Formal logic includes axioms such as the law of non-contradiction that make it less expressive than natural language (again, referring to Chomsky's hierarchy of languages).
We might not think about music in words, but that does not mean that what we're thinking can't be expressed in words. We may not be able to explain in words how we know if a singer is in tune, but pitch correction machines do it in words. They use formal language of course; I think natural language can express those remaining places where the pitch machine gets it wrong. "This note is both C and G due to vibrato..."
Chess is a strange example, since machines can beat the best human at it.
I think musicians can find more expressive means of describing the concepts they refer to as color or shape. It isn't a limitation of language, but of human will.
And the feds used their unlimited power of money creation to bail out the same financial institutions that are now holding the government hostage. Time for the ppl to demand that the created money goes to us directly instead of rewarding middlemen.
"As a result of this audit, we now know that the Federal Reserve provided more than $16 trillion in total financial assistance to some of the largest financial institutions and corporations in the United States and throughout the world," said Sanders.
I'll start with tools that represent what we can express with language.
My guess is that since language is infinite, it can represent anything. Language has the potential to be complete, in the sense of Goedel's incompleteness theorem. That also means inconsistency but I don't see that as much of a problem, since you can define subsets of natural language that are consistent, if you want to.
Natural language is a superset of formal languages, according to the Chomsky hierarchy. So natural language is more expressive. Natural language is also ambiguous, inconsistent, and can change without needing any central governing authority. I think all of these traits are very important for a knowledge representation language. Ambiguity helps the language adapt to changing conditions, by allowing multiple meanings for words and sentences; so new meanings that are made necessary, for example, by technology, can be attached to existing words ("net", "keyboard", "computer"). Natural language also allows the creation of new words ("internet", "hertz", "jazz"). In a formal language, words or symbols are defined at one time and don't change; new symbols are added by individuals in a top-down manner, rather than the more democratic fashion by which natural language changes. In tolerating inconsistency, natural language allows us to express observations such as "a photon is a wave and not a wave". In trying to eliminate inconsistency, formal languages lose expressiveness and completeness.
I think knowledge representation should be flexible; which probably means it will include ambiguity, inconsistency, and the ability to change from the bottom-up (as opposed to having changes dictated from the top down). Natural language already has these characteristics, and so much of our knowledge is expressed in natural language already; why not simply use it for knowledge representation?
I think Cyc might be useful for certain limited domains. I would make it one agent in a multi-agent system, and use feedback to reinforce it when it provides the best response, according to user feedback.
I think if a formal language can represent how the brain stores information, it will have to tolerate inconsistency. Since natural language does this already, why not use it?
The cross-modal response of OR neurons could be larger than either of the modality-specific responses, and even larger than their sum. The modality-specific responses of AND neurons could be non-zero. Other neurons could not be fit into a Boolean scheme at all. For example, the responses of ENHANCED tectal neurons to a stimulus of one modality could be increased by a stimulus of another modality that was ineffective by itself. The responses of all types were significantly magnitude dependent. It would not be possible, on the basis of the data on multisensory neurons in the rattlesnake tectum, to develop a satisfying description of their response properties in terms of Boolean logic.
"If you said that the neuron was effectively voting for its preferred movement, you'd say it is voting for moving left at this time and a tenth of a second later it is voting for moving right and a tenth of a second after that it is voting for something else," Churchland said. "It would not make any sense at all."
--- Note: both the above papers present their own hypotheses about what's really going on, or how to resolve the apparent contradictions and inconsistencies observed.
In the first case, instead of P(X=Apple|S), I would use modalities in natural language ("That is an apple", "It may be an apple", "It looks like an apple" etc.). In the second case, I think I would try to use an agent model; one agent is saying "move", another is saying "don't move", and some controller makes a selection among them.
I think language is best stored in itself. I doubt it's stored in a formal language in the brain. Cyc's tried to falsify my hypothesis...
The link is the source for the idea that language is AI complete (maybe two-thirds of the way down; a search for "AI complete" or "complete" will find it).
Notice how he's fulfilled his own prophecy, by proving that he can't represent his knowledge in words...but I think this line of argument has much more to do with social, rather than artificial, intelligence. At any rate, I'll continue trying to falsify his hypothesis by trying to represent natural language in language.
Natural language has the flexibility to represent the inconsistencies and paradoxes that abound all around us. Formal systems have problems because they try to ban inconsistency; but we have to reason in the presence of inconsistency. Natural language gives us a natural tool to represent inconsistent knowledge, and to reason with it. Formal languages are a subset of natural language; natural language is more expressive than formal languages; natural language can be as explicit as we want it to be, while also retaining its flexibility in other contexts. In allowing multiple meanings for words, for example, natural language can adapt to changing conditions (so that "web", "site", "hit", "mouse", etc. all took on new meanings in the last few decades).
Natural language is AI-complete (http://see.stanford.edu/materials/ainlpcs224n/transcripts/NaturalLanguageProcessing-Lecture01.html). When we deal with it from the ground up, representing our knowledge in it instead of creating impedance mismatches by trying to convert it to a formal language representation, we'll have to figure out the algorithms that can be applied to solve (all?) the other problems of AI.
Instead of community service, hold challenges (which biz can hold to, like Google bug bounties, Netflix challenge, etc.) to stimulate individuals to exploit the natural instinct for creativity and wonder that each of us is born with. As long as we keep innovating, we can create as much money as we want without consequence (as Japan, with its 200% debt-to-gdp ratio and currency they keep trying to devalue, proves).
Children who haven't learned to speak need constant supervision.
Natural language gives us the ability to manipulate our thoughts and do experiments like move subject to object: "The sun moves around the earth", what if "The earth moves around the sun"?
Most of our knowledge is stored in natural language now. When we want to know something, we look it up, or ask someone...
Slashdot Mirror
What if roombas pick up after him?
The benefit is more social than economic for the politicians themselves usually, since they're mostly rich before they got into office. We, the people, control that social reward. We can withhold our attention from them when they are greedy, selfish, short-sighted, patronizing, hypocritical...
Jobs should not be the ultimate goal. We must challenge the idea that jobs are the only way to contribute. Let us free people from the necessity of making a living by doing what a boss tells them, and let them instead pursue their own creative interests. Give everyone the option of a basic income, and have lots of challenges by business and government to stimulate the natural curiosity and scientific spirit that most of us are born with. Knowledge and technology will advance, which is what confers survival fitness by better enabling us to predict and adapt to sudden catastrophic change.
Economics is a tool that we invented to serve us. It is not some God that we must practice human sacrifice to.
If automation allows us to live a life where we are more free to do what we want, that's a good thing. We're closer to utopia.
We should challenge the economics that says we can't create money and give it to people. In fact we created $16 trillion (enough to pay off the entire national debt) in two years to bail out financial unions (source: http://www.sanders.senate.gov/newsroom/news/?id=9e2a4ea8-6e73-4be2-a753-62060dcbb3c3 ).
The best option (that I can think of, at least) is to give everyone a basic income (an idea that goes back to Founding Father Thomas Paine in his 1795 Agrarian Justice), and stimulate innovation and technological progress with challenges from both biz and govt (X Prize, DARPA challenges, Google bug bounties, Netflix prize, etc.). The resulting increase in knowledge advancement will raise our survival fitness fastest because knowledge empowers us to better predict and adapt to sudden catastrophic changes.
We start by challenging the fundamental assumptions of popular economics, one of which is that government can only spend what it takes in. This assumption has been violated by the history of the United States, which has had a national debt since its very founding. Lincoln printed some $480 million greenbacks to raise money without increasing taxes or borrowing it. Japan runs a 230% debt-to-gdp ratio and has a currency they keep trying to devalue. Dick Cheney was right: Reagan proved that deficits don't matter.
"how are you going to continually assess students to make sure the class is learning, and not just following patterns or playing a game."
What do you know about learning? "Those who can't, teach." Teachers try to validate themselves by requiring students to pay attention to them, or else!
I prefer Socrates's method: teach for free, and don't give exams. If you end up in a state of aporia, that's okay. As Confucius says in The Analects, Book II Chapter XVII: "Yu, shall I teach you what knowledge is? When you know a thing, to hold that you know it; and when you do not know a thing, to allow that you do not know it;-- this is knowledge."
Instead of obsessing over whether a student is learning or not, and spending time trying to evaluate others, just concentrate on transferring knowledge; if you want to give assignments, ask the students to figure out something you don't know how to do yet. Work with the students to further knowledge, instead of acting as their adversary and withholding knowledge "with the closed fist of the teacher who keeps some things back" (Maha-parinibbana Sutta: Last Days of the Buddha, Part 2, Verse 32).
"It's widely accepted that the first hidden layer of an ANN serves as a feature detector (possibly sub-symoblic features that you can't put a name to), and each successive layer serves as a detector for higher-order features."
Why can't you put a name to them?
yet?
How much do you think a relatively slow pace of innovation, the lack of a democratic government, and near-continuous wars contributed to inflation?
The fact that suicides keep happening is evidence that memes > genes.
You don't speak for me.
A couple weeks ago I was blocked from posting comments on slashdot. When I tried, i got the accusatory message: "You are not allowed to use this resource.", as if I was trying to do something I shouldn't be doing. I felt unfairly targeted.
It may have been some kind of ban on my mobile broadband ip address. The problem hasn't recurred. (I get a new IP every time I log on.)
If that's what it was, it points out how unfair ip-based blocks are.
So if Alan Turing posted about homosexuality which was illegal in his time, he should never have been hired for anything?
Anyway a better answer is a guaranteed basic income. Then people can post freely and employers can discriminate against them, and each individual can still contribute to society by working on their own projects, and/or towards challenges held by govt and biz (bug bounties, netflix prize, darpa challenges, etc.).
Formal logic includes axioms such as the law of non-contradiction that make it less expressive than natural language (again, referring to Chomsky's hierarchy of languages).
We might not think about music in words, but that does not mean that what we're thinking can't be expressed in words. We may not be able to explain in words how we know if a singer is in tune, but pitch correction machines do it in words. They use formal language of course; I think natural language can express those remaining places where the pitch machine gets it wrong. "This note is both C and G due to vibrato..."
Chess is a strange example, since machines can beat the best human at it.
I think musicians can find more expressive means of describing the concepts they refer to as color or shape. It isn't a limitation of language, but of human will.
And the feds used their unlimited power of money creation to bail out the same financial institutions that are now holding the government hostage. Time for the ppl to demand that the created money goes to us directly instead of rewarding middlemen.
http://www.sanders.senate.gov/newsroom/news/?id=9e2a4ea8-6e73-4be2-a753-62060dcbb3c3
I'll start with tools that represent what we can express with language.
My guess is that since language is infinite, it can represent anything. Language has the potential to be complete, in the sense of Goedel's incompleteness theorem. That also means inconsistency but I don't see that as much of a problem, since you can define subsets of natural language that are consistent, if you want to.
Natural language is a superset of formal languages, according to the Chomsky hierarchy. So natural language is more expressive. Natural language is also ambiguous, inconsistent, and can change without needing any central governing authority. I think all of these traits are very important for a knowledge representation language. Ambiguity helps the language adapt to changing conditions, by allowing multiple meanings for words and sentences; so new meanings that are made necessary, for example, by technology, can be attached to existing words ("net", "keyboard", "computer"). Natural language also allows the creation of new words ("internet", "hertz", "jazz"). In a formal language, words or symbols are defined at one time and don't change; new symbols are added by individuals in a top-down manner, rather than the more democratic fashion by which natural language changes. In tolerating inconsistency, natural language allows us to express observations such as "a photon is a wave and not a wave". In trying to eliminate inconsistency, formal languages lose expressiveness and completeness.
I think knowledge representation should be flexible; which probably means it will include ambiguity, inconsistency, and the ability to change from the bottom-up (as opposed to having changes dictated from the top down). Natural language already has these characteristics, and so much of our knowledge is expressed in natural language already; why not simply use it for knowledge representation?
I think Cyc might be useful for certain limited domains. I would make it one agent in a multi-agent system, and use feedback to reinforce it when it provides the best response, according to user feedback.
I think if a formal language can represent how the brain stores information, it will have to tolerate inconsistency. Since natural language does this already, why not use it?
http://www.lel.ed.ac.uk/~jim/BBSNEURO/anastasio.html:
http://news.stanford.edu/news/2010/november/neutrons-brain-movement-110810.html:
---
Note: both the above papers present their own hypotheses about what's really going on, or how to resolve the apparent contradictions and inconsistencies observed.
In the first case, instead of P(X=Apple|S), I would use modalities in natural language ("That is an apple", "It may be an apple", "It looks like an apple" etc.). In the second case, I think I would try to use an agent model; one agent is saying "move", another is saying "don't move", and some controller makes a selection among them.
I think language is best stored in itself. I doubt it's stored in a formal language in the brain. Cyc's tried to falsify my hypothesis...
The link is the source for the idea that language is AI complete (maybe two-thirds of the way down; a search for "AI complete" or "complete" will find it).
Notice how he's fulfilled his own prophecy, by proving that he can't represent his knowledge in words...but I think this line of argument has much more to do with social, rather than artificial, intelligence. At any rate, I'll continue trying to falsify his hypothesis by trying to represent natural language in language.
Natural language has the flexibility to represent the inconsistencies and paradoxes that abound all around us. Formal systems have problems because they try to ban inconsistency; but we have to reason in the presence of inconsistency. Natural language gives us a natural tool to represent inconsistent knowledge, and to reason with it. Formal languages are a subset of natural language; natural language is more expressive than formal languages; natural language can be as explicit as we want it to be, while also retaining its flexibility in other contexts. In allowing multiple meanings for words, for example, natural language can adapt to changing conditions (so that "web", "site", "hit", "mouse", etc. all took on new meanings in the last few decades).
Natural language is AI-complete (http://see.stanford.edu/materials/ainlpcs224n/transcripts/NaturalLanguageProcessing-Lecture01.html). When we deal with it from the ground up, representing our knowledge in it instead of creating impedance mismatches by trying to convert it to a formal language representation, we'll have to figure out the algorithms that can be applied to solve (all?) the other problems of AI.
Note how this reply is trying to distract from my words by focusing attention on me (and the poster's feeling sorry for me)?
Newton had no children. Was he an evolutionary failure? Genghis Khan has millions of descendants, but they study Newton's memes in school...
Instead of community service, hold challenges (which biz can hold to, like Google bug bounties, Netflix challenge, etc.) to stimulate individuals to exploit the natural instinct for creativity and wonder that each of us is born with. As long as we keep innovating, we can create as much money as we want without consequence (as Japan, with its 200% debt-to-gdp ratio and currency they keep trying to devalue, proves).
Children who haven't learned to speak need constant supervision.
Natural language gives us the ability to manipulate our thoughts and do experiments like move subject to object: "The sun moves around the earth", what if "The earth moves around the sun"?
Most of our knowledge is stored in natural language now. When we want to know something, we look it up, or ask someone...
Knowledge is the key. Memes > genes.