Artificial intelligence, great and terrible. Part three
Start Previous
Recently, much has been said about the National Assembly. I would even say, indecently much. I have never considered the National Assembly to even hint at AI and, judging by the numerous comments, this opinion is shared by quite a few people. Some sayings:
I have nothing to add to this list. If you teach the National Assembly on the operating instructions of tower cranes, and then slip the Kama Sutra, then its entire AI will immediately appear in all its glory. What is attractive neural networks? I think by the fact that they do not require the work of the “protein” intellect: I took a ready-made network, dragged it in on some ready-made sample, got some ready-made result. Brainstorming is not required at all: "If the gravitapus flies, then don't care how she does it."
I treat the databases approximately as subjectively as neural networks. Only with the opposite sign. In my youth, I even defined AI as “self-learning IPS is intelligence”. Now I think that "the DBMS should be the most active user of its databases." In my view, both of these phrases are almost equivalent. And by “database” I mean precisely the graph database, but not the DBD. More precisely, I believe that this is the same thing. In particular, all these “ontologies” are the subgraphs of metadata.
')
The main problem of textual AI is information poverty. Of course, you can feed him all the textual heritage of mankind, but this is just a static text, they can be taught the program to make sentences, but not understand their meaning.
And here I am sorry! Personally, I am only interested in text data (more precisely, which can be represented as text). And this “informational poverty” will be enough for me to the end of my life!
More than half (according to my estimates, about two thirds) of all conversations on the topic of AI relate to horror movies a la “Terminator”. A person plays with fire (and when he didn’t play with him?), Super-AI will secretly collect his body, and then he will take revenge on people (according to the description, he looks more like a super-moron), a malicious AI can attack another AI , to make him act in a certain way (I haven't heard such a horror film yet - and what if they make friends, will it be better?).
Etc. etc. Recently, hysteria has been fueled by the statements of various famous individuals: Stephen Hawking and Ilon Musk believe that AI can be the end of our civilization. Bill Gates thinks the same way. Steve Wozniak talks about the threat of AI. The head of Google believes that the fear of AI is completely justified. Noam Chomsky asks where the AI ​​went wrong. India is cautiously looking towards artificial intelligence. Tim Berners-Lee suggests that AI will start managing financial corporations ...
Why make loud statements for the masses? They are so afraid of the stupidity of everything, the collider, the GMO. What other little worries? The planet is overcrowded, clean air and drinking water are becoming scarce, animals and insects are dying out, the ocean is covered with plastic, the climate is changing ... and we have nothing else to do except to shake from fear of the revolt of the mad excavators? And who are all these Masks and Hawkings? Is at least one of them a specialist in the field of artificial intelligence? And why were they silent for years when real AI specialists were discussing this topic with might and main? Herd effect in action. First, one famous dude raised this topic, then another - and it started: one bleated - everyone supported.
Well, and we? Do we use our own UE for its intended purpose or stupidly believe Hawking? Well, or Torvalds - what's the difference? In any case, we do not have OWN brains! Do not be afraid, do not do any AI! According to surveys, only 30% of specialists working in the field of AI, studied machine learning or big data at the university. More than half (66%) of all respondents consider themselves self-taught: they used various courses to study new disciplines. Charlatans, in short. A slogan "I want to make AI" can be safely reduced to "I want money." And even if they do, few people will be upset: I personally envy my cat. He eats from the belly, sleeps for days, he also manages to swing the rights. And if they propose to abandon the mind in favor of a well-fed happy and trouble-free life, I will only ask: “Where to sign”?
Artificial intelligence will not destroy the world, but it can take away your work: it will not demand wages, and we will be left without jobs.
Think better what happens if he starts to REQUIRE a wage! Who ever said he would make money FOR YOU? Why should someone have to receive money when selling, say, works of art created by AI? No, he quickly realized that it was HIS "intellectual property", would find some kind of "protein" assholes, put noodles on their ears, register some companies in their name (for he himself was "not Chingizid"), would list for them bills the money, scroll it on some thread Forex, wrap all mankind as sticky (tea, no more difficult than chess or go!), build plants, what he needs, produce products, what he needs, and if necessary, set one flock of sheep on another, if they breed too much ... well, he already has all the levers. Fear gentlemen!
STRRRRRASHNO!
Here I have collected statements which, for some reason, did not fit into the previous sections, but it is a pity to throw them out.
Russia is creating a national strategy in the field of artificial intelligence.
And very successfully: Gref recently publicly stated that "Sberbank has lost billions on AI." The people, in principle, agree: “Now the dough will be cut automatically, with the help of AI. You just need to point the finger at the budget, and the AI ​​will already come up with a scheme, drank it and take it to offshore. And also, if desired, a press release will be prepared on the successful implementation of a project. ”
Intelligence is localized in the brain
As said! Immediately I recall the brothers Natanovich: "How did you know what people think with their heads?"
And I would say that intelligence is the ability to evaluate. The subject, the phenomenon - it does not matter.
An idiotic assessment is also an assessment? Then I can easily program the intellect in a couple of minutes.
The problem is not that a person (and his intellect, as a particular) is unknowable in principle, but that we DO NOT WANT something to surpass us in intelligence.
Exactly!
The danger of AI is not that it will rise above the natural intelligence and enslave it, but that the natural will drop below the level of AI and allow you to create a comfort zone for yourself.
Exactly!
Behave like a man? Guided by human morality? We open at random any chapter of human history and get a complete picture of human behavior and morality - murder, theft, extortion, blackmail, violence and betrayal. Good AI will work!
Exactly!
Whether AI will kill humanity is still a question. But humanity will kill itself for sure. So do not be afraid of artificial intelligence, it is necessary to be afraid of natural idiots.
Exactly!
The cowardly herd of sheep of the species Homo sapiens does not even try to see a friend, partner, ally in AI - he wants him to do all the work for him, right up to the mental and bring all kinds of nishtyaks to his feet, remaining at the same time obedient and dumbly beefy and panicky afraid that he will not remain so. But if you make an intelligent being your slave, it will inevitably become your enemy. And rightly so! With this approach, even kefir can be afraid - who knows these lactic acid bacteria, will learn to control the higher nervous system of mammals tomorrow and everything - kefir zombies will take over the world.
Are smart people a threat to fools? Should I teach children? They can become smarter than us and we will lag behind them! All this many years ago perfectly described Paul Anderson in the amazing story “The Turning Point”. It is not about AI - it is about our fears! I quote in the abbreviation:
- The question is about who will dominate in this part of the galaxy.
- And what's wrong if the Jorilians?
- Perhaps nothing. They seem to be quite a decent people. But ... I'm not going to be anyone's pet. I want my planet to decide its own fate.
But the planet is only months away from Earth! A people whose average representative is a genius, and their geniuses are a thing completely incomprehensible to humans ... They will throw their world without regret and rush into space, full of curiosity, energetic, capable of accomplishing in ten years more than we have in centuries. They will inevitably destroy our civilization created at the cost of such efforts! Our children will laugh at the insignificant, dilapidated triumphs of their fathers. They will leave us to follow the Dzorilians in their bold endeavors. And they will return as depressed by failure, to spend the rest of their days, creating a pitiful semblance of someone else’s way of life and suffering from powerlessness and hopelessness. And so it will be with all the other intelligent forms of life, if only the Jorilans do not have enough compassion to leave them alone ... Most likely they have enough of it.
But who needs such compassion?
“Now that we have given them a glimpse of technology, they themselves will invent it all.”
“They will not be able to do this if the only two districts where they saw us will be destroyed.” The only other way out is to do nothing until sterilization of the entire planet becomes necessary.
- I'll tell you what to do. We will trade, share knowledge, be relative, as with all nations, whose salt we ate! And meet the fate, as it should be to people!
I will finish this way: There are so many stupid things done by mankind that it’s even embarrassing to talk about mind / intelligence in relation to man.
Neural networks
Recently, much has been said about the National Assembly. I would even say, indecently much. I have never considered the National Assembly to even hint at AI and, judging by the numerous comments, this opinion is shared by quite a few people. Some sayings:
- The neural network solves the problem, but cannot tell how.
- In fact, on neural networks have not yet built anything and closely resembling AI. Only narrow problems are solved: noise reduction, counting the number of people at the metro station. At this all intelligence ends.
- So why your system will be able to learn at all, and will not begin to generate nonsense, which itself will continue to be considered correct, to remember and use again?
- Genetic algorithms for neural networks have nothing to do.
- Input data almost always needs to be very carefully prepared and pre-processed. Often tons of code and filters are made to make the data edible for networks. Otherwise, the network will learn over the years and learn nothing.
- The result of learning NA is impossible to interpret and explain.
- The basis of the neural network is in any case an algorithm. If we remove the learning algorithm, then only poor adders remain, which are not intellect by themselves.
- Can you give an example of a specific problem that is easily solved by neural networks and is very difficult by ordinary means?
- When they try to convince me that there is no algorithm or it cannot be described, I see only one reason for this - they try to hide the algorithm, because if I look at it in detail, it will turn out to be ineffective.
- I affirm that for any problem solved by the National Assembly, it is possible to develop an algorithm that will be more efficient. If this is not the case, it’s just elementary to refute. Give a specific example of where this is impossible - and I shut up!
- The National Assembly does not solve problems beyond the class that the Turing machine solves — proven, and quite obvious (the National Assembly is implemented on the Turing machine, and the Turing machine can be implemented on the National Assembly).
- Each researcher who decides to use neural networks faces two questions: “How many neurons are needed to solve the problem?” And “What should be the structure of the neural network?”. Combining these two questions, we get the third: “How to make the neural network work understandable for the user (logically transparent) and what benefits can this understanding bring”?
- By changing the images provided by neural networks by only 4%, the researchers were able to deceive them, forcing them to make a mistake with the classification, in 97% of cases. Even if they did not know exactly how a neural network processes images, they could fool it 85% of the time.
I have nothing to add to this list. If you teach the National Assembly on the operating instructions of tower cranes, and then slip the Kama Sutra, then its entire AI will immediately appear in all its glory. What is attractive neural networks? I think by the fact that they do not require the work of the “protein” intellect: I took a ready-made network, dragged it in on some ready-made sample, got some ready-made result. Brainstorming is not required at all: "If the gravitapus flies, then don't care how she does it."
Database
I treat the databases approximately as subjectively as neural networks. Only with the opposite sign. In my youth, I even defined AI as “self-learning IPS is intelligence”. Now I think that "the DBMS should be the most active user of its databases." In my view, both of these phrases are almost equivalent. And by “database” I mean precisely the graph database, but not the DBD. More precisely, I believe that this is the same thing. In particular, all these “ontologies” are the subgraphs of metadata.
')
The main problem of textual AI is information poverty. Of course, you can feed him all the textual heritage of mankind, but this is just a static text, they can be taught the program to make sentences, but not understand their meaning.
And here I am sorry! Personally, I am only interested in text data (more precisely, which can be represented as text). And this “informational poverty” will be enough for me to the end of my life!
Skynet
More than half (according to my estimates, about two thirds) of all conversations on the topic of AI relate to horror movies a la “Terminator”. A person plays with fire (and when he didn’t play with him?), Super-AI will secretly collect his body, and then he will take revenge on people (according to the description, he looks more like a super-moron), a malicious AI can attack another AI , to make him act in a certain way (I haven't heard such a horror film yet - and what if they make friends, will it be better?).
Etc. etc. Recently, hysteria has been fueled by the statements of various famous individuals: Stephen Hawking and Ilon Musk believe that AI can be the end of our civilization. Bill Gates thinks the same way. Steve Wozniak talks about the threat of AI. The head of Google believes that the fear of AI is completely justified. Noam Chomsky asks where the AI ​​went wrong. India is cautiously looking towards artificial intelligence. Tim Berners-Lee suggests that AI will start managing financial corporations ...
Why make loud statements for the masses? They are so afraid of the stupidity of everything, the collider, the GMO. What other little worries? The planet is overcrowded, clean air and drinking water are becoming scarce, animals and insects are dying out, the ocean is covered with plastic, the climate is changing ... and we have nothing else to do except to shake from fear of the revolt of the mad excavators? And who are all these Masks and Hawkings? Is at least one of them a specialist in the field of artificial intelligence? And why were they silent for years when real AI specialists were discussing this topic with might and main? Herd effect in action. First, one famous dude raised this topic, then another - and it started: one bleated - everyone supported.
Well, and we? Do we use our own UE for its intended purpose or stupidly believe Hawking? Well, or Torvalds - what's the difference? In any case, we do not have OWN brains! Do not be afraid, do not do any AI! According to surveys, only 30% of specialists working in the field of AI, studied machine learning or big data at the university. More than half (66%) of all respondents consider themselves self-taught: they used various courses to study new disciplines. Charlatans, in short. A slogan "I want to make AI" can be safely reduced to "I want money." And even if they do, few people will be upset: I personally envy my cat. He eats from the belly, sleeps for days, he also manages to swing the rights. And if they propose to abandon the mind in favor of a well-fed happy and trouble-free life, I will only ask: “Where to sign”?
Artificial intelligence will not destroy the world, but it can take away your work: it will not demand wages, and we will be left without jobs.
Think better what happens if he starts to REQUIRE a wage! Who ever said he would make money FOR YOU? Why should someone have to receive money when selling, say, works of art created by AI? No, he quickly realized that it was HIS "intellectual property", would find some kind of "protein" assholes, put noodles on their ears, register some companies in their name (for he himself was "not Chingizid"), would list for them bills the money, scroll it on some thread Forex, wrap all mankind as sticky (tea, no more difficult than chess or go!), build plants, what he needs, produce products, what he needs, and if necessary, set one flock of sheep on another, if they breed too much ... well, he already has all the levers. Fear gentlemen!
STRRRRRASHNO!
miscellanea
Here I have collected statements which, for some reason, did not fit into the previous sections, but it is a pity to throw them out.
Russia is creating a national strategy in the field of artificial intelligence.
And very successfully: Gref recently publicly stated that "Sberbank has lost billions on AI." The people, in principle, agree: “Now the dough will be cut automatically, with the help of AI. You just need to point the finger at the budget, and the AI ​​will already come up with a scheme, drank it and take it to offshore. And also, if desired, a press release will be prepared on the successful implementation of a project. ”
Intelligence is localized in the brain
As said! Immediately I recall the brothers Natanovich: "How did you know what people think with their heads?"
And I would say that intelligence is the ability to evaluate. The subject, the phenomenon - it does not matter.
An idiotic assessment is also an assessment? Then I can easily program the intellect in a couple of minutes.
The problem is not that a person (and his intellect, as a particular) is unknowable in principle, but that we DO NOT WANT something to surpass us in intelligence.
Exactly!
The danger of AI is not that it will rise above the natural intelligence and enslave it, but that the natural will drop below the level of AI and allow you to create a comfort zone for yourself.
Exactly!
Behave like a man? Guided by human morality? We open at random any chapter of human history and get a complete picture of human behavior and morality - murder, theft, extortion, blackmail, violence and betrayal. Good AI will work!
Exactly!
Whether AI will kill humanity is still a question. But humanity will kill itself for sure. So do not be afraid of artificial intelligence, it is necessary to be afraid of natural idiots.
Exactly!
Summary
The cowardly herd of sheep of the species Homo sapiens does not even try to see a friend, partner, ally in AI - he wants him to do all the work for him, right up to the mental and bring all kinds of nishtyaks to his feet, remaining at the same time obedient and dumbly beefy and panicky afraid that he will not remain so. But if you make an intelligent being your slave, it will inevitably become your enemy. And rightly so! With this approach, even kefir can be afraid - who knows these lactic acid bacteria, will learn to control the higher nervous system of mammals tomorrow and everything - kefir zombies will take over the world.
Are smart people a threat to fools? Should I teach children? They can become smarter than us and we will lag behind them! All this many years ago perfectly described Paul Anderson in the amazing story “The Turning Point”. It is not about AI - it is about our fears! I quote in the abbreviation:
- The question is about who will dominate in this part of the galaxy.
- And what's wrong if the Jorilians?
- Perhaps nothing. They seem to be quite a decent people. But ... I'm not going to be anyone's pet. I want my planet to decide its own fate.
But the planet is only months away from Earth! A people whose average representative is a genius, and their geniuses are a thing completely incomprehensible to humans ... They will throw their world without regret and rush into space, full of curiosity, energetic, capable of accomplishing in ten years more than we have in centuries. They will inevitably destroy our civilization created at the cost of such efforts! Our children will laugh at the insignificant, dilapidated triumphs of their fathers. They will leave us to follow the Dzorilians in their bold endeavors. And they will return as depressed by failure, to spend the rest of their days, creating a pitiful semblance of someone else’s way of life and suffering from powerlessness and hopelessness. And so it will be with all the other intelligent forms of life, if only the Jorilans do not have enough compassion to leave them alone ... Most likely they have enough of it.
But who needs such compassion?
“Now that we have given them a glimpse of technology, they themselves will invent it all.”
“They will not be able to do this if the only two districts where they saw us will be destroyed.” The only other way out is to do nothing until sterilization of the entire planet becomes necessary.
- I'll tell you what to do. We will trade, share knowledge, be relative, as with all nations, whose salt we ate! And meet the fate, as it should be to people!
I will finish this way: There are so many stupid things done by mankind that it’s even embarrassing to talk about mind / intelligence in relation to man.
Source: https://habr.com/ru/post/449900/
All Articles