Watson defeated one competitor who previously achieved 74 consecutive victories in the game and another competitor who won it for a record amount of 3.25 million dollars ● The sum accumulated by the IBM supercomputer was 4,400 dollars ● It competed under conditions identical to those of the human competitors, and processed the answers to questions based on databases that include over 200 million pages ● To decide if he is safe in the competition, he had three seconds at his disposal in which he had to run 100 different algorithms - a process that would have taken two hours on a standard desktop computer
In a video posted on the website Engadget Shown how Watson's supercomputer IBM (IBM) defeats the American TV game show champions Jepardi (known in Israel as the "King of Trivia") Ken Jennings וBrad Rutter. Jennings previously achieved 74 consecutive victories in the game and Rutter won it for a record amount of 3.25 million dollars.
However, the defeat of the human players was not complete: the filmed round was designed to allow them to practice competing against the computer, and they still managed to win certain amounts of money: Watson won $4,400, Jennings came in second place with $3,400 and Rutter - in place the third with 1,200 dollars.
Watson, which is equipped with the Linux operating system and operates at a rate of 80 teraflops, processes the answers to questions based on databases that include over 200 million pages. Like his human opponents, he is not entitled to any help from the Internet during the competition, and to respond he must activate a buzzer. In order to participate in the competition, he is required to pass the official entrance exams.
During the game, the supercomputer demonstrated the strategy that is considered the best in it: if you are not sure of the answer with a high probability - it is better not to answer at all than to guess. To decide if he was safe, Watson had three seconds at his disposal in which he had to run 100 different algorithms - a process according to David Procchi, the manager assigned to the supercomputer, might take two hours on a regular desktop computer.
63 תגובות
Yes. I brought my own little shovel as compensation.
The idea I tried to convey in all the comments is quite simple. I will try to explain briefly:
Intelligence=thinking dynamism.
When you talk about learning systems, then they may be intelligent but in their field. So as Einstein said - everything is relative. For me, they are not intelligent in relation to me or the world in which I examine their intelligence. Therefore, because their abilities are limited to such a narrow field, I can only compare them and say "this one is better than that one" and if you want, replace "good" with intelligent.
These systems are dynamic in thinking when you examine their conduct in their field, but as soon as you expand the framework and add an unrelated field, they will not be dynamic.
But I'm just trying to define. Does not determine decisively of course. This is a topic that is very much related to definition and relativity and it really cannot be summed up in one sentence apparently.
Eyal A,
digging? You wrote a very long response that claimed to give a definition for intelligence and in the end it is not at all clear what this definition is. Does the fact that Leviathan studied show that he is intelligent? Do you think learning and teaching is intelligence? So there are many learning computer systems that improve performance, are they intelligent?
You killed me laughing. I have not been here for a few days and you have already managed to produce another scroll for the glory of the discussions of the site of science, as per the best tradition (and without a single word about evolution!...well, just one). At the beginning of each comment section of an article here, you should put a "Caution, dig!" sign.
Still, I skimmed through the comments. I want to go back and refine my intention regarding intelligence:
Remember Jurassic Park? Just an example from a movie, but for the purpose of illustrating the idea. In the scene where one of the men goes out into the woods with a shotgun and just before being hunted he realizes he's walking into an ambush, he says something expressing surprise at their unexpected cleverness. In fact, the scene depicts a situation where a non-human creature will deliberately, consciously, trick the mind of a human being. The ambush was planned and realized while the predator and prey were making progress. The predator "planted" an illusion in the prey's mind and thus manipulated the man's mind so that he would fall into the trap.
In short, this is an example that I liked that a screenwriter tried to give a creature that no longer exists with a higher level of intelligence than we would expect. So without all the ramblings I wrote about manipulation, etc., it is clear that intelligence is attributed (albeit in the film) to this creature.
And you really don't need to go to the realms of imagination. Although we do not attribute intelligence to a rat in everyday life, but if we take for example animals that are a little more communicative and...intelligent, we can better understand the definition of intelligence that we unwittingly pull out every day: a killer whale, certainly more intelligent than a rat. An interesting example to illustrate - about 10-15 years ago, a case of a white whale being killed by a killer whale was observed and documented for the first time. Then the incident happened again. After years of investigation it turned out that these are whales that belong to a group that mostly lives in a certain area of the Pacific Ocean (if I remember correctly) that due to competition for food and constant friction with a population of sharks, pressure was created on this group of whales. This led them to find a solution by discovering the method for killing sharks (it is not a matter of strength nor a group matter but something more sophisticated consisting of several steps) and studying the other members of the group. It is an intelligent creature. He learns and teaches. When you say that he is learning, you are actually saying that he adapts himself to new conditions - changes the way he thinks and acts in order to "win" in life.
Therefore, what Point said in response 9 is also irrelevant in my opinion. Duplicate the software and hardware of every existing computer with a small twist - that the calculation speed will increase a million times. You'll still get a computer that does the exact same thing(!) just in a much shorter amount of time. that's it.
In conclusion, I tried to give a kind of definition for intelligence (albeit in a somewhat cumbersome way..)
R.H.:
I am not contradicting myself. Important again.
After all, if it was impossible to prepare for psychometrics, there would be no need to ban the preparation!
Therefore, the claim that it is possible to prepare fits perfectly with the claim that it was desirable to ban the preparation. What is so hard to understand about this? And see a contradiction in that?!
I explained that it is possible to prepare and I explained why the tests are planned the way they are planned: these tests do not refer to the material studied normally and therefore - if you do not prepare for them - they can measure intelligence. If you prepare - everything is destroyed.
It is understood, by the way, that if everyone prepares to an equal extent - success in tests can still represent intelligence, but since this is not the case, the goal is not really achieved.
from the devil,
You are contradicting yourself. On the one hand, you claim that it is not true that it was impossible to prepare for the psychometric test, and on the other hand, you say that the preparation should have been banned, that is, as I claim, if you prepared and improved your score, the test whose purpose is to test skills actually failed. After all, what do you learn in the psychometric preparation courses? You don't learn to raise skills or abilities (which is impossible) but you learn how to solve the test effectively and here it fails.
I agree that the Turing test tests the intelligence of the examiner no less than that of the examinee (or whoever programmed it).
Guy, when a computer passes the Turing test, it will be able to develop new strategies for solving problems, it will know when a joke is funny and when it is not, it will know how to recognize sarcasm and the upside down, it will be able to analyze a story on its layers, it will know whether a picture/woman/song is beautiful or not and so on and so forth Maybe he will be intelligent.
It is not true that a psychometric test should not have been possible to prepare for.
Even the very fact that someone could theoretically (by luck or cheating) do exactly the test we will be tested on - even during the preparation and find out in advance what the correct answers are - demolishes this claim.
The claim completely falls apart when you understand how to construct a psychometric test.
The idea is to let people deal with problems in unfamiliar areas.
That's why they invent questions that have nothing to do with the material being studied or material that has any direct use and test the students on it.
In order for the test not to miss its purpose, the preparation should have been banned - but this is probably impossible.
That is why people prepare and when they come to the test - it is no longer about unfamiliar fields and the degree of preparation - so the degree of success.
The Turing test actually tests the intelligence of... the examinee.
I think that when a computer passes the Turing test and fools everyone it will be intelligent.
R.H.:
Indeed, every test checks who can pass the test (the Turing test is no exception in this regard).
In fact, Turing himself predicted that a computer would pass the Turing test for a significant percentage of test takers by the year XNUMX, and in this respect he underestimated the difficulty of the task.
ravine,
In my opinion, the Turing test checks who can pass the Turing test, and I will explain. This is similar to psychometrics that supposedly tests skills, but the very fact that people after a course improve their grades shows that the test does not really show what your ability is. You would not be able to prepare for a real ability test.
The IQ test, which is also supposed to measure intelligence, is actually aimed at a small, western segment of people. An Eskimo or Bedouin aboriginal will fail the test even if they are geniuses in their natural environment.
So even when there is (and there will be, I assume soon) a computer that passes the test and fools everyone in the Turing test, will it really be intelligent in your opinion?
They once said that a computer can't win at chess or recognize faces, and all of this came true, so Turing will also be passed. There are actually already a number of programs that have managed to mislead judges in the test. For example, Mrs. Suzette:
http://www.newscientist.com/article/dn19643-prizewinning-chatbot-steers-the-conversation.html?DCMP=OTC-rss&nsref=online-news
R.H.:
I agree that the concept is elusive, but the Turing test nevertheless manages to capture and delineate some of the obscurity.
Turing began the famous paper by presenting the following test:
There are two people. One is a man, the other is a woman. Everyone sits in a separate sealed room and communicates through a terminal with a third person. The role of the third person is to determine who the woman is, where the man's task is to trick him into thinking he is the woman and the woman's role is to try and convince that she is the woman. Turing then suggests replacing the man with a computer.
What does this test capture? The male actor will do a good job if he manages to "get into" a woman's head at least as well as a woman does. In the same sense, a computer will do a good job if it manages to "get inside" a person's head, i.e. convincingly imitate the human way of thinking.
Another matter that the test sharpens is the lack of dependence on the implementation level - it is not relevant at all what the computer is made of, it can be made of bacteria, chips, or anything else but this is not relevant for the purpose of examining thinking abilities.
What computers currently lack in order to be able to think at all has nothing to do with the material from which computers are made. What is missing is an understanding at the algorithmic/computational level. If we were to assemble a computer out of neurons with the knowledge that exists today it would not become self-aware.
R.H.:
But I'm still sure.
Pay attention to the article I wrote about the development of language:
https://www.hayadan.org.il/birth-of-language-1110084/
In my opinion - as I mentioned in the article - the basis of language is not a collection of symbols but the very ability to use symbols.
That is why a person has no difficulty in learning any language (as long as he is able to produce and absorb its symbols - which is probably limited by the ability of the senses to absorb certain wavelengths - both of sound and of light) be it computer language, sign language, or the languages of other nations.
In my opinion, this is an ability that is essential for the very development of thought at a level that allows the creation of an aircraft.
By the way, in my opinion the Turing test is not really defined and the definition of problem solving is more accurate - even if it is difficult to examine. The degree to which the conversation in the Turing Test challenges the intelligence of the interlocutor depends too much on the intelligence of the person trying to find out if it is a computer or a person.
from the devil,
I'm not sure at all. The Turing test is designed for a person based on the person's communication abilities. Note that of the millions of species on Earth, only humans can pass a Turing test (and not all of them either...). It's hard to even begin to imagine what the communication signals and logic underlying extraterrestrial communication would be.
In addition, and this goes back to the topic of the article, in my opinion very soon there will be computers that will pass the Turing test and you will not be able to tell in a simple conversation that they are not human. In fact, even today there are some who have overthrown the judges in the "Turing" contests.
R.H.:
When something makes me laugh, I don't usually ask people whether to laugh. Imagine if I would act like that in a comedy - it's just not polite 🙂
I have no doubt that alien intelligence (I'm talking about the real intelligence of beings that created flying vehicles and flew to Earth and not about alien bacteria that arrived on a meteor) will pass the Turing test (after a certain period of study in which you will learn the language and the local culture) easily
from the devil,
Funny or not, argue with Eshel ben Yaakov. In any case, I believe that if we intend to build artificial intelligence or, alternatively, one day contract with other intelligences, it is appropriate that we at least have some kind of general and acceptable definition of the concept.
Do you have any doubt that alien intelligence would fail the Turing test for us?
R.H.:
In my opinion, the accepted definition that I quoted - includes humans and does not include bacteria.
I also pointed out a whole series of problems that bacterial colonies are unable to deal with on their own.
This "IQ test" is also quite funny because beyond what the ghost said is true, the grotesqueness of the description stands out in the following passage:
The recent study shows that everyday pathogenic bacteria are not so smart: their S-IQ score is just at the average level. But the social intelligence of the Vortex bacteria is at the "genius range": if compared to human IQ scores it is about 60 points higher than the average IQ at 100. Armed with this kind of information on the social intelligence of bacteria, researchers will be better able to outsmart them, says Prof. Ben-Jacob.
R.H
Regarding the link you posted: this only proves that a different measure than the human one is needed to test the IQ of the bacteria.
This means that it is not possible to measure the level of 'intelligence' of the bacterium with the same indicators as the level of human intelligence is tested with.
If there is another test that can compare the two IQ tests, and from which it will be possible to draw logical conclusions with meaning, only then I suppose it will be possible to compare the intelligence of man with the intelligence of bacteria.
In the same matter, it was just announced today that Prof. Eshel Ben Avraham from Tel Aviv University created an IQ index for bacteria...
http://www.sciencedaily.com/releases/2011/01/110124111138.htm
ravine,
The more I think about it, the more I am convinced that there is no good definition for the concept. It is somewhat similar to the elusiveness of the concept of beauty, everyone knows what is beautiful but it is very difficult to say why.
Apparently, the ability to shoot a basket or drive in a built-up area or play a crazy computer game does not require great intelligence or wisdom, but if you think about it, the brain makes very complex calculations here that, in my opinion, do not fall short of solving a problem in chess, for example. What is more, it is done in some automatic and unconscious way.
I don't remember who said that the difference between a genius and an eccentric is that a genius remembers the entire Bible by heart and an eccentric remembers the phone book.
R.H.:
I did not provide a definition for intelligence, but I am ready for the purpose of the discussion to go for the Turing test. Turing also consciously avoided a definition of the concept but gave a very practical test that could answer the question.
This test is really biased towards humans, but intelligence is a human concept, and the essence of the test is to capture exactly the human characteristics of intelligence (for example, a correct and fastest answer to the question of how much is 10.12321 to the power of 232324 suspects the season, being a computer instead of the opposite).
This test, by its very definition, puts the person as the highest measure of intelligence.
To your question, the ability to shoot the basket is not measured in this test at all. In addition, this test does not distinguish between Einstein and Michael Jordan. It sets a goal that has not yet been reached for the abstraction and understanding of computers (which are "better" than humans in many things but "weaker" than them in others)
And about the question of awareness I don't know. How do you define awareness and how would you define intelligence?
I think that in zero order one should ask, is self-awareness required as a necessary condition for intelligence? or awareness of the problem being solved. This question is sharpened by the paradox of the Chinese room. On the other hand, Turing's approach is, if it looks like a duck, it walks like a duck, and it quacks like a duck, it's probably a duck. Turing tries to bypass the question of whether a machine will ever have self-awareness and what this means.
Michael and Guy,
I'm sorry but I still haven't received from you any definition of intelligence where man will be more intelligent than other creatures. And also you did not answer me how you will recognize that an alien/computer is intelligent.
It is clear, for example, that shooting at the basket while moving forward quickly requires complex calculation and precise timing. Obviously, Michael Jordan beats Einstein in a shooting contest, is that why he is more intelligent than Einstein? If not why? Since the calculation is done subconsciously, does it detract from his intelligence? If so, do you think the definition of intelligence is related to awareness?
R.H.:
Turing was right when he removed the physical properties and the definition based on solving problems in unfamiliar domains is also freed from the physical properties.
The point is that, as I said - a population of bacteria does not even meet the logical characteristics of the matter and it is not capable of solving any of the problems I described.
Now - it is clear that everything is a matter of measure and a bacterial population is more intelligent than a stone and indeed a computer is more intelligent than a stone, but these are such tiny intelligences that there is no point in treating them as intelligence.
Referring to these phenomena as intelligence, in my opinion, only reduces the chance that we will ever succeed in creating something that deserves to be called artificial intelligence, because in order to acquire the real ability to solve (and even identify) problems in a field unknown to the machine, more fundamental insights are necessary and this is not a question of quantity.
ravine,
It is clear to all of us that the collection of cells known as a person is more intelligent than the collection of cells that make up the culture of bacteria or worms or plants. My question is which setting shows this difference? Is the ability to solve a mathematical problem? A chameleon, for example, makes a complex calculation in three dimensions while moving and shoots its tongue exactly at a fly flying in the air, is it intelligent? Birds or salmon navigate as if they have a GPS, do you know how many calculations this navigation involves? Bacteria manage to cope with a hostile environment such as a hospital saturated with antibiotics and cleaning and disinfection agents, is this intelligence? And in my opinion the algorithm they use (in this case trial and error) is completely irrelevant to the question.
If we meet an alien, how will we know if it is intelligent or not? Will whether or not we manage to contact him prove anything?
R.H.:
I don't think that the ability to organize bacteria in a way that gives calculation capabilities makes bacteria smart or intelligent. In the same way, the inability of humans to perform calculations does not make their neurons intelligent. Intelligence is the characteristic created by the way the components are organized and the patterns of information transfer and probably does not depend on the level of "realization"-the neurons/semiconductors/bacteria (as Douglas Hofstetter said: "it aint the meat its the motion").
It is true that man is made up of cells and molecules, but so is the chicken I eat, as well as worms and plants. Why are there groups of cells that will be called intelligent and those that are not - this is exactly the special arrangement of those foundation stones and this special arrangement is given the name "person" or "computer". The bacteria/cells/chips are just the substrate.
Turing tried to tackle the question by setting up his famous test. In this test he removed all the "physical" elements from the question of when a computer will be considered as intelligent as a person. He did not look at the way a person looks, nor how his voice sounds, nor how his skin feels, but only at the way he communicates with a person so that the person thinks that his interlocutor is a person.
It belongs, it shows you that the definition of intelligence as the ability to deal with problems in unfamiliar situations is only partial if at all and does not really provide the difference between a human and a computer or other organisms.
Nice, but what does it belong to?
So you can make a computer out of bacteria.
Strings are also possible!
Although a little with the help of man, but please:
http://www.guardian.co.uk/science/blog/2009/jul/24/bacteria-computer
R.H.:
The molecules also solved the problem of survival and even better than the bacteria.
Many elementary particles solved it better.
The term intelligence was invented by humans to define something that serves them and it should not be amazing or considered out of order if it has a bias towards humans.
When you say something is "delicious" you are not even talking about all humans but only about yourself.
In any case - no bacteria has solved the problems you are talking about and no group of bacteria will pass IQ tests successfully.
Michael,
I strongly disagree with you. Bacterial populations have solved the problem of how to live in changing conditions from Antarctic glaciers through the bowels of the earth, hot water springs, stomachs of creatures to the same arsenic-saturated lake where bacteria were discovered last month. I guess if their survival depended on solving a chess problem they would solve it too.
Additionally, why does the fact that a computer does solve chess problems not make it intelligent?
In any case, the definition of intelligence is vague and very biased towards the person's abilities in my opinion. This is similar to how an Eskimo or Aboriginal survival genius would fail IQ tests. Is that why his intelligence falls short of a western person?
Man is also a collection of cells and he is also a collection of molecules, but it is much more justified to define him as one entity than a collection of bacteria that have almost no cooperation between them and really no specialization.
All this, of course, does not change the important fact that even as a collection the bacteria cannot deal with almost any problem. Not with a problem in geometry, not with a chess problem, not with a problem in education, not with a problem in flying rockets to the moon - simply nothing! All they "know" as individuals is to change and kill themselves and this allows them as a group to (sometimes) find a version that survives.
In my opinion, we are also very far from computers that deserve the title "intelligent"
God,
Man is also a collection of cells, a parallel computer is also a collection of processors and the culture of bacteria is a population of cells that communicate with each other. In short, I very much agree with Elad Kasum and the others above who claim that already today a great many computerized actions (and I would add biological ones) can be defined as intelligent and the difference is in the complexity of the network that produces them and not in anything "beyond". I understand, and correct me if I'm wrong, that you agree with this assumption. Hence, an intelligent computer is probably on the way and not in the distant future.
R.H.:
Not true!
I also clarified the differences in my previous response and will repeat them:
1. This is not a bacteria but the population of all the bacteria in the world together.
2. It is not about solving general problems but about solving a very specific and defined type of problems.
This is, as mentioned, if the entire population of bacteria is treated as a single entity - something that can be opposed in itself, but let's assume that we allow it. No problem is solved at the level of the single bacterium.
Clarification about Watson's abilities is missing here.
Does Watson receive the questions as text or as a voice decoder that works in real time and knows how to convert words to text. Because just this development alone is worth millions.
An example of the "success" of technology today:
http://www.youtube.com/watch?v=KyLqUf4cdwc
Michael,
You are now debating the algorithm and not the definition of intelligence. True, the algorithm of the bacteria is trial and error until the solution is obtained, but it is probably a very effective algorithm because it does solve the problem. Hence, if we agree that people are more intelligent than bacteria or computers, then the definition of solving unfamiliar problems for intelligence is insufficient.
Ghost:
You didn't understand me but the whole thing is really not important so let's get out of it
Machel
You claim there is no debate, so why are we still arguing? In this debate between us, only you claim that the debate is about the programming of the debate itself. I don't want to, nor did I start debating the debate like you do in response 27.
I just stated for you the point on which I disagree about Ail.A in my previous comments.
Ghost:
The debate is only about the question of whether there is a debate.
You claimed there was and I said there wasn't.
So now you say "there is no debate - so what is the debate about".
In other words - you accept my words and ask why I said them (even though if I hadn't said them you would think something else about whether or not there is a debate).
In your reply to Eyal, you repeat over and over the wording he used - which can perhaps be understood in two ways and try to present it as bearing the least likely meaning - just to keep the debate going.
As Eyal said - it is a question of definitions and in fact there is no debate (something you yourself agreed with in part of your response).
R.H.:
I already answered you about this in another discussion.
Even in your description - there is no bacteria that solves any problem.
The bacterial population (sometimes) overcomes the problem and it does so by checking lots of possibilities in parallel (just like the computer - but at the same time because there are many "computers").
Therefore - at most you can claim that the entire population of bacteria is intelligent - but this is also not quite true - because beyond the fact that it is a mass suicide in which some survive by chance, it is also about solving a very specific type of problems and not a general ability to deal with problems in an unfamiliar environment.
Michael and a ghost,
By and large, our opinion seems to be more or less in line. The question of what is the definition of intelligence is indeed correct, as in many other cases, the definition is the problematic factor in the debate.
In my original response to the news, I meant that the computer in question is not much different from the other computers we have known since the invention of the computer, if we compare them to an animal such as a rat. As they sometimes say about certain people - "He is so fixed in his thinking". So the computer is an extreme example of this. It is completely fixed. Let alone that he can't talk to me because all he knows is the game and how to behave in front of the game moderator, nothing except external intervention in his software will make him change his behavior.
And yet, with these data processing capabilities, I believe that the software was written in a way that is more reminiscent of the operation of a brain (for the matter of a rat), probably this would have come at the expense of the huge database that was written for the purpose of the game, but then you know, the computer could have decided to do Things he wasn't designed to do.
But as we know, we currently have no real idea about how the brain works. So there is nothing to expect from computers to come to life in the meantime.
Michael,
What is your definition of intelligence? If it is "the ability to deal with problems in unfamiliar situations" (as you wrote in 13), then according to this definition, bacteria are more intelligent than humans. Both in the comparison of a single bacterial cell against a single human cell and also of a bacterial culture against a multicellular organism (ie a culture against a human) they will win in dealing with an unfamiliar situation especially if we are talking about survival which is what interests them. That's why I think the definition of intelligence is much more complex and elusive than this definition.
Machel
You seem to agree if we both do, so I don't see where the argument is between us. I guess we agree.
You phrased the things I wrote in response 21 in a different way.
Regarding Ail.A he wrote: 'In my opinion there is no intelligence for computers today, not even at the level of a rat.'
I mean, he thinks that computers today do not have intelligence even at the level of a rat. Or, no intelligence at all.
In my opinion this is not true or inaccurate. Computers have a kind of intelligence, of course it is not really intelligence and it is not similar to human intelligence but it is the computer's intelligence.
I guess we all agree on that?
Ghost:
Eyal said that computers don't even have the intelligence of a rat and you agree with that - so where is the argument?
Everyone knows that computers do very beautiful things and many of them are things that a human cannot do at all, but the question of whether what they do deserves to be called intelligence is another question and according to the definition of intelligence, it seems that their actions still do not deserve to be called that.
Different forms of learning also exist in bacteria and no one claims that bacteria are intelligent.
Note - no one here said that artificial intelligence is not possible - all that was said is that what has been achieved in the field still does not deserve to be called intelligence.
I say this as someone who built a system that replaces the night shifts with a computer unit and everyone who watched it said it was more intelligent than the people it came to replace.
She also continuously studied the situation and responded effectively to changing situations.
The point is that she just effectively implemented a theoretical solution I found to the problem of managing night shift tasks, and if you asked her how to thread a needle, she wouldn't understand what you were talking about.
Machal and Eyal.A
From the little that I know on the subject, the mistake that emerges from Ayal.A's words is in the following sentence that he wrote:
'In my opinion, there is no intelligence for computers today, not even at the level of a rat.'
On this point I think he is wrong.
Today there are already computational systems that know how to execute commands at a level that can be considered as 'intelligence'.
Although it is not intelligence but only a collection of commands written in computer language.
A computer (or any computational system) may not yet have rat-level intelligence, but what there is
It is an ability to execute commands that are 'autodidactic'. The robot's self-taught commands are a representation of a type of intelligence (artificial intelligence) even if it is at the most primitive level.
The very fact that a robot can learn by itself or predict an action that will better suit its next step (with the help of pre-written algorithms) is already a confirmation that there is some kind of intelligence in the machine.
See for example:
http://ccsl.mae.cornell.edu/emergent_self_models
That is, computers today have 'intelligence', it is called artificial intelligence. But it is at a very low level compared to human intelligence. It is at such a low level that it cannot be called intelligence but a computational system that executes commands. But, every creature has a system that knows how to execute commands, or in other words, the body executes commands of the brain, but the brain also executes commands somehow.
So how the brain chooses to carry out commands has not yet been definitively discovered, but how the robot chooses to carry out commands is indeed known.
Here is another link (in my opinion, the robots in this movie are much less 'intelligent' than the robot from the previous link)
http://www.youtube.com/watch?v=9vwZ5FQEUFg
Ghost:
It seems to me that you did not understand the intention of Ayal.A and you actually agree.
Eyal. A
You are wrong.
Computers today have what is called 'artificial intelligence'.
Note:
'Artificial intelligence' is similar in its operations to what is called intelligence but it is not intelligence, and it is not human intelligence, although it was created to try to imitate human intelligence.
If you compare artificial intelligence to human intelligence, then its abilities to function in a way that would be considered intelligent are still at a very low and undeveloped level. It still cannot imitate human intelligence.
If Michael Rothschild, who understands this much better than me and you, chooses not to answer you, then I will answer you in detail later.
Commenter 16, in my opinion there is no intelligence for computers today, not even at the level of a rat. The whole point is the unlimited ability to study and manipulate an existing database on which data processing is based. And said Yaffe Michael. Accordingly, a rat will learn in various experiments to obtain what is desired. This supercomputer would have failed miserably (most likely) if they had changed the rules of the game a little. And he wouldn't come back to himself for another million years. It is programmed to work in a very specific way. It doesn't have algorithms to change algorithms, that's my opinion. The fact is that in order to decide whether to answer or not, he has 3 seconds in which he has to run 100 different algorithms. I mean, if I gave him half a second, how would he perform? Would he have changed at all the way he made decisions about whether to answer or not? After all, he can no longer run the same 100 different algorithms. If I am given a task that I am supposed to fulfill at a certain time, and then the rules are changed so that I am supposed to fulfill it under other time conditions, I will change my way of thinking accordingly. I am an intelligent machine - I adapt myself to the environmental conditions and invent the adaptation from scratch. Computers today, do not adapt themselves to the environmental conditions in terms of the way data is analyzed. And those that do adjust themselves, do the adjustment according to coded information, according to how they were programmed. They don't invent how to analyze their data themselves.
The road to artificial intelligence is still light years away...
Nice wording by Michael Rothschild.
And if we can add to his words, then: artificial intelligence, today, is still 'in its infancy'.
Compared to a human, she is as intelligent as a 2-year-old child.
At some stage not far away all these abilities will be concentrated in one computer that will be inside one robot
or in your pocket
Let him play chess with you, give you trivia answers, land a plane for you, identify objects and people
He will write down what was said, he will record important things for you...
Ahhh wait where is my iPhone 🙂
It was artificial intelligence with these algorithms being developed by the computer itself. When a person learns something, he actually develops the algorithm for himself.
This is not intelligence.
Intelligence is the ability to deal with problems in unfamiliar situations.
Making many situations familiar to a computer does not make it intelligent.
You should get some inspiration from this story:
http://en.wikipedia.org/wiki/Profession_(short_story)
Very nice wording by commenter 11.
Very nice wording by Elad
Point, all of this is true on the assumption that we will simulate the brain one by one, as evolution did, but it is very possible that at some point during the research we will understand the principle behind the operation of the brain, and we will be able to reproduce it in a way that is much more economical in resources. This is exactly what Ray Kurzweil is talking about (among other things).
80 teraflops is not enough.
To reach self-aware mind creation the computer must be able to process on orders of magnitude of exaflops. (10 to the 18th power), and this will happen in another 15 years or so.
Totally agree with Elad and someone.
glacial,
Regarding the Chinese room, one can argue. I think that if there is a computer in the room with the knowledge to translate Chinese, you can say that the room understands the language because there is no difference, even in principle, between it and a Mandarin professor who will do the same task.
The difference exists only in our feeling that "software" does not understand what it is doing.
The Chinese professor also has software in his brain that does the same thing, we just don't yet understand exactly how it works, so we are fascinated by intelligence.
As soon as we crack this too, and build appropriate algorithms, translation will suddenly not be considered an intelligent task but rather the ability to calculate and process data.
Although the Google translation is not there yet, but we are on our way...
But all this is in the realm of philosophy.
In the result test of the Chinese room, which is the only quantitative test you can think of, you sent a text to the room in Chinese and received it in English.
I agree about the rest, but that's exactly the point.
We manage to break down specific capabilities into algorithms, and build computers that know how to execute them.
Connect enough of these to each other and it will become more and more difficult to explain why what is in front of you is not intelligent.
And we will add to Elad's list the ability to recognize faces, which until recently was considered an exclusive human ability and today already exists in every new camera or cell phone, the ability to recognize sloppy handwriting, automatic conversion of speech into text, and vice versa - turning text into human speech, and much more.
Every time a computer succeeds in such a task, they decide that it is no longer a real sign of intelligence and look for another obstacle that the computer has not yet reached.
For ever, there is of course the well-known Chinese room test which is an example of the fact that if it quacks like a duck it's not always a duck. Beyond that - it seems to me that one of the main definitions of such intelligence is the ability to learn.. Note that each type of activity you talked about is done by a completely separate computer system without any connection between them. If our Wilson knew beyond playing a trivia game (using the same algorithm) also to fly, travel in a city full of cars, while being a voice response in the communication company that hates you, then I would talk about intelligence. About intelligence in my opinion..
And this is without underestimating Wilson's ability, the ability to understand his text and the ability to come up with a very impressive appropriate answer..
for number 2,
In the book The Near Singularity by Ray Kurzweil there is an interesting definition of artificial intelligence.
He says that every action a computer does is perceived by the public as not intelligent, but only as data processing.
Twenty years ago we would have said that there must be an intelligent pilot at the helm to land a plane, and today it is clear to everyone that intelligence has nothing to do with it, but only data processing...
The same in checkmate, driving, searching databases, automatic answering, landing vehicles on Mars, etc. All of them are done quite easily by computers today, because we found the algorithms underlying these operations. Therefore, they appear to us as something that does not require intelligence, but "only" calculation power.
This means that we are starting to define what intelligence is by what computers cannot do!! (Write a book, discuss moral questions, etc., although it is already possible to start arguing about these as well)
In a very short time, when computers will also succeed in such tasks, we will understand that because in the end everything can be broken down into algorithms, data processing and calculations, organized in very complicated ways, is intelligence.
In entertainment, it's true that it doesn't look like a human being, but if they put the computer there and the people behind a curtain, we wouldn't be able to tell who is who (assuming they synthesized a slightly better voice for him....)
And once it is impossible in principle to differentiate between two things, you can say that they are the same thing
So yes, this is exactly artificial intelligence, at least in my opinion.
Admittedly only in one specific area, but this is exactly how the beginning looks like.
http://www.ynet.co.il/articles/0,7340,L-4014959,00.html
Very nice. Really nice. But still, it doesn't look like intelligence. And with data processing power like that, how could it not be?
This is not the end, and not even the beginning of the end, but it is the end of the beginning..."