What Is Intelligence? On Free Will and Judgment (Column 35)
With God's help
In Column 31 I addressed the question of the rationality of the religious person. That reminded me of several questions and discussions from the past that dealt with the question of intelligence, and I thought this would be a good opportunity to lay out the broader picture on this issue.
What Is Intelligence?
It is very difficult to define the concept of intelligence. It is controversial and heavily burdened with politically correct baggage. Usually it is linked to the ability to understand, draw conclusions, make decisions, and especially to skill in solving problems. The vagueness of the definition has led to its expansion, so that today it is customary to define several types of intelligence.
The definition at the beginning of the Wikipedia entry is:
Intelligence (in Hebrew: miskal) is the totality of abilities by means of which problems requiring thought can be solved efficiently. The word intelligence derives from the Latin word intelligere, meaning "to choose," "to decide," "to discern." Accordingly, intelligence is also defined as the ability to infer one thing from another, as well as the ability to distinguish between one thing and another. Intelligence is not unique to human beings. There is also intelligence in animals and artificial intelligence created in machines.
And yet, this concept was originally associated with human beings, and was even regarded by thinkers (especially earlier ones) as what distinguishes human beings from all other creatures. Is it really true, as is commonly thought today, that animals or machines possess intelligence?
The Turing Test
The renowned mathematician and computer scientist Alan Turing, who was responsible for cracking Enigma (the German code in World War II; this is described in the film The Imitation Game), proposed a test that would determine when machines may be considered intelligent (artificial intelligence). Some formulate it as a test that determines when a machine has become a human being. In this test, a person expresses himself in natural language in textual form (through a keyboard) and receives textual responses in natural language (on a computer screen) from two entities, one human and the other a machine. If the person cannot distinguish which of the two interlocutors is the human and which is the machine, then the machine before him has passed the test, and one may say that it possesses human intelligence (is a human being?).
To be sure, the test is not well defined, since it is unclear who the judge is. A three-year-old child? Or perhaps a ten-year-old? Or a skilled philosopher? Is the judge a human being, or perhaps a machine? How do we know the judge is human? Likewise, one can conduct a one-sentence conversation, or perhaps a three-minute conversation is required, or perhaps shared life over time should be required (as in the well-known film Her, which describes a man who lives with a computer program and falls in love with it). It is also unclear what the topics of such a conversation are supposed to be, and so on and so forth. In any case, this test seems to have the potential to define more clearly vague questions concerning the distinction between human and machine and the definition of intelligence in general, and many people to this day believe that this is indeed the criterion for defining human intelligence.
From then until today, quite a few such attempts have been made, and in 2008 there were two fairly impressive successes. One was a program that managed to fool a quarter of the judges who conversed with it for five minutes and could not determine whether it was a machine or a human being, and the second program, which goes by the name Eugene Goostman, managed to fool a third of them (one should remember that random guessing already yields 50% success, so it is incorrect to compare this to 100% success).
We should remember that the intelligence expressed in conversation is only one among many kinds. In 1996 IBM's Deep Blue program succeeded in defeating Garry Kasparov, the world chess champion. And this is still without mentioning the solving of equations and various mathematical problems that human beings cannot perform, something computers do every single day. A prominent example is the proof of the four-color theorem. The theorem states that there is no two-dimensional map of countries of any type or shape that cannot be colored with four different colors such that no two neighboring countries (sharing a common border) are colored the same color. There is a mathematical proof that with five colors this can be done for any map. The question whether the same is true with four colors had not yet received an agreed-upon answer. In 1976, two researchers from the University of Illinois proved for the first time, by means of a computer, that this is possible for every map. They checked 1,936 different types of maps in a program that ran for about 1,200 hours, while presenting a proof that every map in the world is equivalent to one of those types. Twenty years later Robertson and his colleagues presented a proof that there is another classification in which it is enough to check only 633 types. But to this day there is no proof of this theorem that does not rely on a computer.
Interestingly, mathematicians are divided to this day over whether to accept such a proof as mathematically valid, although I believe none of them has any doubt that the theorem is true. In the eyes of some of them, the concept of "proof" apparently does not coincide with a procedure that brings about complete conviction (think: what would we say about a situation in which hypnosis led us to become convinced of something?).
Critiques: the Chinese Room
The best-known critique of the Turing test is that of the philosopher John Searle, who proposed, by way of illustration, the Chinese Room thought experiment. A person who speaks only Hebrew sits in a closed room, and beside him is an infinite barrel full of Chinese characters. There are two openings in the room. Through one of them, questions written in Chinese are inserted, and he then constructs an answer from the characters in the barrel and sends it out through the second opening. If we had infinite time, and that person received painful feedback for every incorrect answer (for example, an electric shock), it is not inconceivable that over time he would learn to give reasonable and relevant answers to every question. The person outside would be able to converse with him pleasantly and receive answers, without knowing that his interlocutor does not understand at all what he is talking about. If we were to allocate that person finite but lengthy time, along with a large but not infinite barrel of characters, we might perhaps achieve an "intelligent" conversation lasting several months, weeks, days, or hours. And if we cut the resources down even further, we would get a conversation of an hour or a few minutes. It depends on the length of the conversation required by the Turing test (see above). In any case, once you define the required duration of the conversation, there exists some amount of time and some barrel depth that will bring us success on the test.
Searle argued that the person inside the room, even after he has "learned" Chinese, does not really understand the language. He displays a technical skill of replacing symbols, but no understanding accompanies it. In that person's mind there is no parallel semantic process accompanying the exchange of symbols. Precisely for this reason, Searle argued, a computer that passes the Turing test does not really know how to converse. It performs a technical operation of replacing symbols, but conversation also includes a cognitive component of understanding that accompanies the exchange of symbols. Without that, there is no conversation here, and no intelligence, and certainly no human being.
Do Water, Rocks, and Birds Have Intelligence?
In the definition I cited above from Wikipedia, it is said that animals also possess intelligence, and as we have seen there are similar claims about computers. Searle's example shows why this is nonsense. Birds know how to navigate in a very impressive way, and therefore people attribute to them a certain kind of intelligence. But by that logic, water too has intelligence, and rocks as well. Water moves in a way whose mathematical description is astonishingly complex (the Navier-Stokes equations). Physicists do not know how to solve these equations in any nontrivial case, and lo and behold, water does so every day and every hour. It finds the path dictated by those equations in every situation and under every condition without any problem. The conclusion is that, according to the accepted metrics in the field of artificial intelligence, water possesses superintelligence, far beyond the most gifted physicist. Processes also occur inside a rock that are described by equations human beings do not know how to solve—from quantum theory at the level of the microscopic particles inside it, through processes of decay and crystallization that we are unable to compute in detailed and precise fashion. But the rock itself does so without any problem. The unavoidable conclusion is that it too has superintelligence.
When I raised these points before a mathematician acquaintance of mine, he said that this is not intelligence, because it is the solution of one specific problem and not the ability to solve a broad range of many different kinds of problems. I told him that in my opinion this is a weak argument, for two reasons: first, it is merely quantitative. There is very high intelligence here, only it is not broad enough. Is he really prepared to say that water and rocks have intelligence? Second, water knows how to solve a whole array of mathematical problems. The flow of water under every environmental condition is a different problem. Flow in a bathtub, or in a sink, or along some topographical layout, or in a river, or in an ocean, at different temperatures and pressures, with environmental influences (such as a whale swimming in the water). Water does all this without any difficulty, under every condition and in every situation. Does the ability to solve such a variety of problems not deserve to be called intelligence?
So What Is Missing: Consciousness or Judgment?
It seems to me that no sane person would define water or rocks as intelligent. But regarding birds, many people I know think that they do indeed have intelligence. What distinguishes the latter from the former? What do water lack that birds have? Is this the thing that lies at the foundation of the elusive concept of intelligence? Searle's example speaks of consciousness, that is, processes that accompany the technical-motor acts of conversation. In the case of water or rocks as well, one can say that there is no consciousness accompanying these "calculations," and therefore this is not intelligence. But birds may perhaps have such consciousness, and therefore people feel that one can speak of intelligence there. And what about computers? In this respect they resemble rocks and water, not birds. True, their intelligence is called "artificial," but that is only because it was created by human beings rather than occurring naturally. There is no other essential difference between them and water. Is the conclusion, then, that water have non-artificial intelligence? That is not plausible.
But in my judgment, hanging the issue on consciousness is a mistake. The point is not consciousness but judgment. A creature that acts mechanically, as dictated by its structure, is not an intelligent creature. The difference between water and a human being is not consciousness, nor the range of problems each can solve, but rather the decision and choice of the technique of action and thought. A human being decides which route to take in order to solve one problem or another. After the decision, he also has the skills to carry it out, but birds, water, and rocks have that as well. What all of these lack (including the bird) is judgment or decision. They act as they do because of an innate structure within them that they themselves did not choose. This is a mechanical action that does not arise from judgment and choice, and therefore it should not be subsumed under the heading of intelligence. Only a creature that chooses a course of action for solving the problem before it is a creature possessed of intelligence (low or high, depending on the degree of its success). The existence of a toolbox is not intelligence. The ability to use it wisely, exercising judgment, is intelligence. The conclusion is that it is mistaken to say that water, rocks, or birds have low intelligence. They have no intelligence at all.
The same is true of computers. Even if one speaks of decisions being made within them, that process takes place mechanically and is dictated to them compulsorily, just like water or rocks. The conclusion is that a deterministic creature cannot be considered intelligent.
The Accepted Definitions in the Field of Artificial Intelligence
The conclusion is that the definitions of intelligence commonly accepted in the field of artificial intelligence lack any philosophical significance. What is meant there is skill in solving complex problems, but as stated, this captures only one of the two parts of the equation. A mechanical creature that is skilled at solving problems will, on that view, be considered highly intelligent, and the same is true of water or rocks. According to the definition I have proposed here, that is not the case. Only if that creature chooses the solution by means of judgment, and the solution is indeed correct (that is, it solves the problem), only then is it entitled to be considered an intelligent creature. A computer is not such a creature, and therefore a computer has no intelligence.[1]
Freedom in the Intellectual Realm
We are accustomed to speaking of free will, which is responsible for our ability to choose freely and autonomously values and conduct in different situations. The essence of free will is that the step or value-position we choose is not a result dictated by the circumstances that prevailed before the decision. That is the meaning of our freedom of choice. The decision is entrusted to us, and in those very same circumstances, the path is open before us to decide one way or the other, as we see fit.[2]
This discussion concerns the value plane, and there the freedom is that of the will. In contrast, here I have argued that there is similar freedom on the intellectual plane as well; that is, there is freedom of the intellect and not only of the will. Choosing a course of action for solving a factual problem, or choosing one scientific theory from among several possible ones, also involves judgment, where judgment is the opposite of mechanical calculation. This is a free decision, which freely chooses among several options the correct theory, or the correct facts, or the correct way (factual, not evaluative) to deal with some problem. The more this free creature can cope with more complicated problems, the higher its intelligence. But a necessary condition for speaking of its intelligence is freedom, that is, that it possess judgment.
A Definition of Intelligence
The conclusion is that a definition of intelligence includes another component, beyond the one discussed in the field of artificial intelligence. Our intelligence is the ability to decide freely in the correct direction when solving complex problems. It is important to understand that, at least in the realm of facts (and in my opinion also in the realm of values), the correct direction does not depend on us but on the world itself. Our decision that some scientific theory is correct does not make it correct. Unsuccessful scientists will freely choose an incorrect theory, and people who are weak at solving problems will freely choose a poor solution to a problem they have encountered (it simply will not solve it). But the decision depends only on us, and that is our freedom.
In other words, intelligence measures the relation between two things: the decision we chose (freely) and the correct decision. The greater the overlap between them, the higher the intelligence of that creature. If the correlation is low, then this is a creature with low intelligence. But the moment some creature has no ability to choose freely the method or form of solution to a problem, then one cannot speak of its intelligence at all (as noted, not even of low intelligence. It has no intelligence whatsoever).
In the cases of a bird, water, or a computer, we are speaking about the intelligence of whoever or whatever created them, not their own. A computer has no ability to choose freely a way of solving a problem, and therefore the intelligence (the fit between its decisions and the correct solution) is that of its programmer, not its own. The programmer chose to write software that enables the computer to solve problems. He is the one who exercised judgment, not the computer. The same applies to the bird, the water, or the rocks: the intelligence is that of whoever created them, not their own.
If birds were formed through processes of evolution, the intelligence is certainly not evolution's, for evolution too is a mechanical process. If anything, the intelligence belongs to whoever created the laws that govern it and generate it. If we have reached the conclusion that some intelligence underlies our world, then necessarily at the foundation of all the processes taking place within it stands a being with free decision and judgment who established the laws that govern it, and only to Him may the intelligence that appears in the inanimate and living creatures of our world be attributed. No mechanical process along the way can be the source of the intelligence we see around us. This is in fact a version of the physicotheological proof for the existence of God (see my discussion of it in the third notebook here on the site).
Of course, one can also reach the conclusion that the world is deterministic, and that the wisdom embedded in it is mechanical from beginning to end. There exists no creature with intelligence; in fact, there are only mechanical and blind systems without judgment. In such a situation there are no intelligent beings in the world at all—not human beings, not animals, not inanimate objects, and not even God. But on such a view the concept of intelligence itself has simply been emptied of content.
An Argument Against Determinism
If we reverse the picture described here, then the very concept of intelligence is itself an argument against determinism (see also here). In a deterministic-materialist world, the concept of intelligence has no meaning, and therefore anyone who thinks this concept has content is necessarily not a determinist. This is a "theological" proof (see the fourth notebook, which is devoted entirely to this inverse logic), because it proceeds from the conclusion and from it arrives at the premises. The straightforward path is that if we have judgment, then—and only then—can intelligence be defined. Here I have proposed reversing the direction and arguing that if there is intelligence, then necessarily we have judgment (intelligence cannot exist without judgment).
On Religious Rationalism
In Column 31 I dealt with religious rationality. It seems to me that it also finds expression with respect to this issue of artificial intelligence. You can see my remarks on this in the correspondence here (it refers to my correspondence with Yonatan here, all of which concerns this column).
Emotional Intelligence
As I have already mentioned, in recent years the concept of intelligence has been expanded, and several additional kinds of abilities have been brought under this heading. Today people speak of multiple intelligences (eight in number), among them emotional, motor, and several others. It seems to me that this expansion contains several flaws and provides a good example of science in the service of political correctness. I will explain one of them here.
This expansion was carried out in the following way. There was a concept of intelligence commonly accepted by the public. People knew how to point out that Einstein or Maimonides were very intelligent, and Reuven less intelligent than they were. This was generally connected to intellectual and academic abilities. Then the sages of intelligence theory (Howard Gardner, Perkins, and Sternberg) came along and distilled from this intuitive conception several characteristics that define the concept. And lo and behold, there are other human abilities that possess those characteristics. Therefore, those sages concluded, these abilities too belong under the heading of intelligence.
But this gain comes with a sting. After all, the point of departure was an initial intuition about the concept, and the distillation was meant only to conceptualize it and define its content more precisely. Yet it turns out that this conceptualization yields results different from those that follow from the intuitive conception. The unavoidable conclusion is that the conceptualization simply failed. The characteristics that were distilled do not really capture the intuitive concept of intelligence. So why did those (intelligent?) sages choose specifically the second option: to discard the intuition and remain with what had been distilled from it? Such are the wonders of political correctness. This theory is attractive to many people today, because it allows us to regard every person as intelligent. Now not only Einstein or Maimonides are wise and intelligent, but also the cobbler beside me and the shoe seller on the next street, and the soccer player on team X and the artist Y. The former have intellectual intelligence, and the latter emotional, motor, musical, or naturalistic intelligence.
But in light of what we have said so far, it seems that one can raise another claim against the definition of multiple intelligences (or at least some of them). Think of a talented soccer player (for example Franz Beckenbauer, the libero, who was known as a genius). Does he have high intelligence? Perhaps. But one can say this only if he makes decisions by means of judgment. If those decisions are derived mechanically from an intuitive understanding with which he is endowed, and this ability simply flows through him naturally, then it seems to me that it is not correct to attribute high intelligence to him. As stated, intelligence is the result of exercising judgment. The same is true of emotional intelligence. If a person intuitively understands another and behaves properly toward him, but without judgment and conscious thought, it is highly doubtful whether one can regard this as intelligence.
For a similar reason, a sheep is not moral (see the fourth notebook at the beginning of chapter 9), even though it harms no one. It behaves this way because that is its nature and not because it chose to do so, and therefore it is incorrect to regard it as moral. But above we have already seen that, just as choice is a condition for moral evaluation, judgment is a condition for intellectual evaluation.
[1] Computer scientists may perhaps think that a computer that operates by means of a neural network technique does make decisions and does not merely operate mechanically. But this is a mistake. A neural network is an entirely mechanical-deterministic procedure, except that its underlying philosophy differs from that of classical programs.
[2] I am not entering here into the distinction between free will and randomness, which ostensibly can also be described in the same way. I have expanded on this matter in my book The Sciences of Freedom.
Discussion
Hello Oren.
When that person makes a mistaken free decision, his error is under coercion. That is when, in light of the information he has, it is a reasonable decision. A person who grew up secular and those are the data available to him may reasonably decide to remain secular. So he is coerced even though his decision is free. He is not coerced to decide, but a reasonable person certainly can (even if he need not) decide as he does.
That sounds a bit contradictory. On the one hand, you say that his error is under coercion. On the other hand, you say that he could have decided otherwise and thus not erred. So where is the coercion here? Usually coercion refers to situations in which a person could not have chosen otherwise.
Perhaps you mean that there are certain intellectual choices that are not free choices because there is only one option that seems reasonable, and all the others do not. And yet there are certain intellectual choices in which there are several reasonable paths, and only there does intellectual free choice come into play. Correct me if I'm wrong.
I do not see a contradiction here. I claim that the person decides freely, but he is not guilty for his decision. Freedom is a necessary condition for guilt and responsibility for your decision, but not a sufficient one. Take, for example, a person who decides on the basis of partial information. He decides freely, but his decision is mistaken, and he is not to blame for that and therefore not responsible for it.
When a person is in a situation in which his mistaken decision is reasonable, that is, any reasonable person could have made it (though not must have made it—it is a free decision), then he bears no moral responsibility.
I am reminded of a case that reached the Supreme Court in the merry days of Aharon Barak. A woman was driving and a cat crossed in front of her. She swerved and hit a person on the sidewalk beside her and killed him. Aharon Barak convicted her despite the defense's claim that this was the action of a reasonable person (any reasonable person would have done so in her situation), and argued that this was an unreasonable act of the reasonable person. I of course do not agree with him. This is an action under partial information or under circumstances in which it is hard to make the right decision, and therefore, although it was made freely, there is no moral responsibility here for the result.
So too with a secular person who freely chooses not to observe the commandments: I claim that if this is truly what he thinks, then he bears no responsibility. And likewise in the case discussed by the Radbaz. That delusional person's conclusion was not necessary. Not everyone in his situation would have reached it, but it was his conclusion. And that is enough to exempt him from responsibility.
Hello,
And what about Einstein, who was endowed with intuitive genius? Who can guarantee that his genius stems from deliberation (like the soccer player above)? In fact, with this definition, it seems we have no way at all to assess a person regarding his intelligence.
Note that the criterion for intelligence according to my proposal has two levels: 1. A preliminary condition—that he possess deliberation (choice). 2. That the content of his choice be wise (he solves problems correctly, quickly, and efficiently). The degree of intelligence (whether a genius or not) depends only on parameter 2. Parameter 1 is merely a necessary preliminary condition, but not a sufficient one.
As for Einstein, all of us can see 2 in him. As for 1, this is an assumption just as it is with any other person. I assume he has deliberation, meaning that he is not a machine. True, there is no guarantee of anything. By the same token, one may ask how we know that you or I have free choice, or a soul, or consciousness, or anything else. The assumption is that he has deliberation and choice like any other person.
But you wrote regarding the soccer player:
"But this can be said only if he makes decisions through deliberation. If those decisions are mechanically determined by an intuitive understanding with which he is endowed, if that ability simply flows through him naturally, then I think it is not correct to attribute high intelligence to him."
Why, in Einstein's case, can everyone see 2, but in the soccer player's case you qualify it? The distinction is not clear to me.
If you can see it, then it is the same thing. My impression is different. With a soccer player it happens in real time and very quickly, and so it is not likely that there is deliberation there; rather, it is an ability embedded in him. Einstein, by contrast, thought before arriving at his solutions and theories. But as I said, that is my impression. You can of course have a different impression. Bottom line, there is no point arguing about that. What matters is the principle: intelligence is connected only to decisions based on deliberation and not on abilities built into me.
1. Define intelligence as something that requires non-deterministic "free will"
2. From this it follows that in a deterministic world there are no intelligent beings at all (as defined in 1)
3. Such a world sucks
4. Conclusion – there is a God and the world is not deterministic
I don't know exactly what the definition of a theological proof is, but I would not define this move as a proof of any kind.
By the way, I don't see any definition (except one involving potential) under which a chimpanzee (or at least a hybrid of chimpanzee, dolphin, raven, and elephant) is less intelligent than a one-year-old baby.
If that is true, then either they have free choice, or a baby is not intelligent.
Hello Shachar. Ridicule is not an argument, especially when the alternative you proposed is logically flawed.
A. I did not define intelligence that way; rather, I described how we relate to intelligence (unless we want to say that a rock, water, or an electron also has intelligence—which many of us would not say. Perhaps you think water has intelligence, but I think you are in the minority. In any case, that is a semantic debate).
B. This is not about free will but about deliberation (free intellect, not free will).
C. My conclusion was against determinism, not in favor of God. The conclusions regarding God are left to the reader's judgment. But the gates of demagoguery have not been locked.
D. Such a world is not only depressing but incorrect, meaning it does not fit how we conceive of intelligence (cf. water).
E. If you would like to learn what a theological proof is (by the way, almost every proof is one) – go to the fourth notebook. Every proof is based on unproven assumptions. A theological proof starts from an accepted definition of a concept and derives from it, as a conclusion, the assumptions that lead to such a definition. That is exactly what I did here, and what every philosopher does (even if sometimes unconsciously).
E. You yourself make such a move in the last paragraph. You assume what you want and derive conclusions from it.
F. Except that unlike me, you do it unsuccessfully. Why? Because when you moved from actual abilities to potential, you did not escape the need to define (though you tried to run from it. I must say that escape does not seem to me a better alternative than a theological proof). To see this, now think about the question that arises regarding the definition you proposed: what abilities must the adult creature have (who is actually intelligent) so that the infant/young one (who has the potential to become such) will also be considered intelligent?
In other words: you say you cannot define actual intelligence (which is what I tried to do), and then you propose defining the potential for those undefined abilities as intelligence. Do you think that is a definition? You do not meet even the most basic logical threshold here. Not that it is incorrect—it is simply nonsense.
Now, to complete the philosophical picture, note that beyond its being nonsense, it is also a circular and bizarre argument: essentially you are saying that X cannot be defined, and therefore we should define the potential to be X as X. You have not solved the problem and have not answered the need to define it; in fact, you have used the term to define itself.
And this is the alternative you offer to a theological proof? Strange…
G. Now you will surely agree that there is no need to choose one of the two options you offered me at the end of your remarks.
Hello,
I did not mean to ridicule; I meant to present a schema of the proof and show that it is not a proof.
I will address the points you raised:
A. You described how you relate to intelligence. I do not think water is intelligent, but I—and, it seems to me, every other materialist—do think that a human being is intelligent (even without a free intellect). As I wrote, and as you wrote, I think many people also relate to certain animals as intelligent. I do not think that this relation is because of non-materialist deliberation (which I think does not exist), but rather because of a combination of the breadth and depth of the toolbox and consciousness.
C. True, the main argument spoke against determinism, but the way you described intelligence indirectly leads (at least in your view; I am not sure whether I disagree with you) to the existence of God ("If birds were created in processes of evolution… this is essentially a version of the physico-theological proof of the existence of God").
D. Water is not the right example here. At this stage of the argument one assumes there are no intelligent beings at all, so "human being" would have been more appropriate. In any case, the use of "sucks" was the only thing that came close to ridicule. It could have been omitted.
E–G. An alternative definition—vague, because I have no sharper one, and in my view there may indeed be a continuous transition—was mentioned in A. I think my last claim was not properly understood. It was not an alternative definition of intelligence, but an either-or claim.
According to your view, that intelligence requires deliberation, it follows that if "smart" animals are not intelligent, then a baby is not intelligent either. And if they are intelligent, then they must have real, non-materialist deliberation.
I only added that you could evade the either-or claim if you were willing to regard a baby as intelligent solely on the basis of potential—something I *do not* think is justified.
Hello Shachar.
A. The question is whether, according to your view, water has a little intelligence or no intelligence at all. On your view one should say it has a little, not none, since it too can solve problems, just less well (and in your words, it has a toolbox, only a somewhat meager one. As for consciousness, see below). But I think any reasonable person would see such a statement as absurd. By the same token, a sleeping person who happens to move his hands and write a correct solution to an equation, or even solve the unified field equations that Einstein did not find, does not express the slightest shred of intelligence. I do not see how one can argue about that.
I do not understand the whole matter of consciousness at all. Why should consciousness matter? After all, I solve the problem completely mechanically, so why is it important that I am conscious of what I am doing? Consciousness merely passively accompanies the solution and does not affect it. So why should it be a condition for intelligence? That is roughly like saying that because I love watching basketball games on television, I am therefore defined as an athlete.
Let me clarify that the sleeping person I cited above as an example of an action that does not indicate intelligence was not brought up because of lack of consciousness but because of lack of deliberation. Consciousness is not relevant to the question of intelligence. If you meant deliberation, then that is certainly relevant—but that is our dispute.
E–G. This is not an alternative definition but a lack of definition. In my view intelligence requires deliberation, and therefore animals have no intelligence (assuming they have no deliberation. I am not completely sure about that). As for the either-or question, clearly a baby also has no intelligence, not only because it has no deliberation but also because it has no ability to solve problems. What is the question?
The potential-based definition is not an alternative definition that constitutes some sort of possibility one can agree or disagree with. It is circular nonsense.
Bottom line, I think that now you yourself have admitted that a determinist cannot define intelligence in a coherent way that accords with what we regard as intelligence (that is, without saying that water has low intelligence, or that a sleeping person who just waves his hands around and writes, entirely at random, solutions to Einstein's equations has high intelligence). But that is precisely the "theological" proof that determinism is mistaken.
Hello Michi,
Following up on our previous conversation, and on this post, which repeats some of the points you raised there.
My position (I don’t know how representative it is of all AI researchers) is as follows:
1. I have no way of knowing that any other person besides me possesses consciousness (i.e. a "soul," i.e. "free will"), but in order to function in this world I make an operative assumption: whoever speaks and responds to the world as though he has consciousness is, for all intents and purposes, conscious. You can call this faith or an axiom. Therefore I accept that other human beings besides me have consciousness.
2. If a computer passes the Turing test, that is, if it speaks and responds to the world as though it has consciousness (this is not exactly what Turing defined, but for me that is the meaning), then rationally I will have no choice but to let it benefit from the conclusions of that same assumption in 1, namely to assume that it too has consciousness (i.e. a "soul," i.e. "free will"), even if it is made of iron.
3. Conclusions 1 and 2 stand on their own and are not connected to the question of what intelligence is, whether the world is deterministic, and whether or not free will exists in it.
Now regarding the question of intelligence and determinism:
I accept your logic that if the world is deterministic (there is no "freedom" in it), then there is no meaning to defining intelligence as "the ability to decide freely in the right direction in solving complex problems," because then the implication is that there is no intelligence in the world. From here, one possibility (which you chose) is to infer that the world is not deterministic, but the second possibility (my choice) is that the above definition of intelligence is not a good definition.
In a deterministic world in which there is no free choice but only mechanical systems that can be distinguished by the way they process information, the definition of intelligence would be derived from the complexity of information processing in those systems. I do not yet have a sufficiently precise definition of how to look at a mechanical system (such as a water molecule, a bird, or a human being) and determine whether it is intelligent or not. I have an intuition that this can be defined in terms of entropy and information theory, and I need to think about the matter more deeply before I come with a definition of my own.
So why do some of us nevertheless define intelligence as "the ability to decide freely in the right direction in solving complex problems"? Because the model in which we operate in the world is one that contains free will. This does not determine that this model is the "correct" one, only that it is useful—as the saying goes: "All models are wrong, but some are useful." It may very well be that the deterministic model is the "correct" one, but it is not useful for interpreting the world (for various constraints). The definition of intelligence as "the ability to decide freely etc." is useful under the free-will model, and I can only conjecture that whatever the definition of intelligence in the deterministic model turns out to be, it will be correlative to it.
And one final note regarding "free will":
I assume that when you speak about "determinism," you mean it in the broad sense, namely as including dice throws—responses that occur as a result of pure randomness. It may be that what is attributed in our consciousness model to "free will" is (in the deterministic model) the result of a chain of events some of which arose from such pure randomness.
Hello, and thank you for your response. I will address the points you raised according to your numbering.
1.
A. Let me begin by saying that from your remarks in this section it is not clear what you mean by an "operative assumption." This could be interpreted as an assumption about the world that is not certain. Alternatively, it could mean an assumption of fact that claims nothing at all about the world but only about yourself and your own thought. From what you write later, it seems that you probably mean the second sense, and I will address that there.
B. For some reason you identify consciousness with soul and free will. They are not identical.
C. In the discussion of intelligence I am speaking about free intellect, not free will. Intelligence is not connected to desires and values but mainly to thought and deliberation.
D. On your view it follows that at least with regard to yourself, you can say (not merely "operatively") that you have free will, a soul, and consciousness.
E. If that is indeed your conclusion, then I do not see why not generalize and infer this regarding other human beings as well. Just as you infer that all bodies with mass fall to the earth, even though you have seen this only with some of them, or that all photons move at the speed of light, even though you have seen this only with very few of them, if any at all.
Perhaps you mean to say that a generalization based on one case is weak? Fine—but that is still better than adopting the opposite conclusion, which is based on zero cases (after all, you have never seen a human being whom you can say lacks consciousness/soul/freedom).
2. I disagree with you. If I photographed a person, that does not mean the photograph is a person because it looks like him. The fact that two objects are similar does not mean they are the same kind of object. The person standing before me is similar to me, and therefore I infer from my own consciousness to him. A computer is not similar to me, and I know exactly how it was built and that it is only an imitation of certain aspects of human beings, made of iron because someone programmed it to do so. There is no reason in the world to assume that it has consciousness. In my view this is nonsense.
3. Here I did not really understand. You infer that you have free will and operatively assume that others do too, but this has no connection to the question of whether they really do (that is, no connection to the question of determinism)? I assume that here you mean to answer my question in 1A and say that you are speaking in the second sense. But regarding yourself you spoke in the first sense (because there you see it directly). So that certainly does have a connection to whether the world is deterministic or not. If the truth is that it is deterministic, how can you say that you yourself have free will/free intellect?
Moreover, just as with the question of free will, so too with determinism you can discuss only on the "operative" plane. If so, you must assume at least operatively that the world is not deterministic (for operatively you assume that human beings have free will). But science too is an operative discussion by the same logic, and all its laws are operative (since they are generalizations from several facts we have seen to a general law). If so, in any event we are speaking only on the operative plane, so what is the dispute about?
In other words, you are only raising a skeptical question, which, as noted, is relevant not only to the question of intelligence and free will but to every claim that I or you might make about the world. So in essence you could have said all your remarks briefly as follows:
Michi, I agree with every word you wrote, except that I have a doubt whether all this (like any other claim anyone makes) is merely our subjective model or whether it describes the world itself.
Now regarding the question of intelligence and determinism:
Indeed, both possibilities can be discussed. Therefore I asked whether, in your view, water has intelligence. Or a sleeping person whose hands move and happen to write Schrödinger's equation and also solve it. If your answer is yes—then you have a very strange (or in your words "not good") definition of intelligence. And if it is no—then you arrive at my definition.
What I claimed is that in a deterministic worldview one cannot offer a consistent definition of the term intelligence that fits the way we conceive of it and use it. You claim that one can, but you do not propose such a definition. In mathematics this is called a non-constructive existence theorem, and as long as it has no proof it has no significance for our discussion.
It is important to note that complexity (entropy is merely a mathematical measure of complexity) cannot define intelligence; it can only serve as a measure of the level of intelligence. It is important to distinguish between the claim "water has low intelligence" and the claim "water has no intelligence." Criteria of entropy and complexity will lead you to the first claim (that is, that water has intelligence, which really does not fit the meaning we are accustomed to giving the term intelligence), whereas I claim the second.
In the end you make it clear here that all you mean is to make a skeptical claim. This is indeed our model, but who says it describes the world? In other words, it is always possible that we are living in a movie (The Matrix). But that can be said about any claim whatsoever, and I do not see this as something really worth arguing about. Our discourse is operative, and on that plane we agree. Everything else is a philosophical discussion about skepticism. I already suggested above how you should have phrased your response to my remarks very briefly ("I completely agree…").
As for your final note about free will, I addressed it extensively in my book Sciences of Freedom. Here I will say briefly that conflating randomness with free will is a conceptual mistake. Moreover, as I explained there, genuinely random processes (not like a die roll but like quantum processes) are not familiar to us at all from our experience and our world (except in quantum theory—that is, on very small scales, much smaller than a neuron, and certainly not from our lives, and even that is not really agreed upon among interpreters of that strange theory. See hidden variables, etc.). By contrast, free will is known to us with utterly intimate and solid acquaintance from within ourselves (as you yourself agreed at the outset). So it is really not clear to me what the logic is in assuming that what is known to us in the most intimate way is a fictional movie, and preferring an interpretation according to which what we see is actually a mechanism (insofar as randomness is a mechanism) unfamiliar to us in our experience altogether.
By the way, a die roll contains nothing random, of course. Newton's laws through and through. It is indeed complex and chaotic (because of the initial conditions), but not random. I also explained this distinction in my aforementioned book. This is not merely a pedantic remark; it is connected to what I wrote here. The die roll is indeed familiar to us, but it is not random. True randomness we have never seen in our experience.
I do not have a precise definition of intelligence, and even if I did, it would only be my definition (or way of looking at it, or conception—it does not really matter). Certainly there is no agreed-upon and closed definition of it. I think a vague, non-sharp definition is preferable to your sharp definition, which in my view simply does not fit how we conceive of intelligence.
I am not sure I would require consciousness for intelligence, but since this is a definition, it is not clear to me how one can rule it out (from your perspective, on my view). Again, I am not sure, but it may be that I would not define AI, for example, as truly intelligent without self-consciousness.
Even in a definition with non-binary values like mine there can be points of discontinuity, or at least jumps, even if it is continuous. There can also be actual values of 0, and values of epsilon that are practically zero. To water I would give an actual 0, and to an ant probably a small value (assuming I do not require consciousness).
As for the baby: again, since this is a definition, you can argue otherwise (and indeed you escaped the either-or), but if we are talking about what the reasonable person thinks, then in my view a baby is certainly intelligent (quite apart from potential, which I already said earlier I brought up as an example of something irrelevant). He meets every criterion I set up (and in my opinion also what the reasonable person would set up). He learns a variety of things, and he is also self-aware (in a limited way). I am not sure the criterion of "what the reasonable person thinks" is a good one for a proof of the type "if you claim X, it follows that what the reasonable person thinks is emptied of meaning," but if we are already looking at it as a criterion, then I do not think there are many reasonable people who think that a baby is not intelligent.
I of course do not agree with your conclusion. I was once surprised by how two people, both with healthy logic and not stupid, can completely disagree on such issues.
As you often write, this is not just a matter of logical inference and tautology, but of many assumptions and probabilities and agreements.
To some extent these exchanges are a bit superfluous, though sometimes enjoyable. I do not think there is any non-stupid person who would disagree with you after the initial reading, but then after "clarifications" would suddenly understand and agree.
Maybe a teenager thinking about these things for the first time.
Hello Shachar.
I do not agree even on the methodological level (which is what you are focusing on now). Sometimes a person grasps a certain concept in one way, and when he is shown the implications he understands that his conception is mistaken (and then he changes his conception), or at least that his understanding of his own intuitive conception is mistaken (and then he realizes that this was not his conception, but that all along it had been different. The problem was in how he defined his conception).
That is what happened, for example, with emotional intelligence and multiple intelligences. They claim to have shown (in my opinion this is not true) that although if you had once asked some person whether the carpenter next to him is intelligent he would have said no, and regarding Einstein he would have said yes, now after renewed examination he will say yes to both. And this has not necessarily changed his conception; rather, he has become aware of what he had intuitively understood even earlier, and he needed help in conceptualizing it and bringing it to consciousness.
I think my arguments here are trying to do something similar for the concept of intelligence. I claim that although at first glance people think a baby is intelligent, consistent follow-through should lead them to the view that water is also intelligent and that a person moving in his sleep can also be intelligent. I also reject your and others' reliance on consciousness, since consciousness is like watching a basketball game and saying that I myself am playing basketball. Why should there be any connection between consciousness and intelligence? Consciousness is part of the requirements for deliberation (unconscious deliberation is of no importance to our discussion), but it is not the relevant parameter. Just as even if all intelligent creatures have legs, it does not follow that having legs is a relevant parameter for defining intelligence.
Through these arguments I am trying to direct determinists to reexamine their position. Such an examination may reveal to them that they themselves did not understand the meaning of intelligence according to their own conception. They understand that they attached it to consciousness, but that is not it. Others will realize that they attached it only to the ability to solve problems, but that is not it either.
To defend oneself against my arguments, it is not enough to propose another definition, and certainly not a non-constructive declaration that there is another definition without presenting it, unless one proposes a definition that also fits the way people think. In my opinion, people do not think that water or a sleeping person act intelligently, and now they need to cope with their attitude toward AI.
Digging in to the initial intuition, as you describe people doing, is not a good method for making decisions and for intellectual clarification. Perhaps I am optimistic, but I at least hope that people (including me) are willing to reexamine their positions, and in particular to examine what their real position is (whether it is identical to what they thought it was—see above about that distinction). As stated, such an examination does not necessarily change the position; sometimes it clarifies for you the position you had already believed beforehand.
In short, even if you do not have a definition of your own, I claim that there cannot be one. That is what you have to deal with. In other words, explain how, on your view, a baby is intelligent but water or a sleeping person are not (even the suggestion of seeing consciousness as the distinction does not really help here, since I very much doubt how much consciousness a tiny baby has. But as I said, consciousness is a very problematic proposal as a basis for defining intelligence; cf. basketball).
You can of course deny this and say that water and a sleeping person also act intelligently, but that already seems to me really unreasonable entrenchment. I find it hard to believe that you yourself think so. This is the mirror I tried to hold up to determinists who think they understand the concept of intelligence and use it clearly and consistently. In my opinion, absolutely not. So long as you do not propose an alternative definition that explains your attitude toward water, rocks, a baby, or a sleeping person, in a consistent way that accords with our conceptions of the term intelligence, your declaration seems to me basically like saying: "I have no alternative, but I refuse to change my position because I am conservative. I prefer to remain in contradiction." This is the kind of entrenchment of religious thinkers that always bothered me. If there is a difficulty, they say: fine, I am still right even though I have no explanation and there is a contradiction in my doctrine. It seems strange to me (though not surprising) that I find such a "religious" attitude also among determinists, who also have articles of faith that are not open to critical examination.
And finally, I am actually not surprised by the disagreements you described. Sometimes it really is an issue that is not clear-cut. But when good arguments are raised (and that is the case here, in my assessment), then in most cases we are dealing with stubbornness and entrenchment, not a genuine disagreement.
I am reminded that I once wrote similarly here about philosophy. Many have the sense that there is no point dealing with it because there are disputes that cannot be decided (after all, it is not an empirical field), and therefore the field is stuck. I very much disagree. In my opinion there are almost no disputes in philosophy (I mean philosophy based on arguments, not existentialist nonsense that people today also call philosophy). Either the "disputants" are speaking about different aspects of the subject and have no real dispute, or one of them is simply stubborn. In my opinion, in very few cases is there a substantive disagreement (apart from a few open questions regarding which there are of course no arguments). My impression is that the common sense of most of us is actually quite similar, and therefore in most arguments my feeling is that there is a resolution, but sometimes one side simply insists and refuses to accept it. That can of course also be me (it is clear to me that I am very far from the level of an angel or a person free of biases), but I do not accept skepticism and despair in advance regarding every argument and every attempt at clarification and resolution.
As for methodology, I think it is correct; it is just that, as a matter of fact, usually people think (unlike perhaps someone for whom such thoughts are not usually on his mind and who suddenly encounters an article) that they have already clarified these things for themselves to some degree.
There is often the feeling, as you describe, that "I get the impression that the common sense of most of us is actually quite similar, and therefore in most arguments my feeling is that there is a resolution, but sometimes one side simply insists and refuses to accept it," but I think that almost always both sides feel that the other side is the one that is simply insisting for no reason (as is certainly happening here too )
As I wrote, even without a definition that gives an exact intelligence value to every thing in the universe, I can speak of creatures that are more or less intelligent, and of a certain threshold below which, depending on the breadth and depth of learning and problem solving, as well as consciousness (even if in itself it does not help solve problems), a creature is not intelligent (cf. water). At least regarding the breadth and depth of learning and problem solving, I think this is an excellent, consistent, and intuitive definition, even without determinists, even if it is not precise. (I do not know how much weight I assign to each of the features, and in any case I do not know how to measure exactly the toolbox of every creature).
By the way, you also do not have a measurement for intelligence; you only added a requirement (which in my opinion simply does not exist in the universe) as a preliminary condition.
First, I will say that in my opinion most people involved in AI have not clarified these points for themselves. I say this from experience in speaking with quite a few such people. Therefore the fact that it seems to them that they have a consolidated picture does not mean they really do. What characterizes all those I have spoken with is that they object to what I say without answering the difficulties I present. That is not called a consolidated picture, and it is not called having thought about the matter.
As for the content of your remarks, I keep repeating the distinction and you keep ignoring it.
You certainly can speak about creatures that are more or less intelligent, because that distinction speaks about the level of intelligence. By the way, on that point we completely agree (therefore your claim at the end of the message that I did not present measures of intelligence is unrelated to our discussion. The measures are agreed upon). Our dispute does not concern that plane in any way. I am speaking about threshold conditions, not about how intelligence is quantified. Our dispute is about how to define who is an intelligent creature and who is not. I do not understand why we keep returning again and again to the irrelevant point of quantification.
The definition in terms of problem solving is indeed an excellent definition of intelligence, and that is what I myself wrote in the post. So why keep returning to it again and again? That is not the point in dispute. It is a definition for measuring intelligence, and it is agreed upon. But it does not provide the threshold condition. I also agree that someone who solves problems better is more intelligent. What I claim is that if this solving is not done through deliberation, then it does not express intelligence.
I keep repeating my claim that water does not have low intelligence; it has no intelligence at all. I claim that this is what you cannot explain. That is the point and there is no other, and it is a shame to wander into other regions. If you can, tell me what you say about this, or about a sleeping person and the other examples I brought. [I hope we do not return again to the question of consciousness, which I have already ruled out (= the basketball spectator), unless you explain what is wrong with what I said.]
You claim that I am adding a requirement that does not exist in the universe (free deliberation). That is your assumption, of course, but my argument attacks that very assumption. Therefore you cannot simply repeat it as though there were an argument here against my position. Once you explain to me why water is not intelligent, then you can continue to hold your claim that there is no such thing as free deliberation.
Regarding operative definition and so on, let me clarify my words further. In day-to-day life I look at the world through a model in which I and the people I meet have free will and consciousness. This is a very useful model for conducting oneself in the world, although I suspect it is not correct. That is what I mean by operative.
So why don’t I simply say, "Michi, I agree with every word you wrote except that I doubt whether all this is only our subjective model or whether it describes the world itself"? Because I think that under this consciousness model it is inconsistent to assume that other people have consciousness but a computer—given that one cannot distinguish its behavior from human behavior—does not have consciousness. That is for the reasons I wrote in claims 1 and 2. This inconsistency is something I identify even if I do not believe at all in the correctness of the consciousness model—that is what I meant in section 3.
Consciousness / free will / soul: I will leave the distinction between them for another discussion and from now on I will use only the concept of "consciousness" (or "awareness").
The reason I think other people have consciousness is not because I generalize from myself alone (N=1) to other particulars that look like me. I think someone (or something) has consciousness when he behaves in a way that reflects my conscious narrative—because I can imagine intention in his actions, put myself in his place. When that happens, after a long conversation or extended observation (a subjective test that I conduct, if you like), I am willing to assume that whoever (whatever) stands before me has consciousness. After I have met many people (an experiment with a large N) and everyone I examined passed my private test, I am willing to generalize the result and say that anyone who looks like a human being would probably also pass my subjective test, and from here that he has consciousness.
The Turing test is an objective approximation to my subjective test. If the computer succeeded in fooling me and I thought it was a human being, it follows that it passed my subjective test for consciousness. If the computer succeeded in fooling enough human beings, then it follows that among human beings there is a consensus that this computer possesses consciousness.
Now let us return to the deterministic model and intelligence. "What I claimed is that in a deterministic worldview one cannot offer a consistent definition of the term intelligence that fits the way we conceive of it and use it. You claim that one can, but you do not propose such a definition. In mathematics this is called a non-constructive existence theorem, and as long as it has no proof it has no significance for our discussion."
The opposite. You are making an impossibility claim. The burden of proof is on you. I claim that there can be a definition of intelligence in a deterministic world (and I give general outlines of its character) that will characterize precisely those entities that, in the consciousness model, you would call "intelligent." In your text you did not succeed in proving that there cannot be a definition of intelligence in deterministic terms.
"Do you think water has intelligence? Or a sleeping person whose hands move and happen to write Schrödinger's equation and also solve it. If your answer is yes—then you have a very strange (or in your words 'not good') definition of intelligence. And if it is no—then you arrive at my definition."
So here, I disagree with you—in my opinion it is possible, in a deterministic world, to arrive at a definition that characterizes the flow of information in the examples you gave as "non-intelligent." If you think not, the burden of proof is on you, and not on me to bring you an alternative definition of intelligence in a deterministic world.
It is still not clear to me what operative means. If you merely suspect that it is not correct, that is true of every claim about the world. No claim about the world is certain (including this one itself). If you mean to say that in your opinion it is not correct (though of course even that is not certain), then you are basically saying that you choose to live in a movie. This is not even a useful/helpful movie, but only one you have become accustomed to. In what way is it useful? With all due respect, in my opinion this is an excuse of someone who is unwilling to admit that he in fact believes in free will, even though he does believe in it. By the way, I have met many such people.
As for myself, as a rational person, I do not chant words that in my opinion are untrue, and certainly do not live by them if in my opinion they are untrue. I must add that it is rather hard to speak and discuss with a person for whom what he says is really true only operatively. So am I supposed to argue about whether it is useful rather than whether it is true? And when you say that this truth is operative—is that itself an operative statement, or truth in the ordinary sense? You understand that one cannot speak this way.
I do not argue about the consistency of the doctrine of a person who wants to live in a consistent movie. Let each person choose his own movies. I also do not really understand why you want to convince me to live in your movie.
When you use the term consciousness to express soul or free will, you are making a big mistake, or at least using very confusing language, since it is really not the same thing. And especially since I claim that intelligence depends not on consciousness but on free intellect (not free will), so why drag the matter of consciousness in here at all?!
It is hard for me to respond to all the rest because I cannot understand whether, when you say that you attribute or do not attribute something to someone, or that the Turing test says something, you mean on the operative plane or as a claim that says something. I see no point in arguing about operative truths.
We are not dealing with proofs but with arguments and reasons. I gave good reasons for why, in my opinion, there is no such definition. My reasoning included several questions that you have not yet answered. You claim there is such a definition, without reasoning. Just like that, and then you move on to "operative" talk that I do not know how to relate to except as living in denial (as I explained).
I did not notice in your remarks any definition (or explanation in any sense) that gathers under its wings the entities we call intelligent and excludes those we do not. Again and again you do not answer my simple and focused questions:
1. Is water an intelligent creature, and why?
2. Is a sleeping person who happens to write the solution to a complex equation intelligent?
3. Why is consciousness relevant to intelligence, if it is merely passive reflection (that is, the thinking proceeds mechanically and consciousness only watches it and is aware of it, a kind of epiphenomenon)?
Through these questions I am making a reasoned claim that the determinist has no definition that fits what we all understand as intelligence, and you keep saying that there is one (an operative definition?). Very well, show it (in my language: make a constructive claim).
After all, you yourself say that you adopt libertarian discourse for the sake of usefulness, meaning that you too cannot manage without it. What is wrong with that very way of seeing things?
And by the way, you still have not answered me how, regarding yourself, you said in the opening message that you genuinely see consciousness and free will (the move to the operative plane was made because of the problem of diagnosing others). How does that fit with your deterministic picture of the rest of the world? Does the claim that everyone is different from you require no justification, while only the claim that they are like you requires one? I really do not understand the logic.
I keep answering regarding water. My definition of an intelligent creature is a creature above a certain level of learning and problem solving (I will return in a moment to consciousness, which keeps coming up). I assign a score composed of the toolbox (including learning and abilities). In my view there is a threshold score above which the creature is intelligent and below which it is not. If it is below the threshold, like water, then it is entirely non-intelligent. Above the threshold, it is intelligent to varying degrees. Again, this is just semantics of the word intelligent, in my opinion.
You are simply imposing a different step function on the matter, one that uses a non-physical characteristic (if it is not deterministic, then in the end it is not physical). I still do not see why that is a preferable claim.
As for consciousness: for me it is simply an important component (and perhaps even a necessary one) in the basket of abilities of an intelligent creature. You may think otherwise, but this is not like the spectator at a basketball game. Part of my basketball game is self-consciousness. It may be that consciousness is only a function of cognitive complexity, and if so it is not an independent feature, but in any case when one breaks abilities down into details it is hard to define which features are independent.
A technical note to the blog administrator: I think that on some of the comments one can reply directly, and on some one cannot.
Hello Shachar.
Without noticing it, you are again answering the question of how much intelligence water has, not the question of whether it has any. What you are saying here is that with respect to the question of "how much," it is customary to draw a line above a certain level and call whoever is above it an intelligent creature and whoever is below it not. What I claimed is that there is no continuum here within which you can draw a line. Water is not on the same axis as we are at all, and therefore this is not a quantitative question. It is on the continuum of computational ability, but that is not the only measure of intelligence. You are basically claiming that it is only computational ability, and therefore water lies on the same continuum. That sounds bizarre to me (essentially an excuse to keep your deterministic assumption in force). But apparently we have a dispute that will not be resolved here.
As for consciousness as well, you did not really answer. After all, I asked why the fact that someone is conscious of the calculations he performs changes his definition as intelligent. Does having consciousness improve his computational ability? For some reason consciousness too lies on the continuum for you. Moreover, earlier you defined intelligence by measures of problem-solving ability; now you insert consciousness into it (without any logic in that). It seems that you are playing with the definitions as you please, without commitment to accepted meanings and usages, simply in order to remain in your position.
Note that even the brain researcher who is arguing with me here in parallel agreed, as a starting point, that intelligence is attributed to creatures with free will (in my opinion: free intellect). Except that because of his determinism (which is not fully clear to me, since regarding himself he wrote that there is freedom even not on the operative plane, and I still have not received an answer from him on this), he prefers to see it as an operative definition or requirement, rather than a requirement about the world. His remarks too, in my opinion, express insistence merely in order to remain in his position (the deterministic one?), but he at least agrees with the simple fact that it is meaningless to talk about the intelligence of water (because it has no freedom, at least on the operative plane).
Well, that is at least how it seems to me. I understand that my remarks seem that way to you as well. Let the readers judge.
When I say an operative model, I mean a model I have chosen to use (because it is useful for achieving my goals at a given moment). A physicist can look at a tiger and determine that this creature is entirely imaginary, since everything is atoms, and the fact that we called a certain organization of those atoms "a tiger" is already an invention of our mind. When we look at the world, we choose to use a model more abstract than the atomic physical model because it is more useful, but when we do that we run the risk that the concepts we invent for those organizations of atoms in the "lower" model are not well defined, or may even contradict themselves. Nevertheless, they do the job in most cases, and so we use them.
And yet, if you ask me to define a tiger in terms of atoms, it will be very difficult for me. I am fairly convinced that one can describe the characteristic statistics of the atomic organization of a tiger in a way that will be correlative to our use of the concept "tiger," but if you demand that I provide you with that statistical characterization I will protest vigorously—it is hard work and I do not see the point. If you insist anyway, I will train a computerized algorithm to look at organizations of atoms in the world and determine which of them represent a tiger. I am fairly convinced I will succeed at this; algorithms nowadays are getting pretty good at it, and for me that is enough to determine that it is possible to characterize the concept "tiger" that we use in the "higher" model by means of concepts from the "lower" atomic model, even if I have no simple way to describe that characterization to you except by pointing to the file where the parameters of the algorithm were saved.
Such is the relation between the deterministic model and the consciousness model in relation to ourselves and the world. I do not see anything problematic in the fact that the "lower" model that describes how the world works is a deterministic model, while the "higher" model describes all sorts of statistical phenomena of information processing and gives them names like "consciousness," "intelligence," "free will," or "free intellect." And as stated, it is hard for me to characterize what these phenomena are in the language of the lower model; I personally do not have the tools for that. At most I can build a machine such that, if you did not know in advance that it was a computer, you could swear that it has all the properties mentioned above.
As for your questions: water is not an intelligent creature, and the person who is asleep is an intelligent creature, but his hands, which solve the complex equation (by chance, presumably), are not moving as a result of an intelligent computation. What is an "intelligent computation"? I am willing to start thinking about that with you. Are you willing to toy with the idea? Say, I think that one of the characteristics of an intelligent computation is a computation that is able to predict consistently the future to come by means of a representation that is significantly simpler than the input it receives. Water does not make such a computation, and neither did the sleeping person's hands move because of such a computation.
As for myself, I know that I have consciousness, and I have will; I behave as though it is free (it is not practical for me at the moment to behave otherwise), but I believe that behind the scenes things happen deterministically, and that is fine.
Hello.
There is an inherent weakness in your arguments, and it recurs again and again.
You claim that you adopt an "operative" conception while at the same time denying its truth. If I perceive reality as A, then my default assumption is that reality is indeed A. True, one can raise doubt about whether I perceive it correctly, and then perhaps it is only an operative model or an illusion. But what you are doing is not skepticism; it is adopting the opposite position. You adopt the conception opposite to your intuitive perception, that is, you decide that reality is "not A." But that is not healthy skepticism; it is confidence in the opposite of what you yourself think. I cannot grasp what the justification for that is.
In short, if indeed it is hard for you to relate to reality as if it were deterministic, and therefore you adopt a model of free will, the obvious conclusion is that we indeed have free will. The burden of proof is on whoever claims the opposite. Again, I am not saying he is necessarily mistaken, but I am saying that adopting, for no reason at all, a position that says I am living in a movie is a bizarre and baseless outlook (it itself reflects living in a movie). It is like saying that in fact there is no law of gravity, but it is hard for me to conceive of the world as though there were no such law, so I assume operatively that it exists. If it is hard for me—then that is probably because such a law really does exist. This is a twist that is simply incomprehensible to me. It is like saying: you do indeed sound right to me, but I have decided otherwise anyway. Why?
By the same token, it is also possible that there are actually fairies, and only I am uncomfortable conceiving of the world as though there are fairies in it. Is the claim that there are no fairies, in your view, also only "operative"?
If indeed we have no freedom—please explain why it is difficult for you to construct a deterministic worldview. Is it not simply because it is not correct? With all due respect, it seems to me embarrassing to repeat something so simple again and again, but I do not see in your remarks an answer to this point.
And beyond all this, you did not answer the following question I asked you: you opened your first comment by saying that regarding yourself you do adopt a conception of freedom (a non-deterministic one) for real, and only with regard to the surrounding world do you see it as "operative" (because of the problem of other minds).
So I ask (and have already asked): if regarding yourself this is not an operative conception, why assume that regarding others it is? After all, regarding yourself you have already accepted that there is freedom. In principle, of course, it is possible that you are a solitary creature, but a reason is needed to infer that (especially when you yourself assume that if you and others have consciousness, then a computer that passes the Turing test does too. That is, you are not inclined to distinguish between similar creatures).
To Rabbi Michael Abraham, שלום,
Suppose two people are given a mathematical problem to solve, one intelligent and one less so. The first succeeds in solving it and the second does not. Both of them exerted themselves greatly and used deliberation in order to find the correct way to solve the problem. What is the difference between them? As I understand it, the difference is that the first has data (intellectual or otherwise) that enable him to understand the problem better and/or read the mathematical "situation" more correctly and/or better understand the methods of solution, etc., and thus solve the problem. The second suffers from dyscalculia, and every time he sees an equation his brain fills with meaningless numbers. No matter how much effort he exerts and how much deliberation he uses, he will not be able to solve the problem because he does not have the basis for doing so.
And from here to my claim—this is also the case in other areas. For example, in the interpersonal sphere. Two people face an interpersonal dilemma (for example, a friend invited them to his house and they cannot come, but it is very unpleasant to say "no"). Both of them try very hard and use deliberation in how to get out of the situation safely. One succeeds and the other does not. That is because the first understands the situation better, understands better what the friend is going through, etc., and therefore, after deliberation, he succeeds in finding the best way to solve the problem. The second does not have those skills. Why should we not say that the first has intelligence in this area? (Of course, we are speaking of the fact that the first, generally speaking, in many cases, reaches a better result in a clearly evident way than the second.) How is this different from ordinary intellectual intelligence? The soccer player too—it may be that there are things he does automatically, but there are certainly things he does through deliberation, and not everyone has that ability. After all, Einstein also solved the exercise 1+2 automatically (unlike a first-grade child who tries to find the best way to solve it—fingers), and it may be that he also solved equations with one unknown automatically without using deliberation.
In short, I claim that intelligence rests on various abilities a person has, abilities that make it easier for him to find the best way to solve problems with the help of deliberation. Deliberation alone is not enough; he needs tools to work with. And such abilities exist in many fields—social, emotional, etc. This does not necessarily mean that every "intelligence" that has been defined is indeed an intelligence.
Best regards,
Moshe
There are two different claims in your remarks: 1. That ability is also part of the measures of intelligence (and not only deliberation). I completely agree. I wrote that deliberation is a necessary but not sufficient condition. 2. That if deliberation is exercised, then every ability expresses intelligence. Here I tend not to agree. But that requires entering into definitions of different abilities and distinguishing them from one another, and of course there is no decisive argument one way or the other. But of course all this is only if deliberation is exercised and the solution is not produced mechanically and without consciousness.
A. Does that mean prophecy is not intelligence? ("Who creates the fruit of the lips")
B. How would you account for the Rambam's statement that the condition for prophecy is that one be a wise man?
I do not know what prophecy is, and I also doubt how much the Rambam knew. If it is passive reception, then it seems there is no intelligence in it. If the prophet is active (and that seems to follow from the fact that no two prophets prophesy in the same style), then perhaps there is.
I would like to sharpen one point raised by Moshe (above) in his remarks (and apologies for the length).
In his article, Michi distinguishes between a creature that solves a problem automatically (including one that is conscious of doing so) and one that needs to decide to solve it, and without such a decision and desire, it would remain unsolved.
Moshe defined the difference between the wise man and the idiot (fool) as follows: "The difference is that the first has data (intellectual or otherwise) that enable him to understand the problem better and/or read the mathematical 'situation' more correctly and/or better understand the ways of solving it, etc., and thus solve the problem. The second suffers from dyscalculia, and every time he sees an equation his brain fills with meaningless numbers. No matter how much effort he exerts and how much deliberation he uses, he will not be able to solve the problem because he does not have the basis for doing so."
That is, the fool, even when he wants to solve a problem, does not succeed, because he lacks the tools and means required for it. That is to say, he has something the computer does not have (the decision), while the computer has something he does not have (the skill).
It turns out that the intelligent one is the one who has both the skill and the will. And as stated in the post, the wise man's skill does not operate automatically. Only if its possessor decides to use it can it solve a problem before it. In this it differs from the computer's skill, which, as stated, operates mechanically.
And here I ask: what distinguishes between them? What is the additional component (or the different one, or the missing one) that causes skill to require a decision, rather than operating automatically?
In my opinion, the answer to this question is that the test by which the mechanical/deterministic creature and the human being are distinguished from one another is the property of "not knowing." That is, for a mechanical creature, whose actions are automatic, there is no "gap" of not-knowing between the "encounter" with the problem and the process of finding the solution. For it, the transition from the former to the latter is equal to the transition between two of its automatic actions. There is no interruption between them that involves hesitation or difficulty.
The absence of such a gap indicates that there is no transition from source to source, that is, from the source of the problem, which is external, to the source of thought. For the computer's thinking too is not driven by an inner source; rather it is an action whose source/cause is outside the computer, only carried out in it (or through it).
Not so with respect to the human being, whose thought arises from his inwardness. For him, encountering the problem means transferring that problem from an external source into the person's inwardness. This transition requires adaptation and identification of the inner self with the information coming from outside. This is a kind of translation, translating the outside into inner/subjective/emotional concepts and recognitions. [This translation differs in essence from translation into the computer's binary language, in that both languages (spoken and binary) are external to the computer (neither contains any inner identification on the part of the computer)]. This transition is what allows for not-knowing and uncertainty in the human being, when it does not succeed, for whatever reason.
From now on, it seems to me that the principal difference between the human being and the computer is not the decision in choosing a solution and its process, but the power of identification (and in Moshe's language: "the difference is that the first has data (intellectual or otherwise) that enable him to understand the problem better"). It is quite possible (at least in certain situations) that a human being will have a skill similar to that of the computer. He can sometimes find a solution to a problem without deciding to do so, because his mechanism worked on its own (automatically). But this will never happen until he understands the problem (that is, identifies with it, until he finds it within himself like an inner contradiction between two desires).
And by this, it is mainly the wise man who will be distinguished from the fool. Not by richness of tools and means, but by the power of identification.
I did not understand the gist of the claim. Translating the problem is an initial stage. After that there is thinking about it, choosing a method, and arriving at a solution (if at all). The stage of translation דווקא does not seem essential to me as compared with a computer, whereas the next stage, in which one chooses the path to the solution, is the important one. As a result, in principle a human being can solve problems of types he has not encountered, and essentially invent a method of solution, something a computer cannot do.
Hi Rabbi,
If I understand you correctly, then you are saying that we are all born with the same amount of "deliberation"; the question that makes the difference is how much "brain" I have. If I have a lot, then I am intelligent, and if only a little, then I am a fool. This creates an interesting situation: the difference between a computer and a human being is the difference in deliberation, and essentially—in his freedom to choose. Sometimes deliberation leads to situations of problem solving that a computer could not do because it is programmed, but sometimes it is really worthless because we have a sufficiently smart brain that automatically understands how to solve the problem. In other words, it is hard for me to assess the value of "deliberation" in the definition of intelligence that you chose. It sounds as though "deliberation" belongs more to the world of free choice than to the world of wisdom, since most (?) of the time when I encounter something that requires thought, I do not think about how I can solve the situation; I simply act automatically and let my logic do its thing… So to what extent do you really think that "deliberation" is a correct component in the definition of "intelligence"? It sounds to me more like it belongs to the definition of "human being"…
I do not know how you got all this from what I said. I said that intelligence requires two conditions: 1. Computational brain capacity, as scientifically defined. 2. Deliberation (intellectual choice). On point 1 I completely agree with the conventional view, except that I add point 2 as a threshold requirement. Intelligence can be assessed only with respect to beings who possess deliberation. That is all.
Linking to a discussion in the responsa section that may help in understanding this post:
https://mikyab.net/%D7%A9%D7%95%D7%AA/%d7%9c%d7%91%d7%97%d7%95%d7%a8-%d7%9c%d7%94%d7%90%d7%9e%d7%99%d7%9f/#answer-63038&comment=23851
We have never encountered true randomness in our experience – really? There is an electronic device, a diode based on the phenomenon of quantum tunneling, a phenomenon in which a particle's wave function shows that it has some probability of randomly jumping over an apparently impassable energy barrier. In highly secure computers this is exploited to generate truly random numbers for encryption purposes, which cannot be cracked by reverse-engineering the algorithms that generated the (pseudo-)random numbers from which the encryption keys were built. (That does not mean the cipher cannot be broken in other ways…) So every time you connect to a properly secured site (personal banking …), you encounter quantum randomness in action before your eyes.
And regarding a die roll: if the system is sufficiently chaotic, we will reach the point where predicting the system's behavior would apparently require measuring the initial conditions with greater precision than the quantum uncertainty principle allows…
I do not think this is the place to clarify scientific issues. Of course there is quantum tunneling, except that no one has seen it. It is a theory. We are talking about interpretations of disputed phenomena (hidden variables, etc.). When you speak about what we know from our experience, it is not correct to speak of tunneling.
In the macroscopic world there is no quantum randomness except in very carefully designed situations, and not spontaneously (except perhaps in a fluid and in a conductor, and even that at very low temperatures).
Good evening,
You wrote in Post 35 – the conclusion is that a deterministic being cannot be considered intelligent.
Professor Sompolinsky claims that brain research shows that man is indeed a deterministic being.
At least according to Wikipedia.
A friend sent me the link to his lecture – but I have not yet had time to look at it.
https://www.youtube.com/watch?v=UPAN-X6BNIo
Perhaps you will find the lecture interesting (it is not short).
Good evening
Moshe
P.S. – It may be that the professor would change his position if he knew that he was thereby condemning himself to non-intelligence.
Sompolinsky was my teacher (formerly) at Bar-Ilan. I took a course with him for research students on neuroscience and determinism, and בעקבותיו I wrote my book 'Sciences of Freedom.' I also heard this lecture, and the arguments we conducted and still conduct from time to time remain as they were.
He is not condemning himself to lack of intelligence because he is mistaken. He has free choice and uses it (among other things, in order to arrive at the conclusion that there is no free choice), and he is indeed a person of very, very high intelligence. But even such people can be mistaken.
Thank you –
I wondered to myself whether, when he received the invitation to lecture, he thought to himself: I must accept—I have no choice.
To say to yourself "I must accept" assumes that there is a discussion at all. But if you are a determinist, then it is not likely that such a discussion takes place within you at all. You are simply supposed to do what is dictated to you and not reflect on it. Unless, of course, the discussion itself is also dictated to you.
Listen, you’re amazing—you completely transformed my whole worldview.
I seem to remember you saying in the last lesson that there is such a thing as coercion regarding opinions. But according to this post, a person decides on his views through free deliberation. If so, how can the concept of coercion apply to a process that is completely free?