The Turing Test: Affirming the Consequent?

[polytropos]

1a The “theological objection” is only one of nine objections which Turing considers.

But it’s the only one you felt necessary to point out.

1b His test isn’t intended to “deliver actually intelligent machines”, it’s intended to decide whether a machine can think without us getting bogged down trying to define “thinking” and “machine”.

Agreed. But then it follows that our judging that the machine can think is limited to what we can adduce about external actions. But we think (I think that is an uncontroversial claim), so we know that thinking is not limited to external actions; so we know that whatever sense we can call a machine “thinking” by the Turing test is limited and analogous.

1c Surely behavior is the only objective measure of intelligence (IQ etc.)?

Depends what you mean by objective. When I think about how to solve a theorem in my head, it seems that I am using my intelligence, even though an external observer can’t tell.

My calculator can compute sums and products much more quickly than I can, but it is doubtful that we should call it “intelligent,” for it could just be more effectively and quickly implementing the algorithm that an abacus uses, and an abacus doesn’t think. It behaves rather well, but it seems that any criteria designating it as intelligent lead to absurd result.

But then, when the (name removed by moderator)uts are spoken human sentences, which the machine parses in order to formulate a response, one must ask where the intelligence suddenly comes in.

3 A dualist might argue that only God can ensoul, but why would a theist argue that only God can make a machine intelligent?

The soul (at least in the Catholic tradition) is taken to be what accounts for intellectual capacities that cannot in principle be accounted for physically. So “being intelligent” and “being ensouled” are the same thing. But this is far afield from this topic, which isn’t about theology or hylemophic arguments for the soul.

4, 5 Turing is only dealing with dualists at that point.

Erm, yes, I’m just pointing out why his comments on dualists seem kind of out of place. He seems to be saying, “Well, even if dualism were true, God could still ensoul machines.” But I doubt he felt that dualism was true… and his point is theologically implausible, so…

6 I think you need to be careful here about the ethics. In what sense are machines not the same kind as us? They are physical, they are made of the same stuff, so at what level of detail do you draw the distinction of kind? There have been those who believed women are not the same kind as men, or black skins not the same kind as white. If space aliens land and have Star Trek matter transporters, mathematics way beyond ours, stunning literature and so on, doesn’t that indicate they’re intelligent, not whether they’re silicon-based seven-legged cyborgs?

You are comparing my distinction between humans and computers to others’ distinction between races and genders. If you think the difference between humans and computers is that nebulous, then I’m not sure what I can say to you.

Regarding aliens, we would need evidence that they were intelligent. We would still regard the intelligence as non-physical (following other arguments that I haven’t given here).

 

[polytropos]

We say that the “human like responses make the assumption that the machine thinks a likely one”.

But the fact is that they don’t, if intentionality and qualia are necessary conditions for a being to be “thinking,” because the Turing test ignores those.

So you say. Do all people always think “intentionally” and “qualitatively”? Or are there exceptions?

No, it does not have to be always for my statement to hold. The claim is that machines don’t think intentionally or qualitatively at all, while humans do some of the time.

A metaphysical statement is vacuous if there is no accompanying epistemology. Why is there no “well-defined” term for thinking? And intelligence?

Intentionality and qualia are not vacuous or lacking in epistemology. Intentionality is just the “aboutness” of thoughts; qualia are the subjective aspects of experience (color, smell, pain, etc. - the qualities that early modern philosophers defined as non-objective, mind-dependent, etc., before they realized that the materialist project would need to explain mind as well). For example, I can think about the color red. I have never “seen” red alone (though I’ve seen red things). I have red in the abstract. (It even seems that my thinking about red need not even be visual/imaginative, since when I think about red, my vision does not “go red.”)

Can a computer “think about” red? It seems doubtful. Suppose we hooked up a camera to the computer, and it converts a photo stream to binary data. Nothing about the data necessitates red. The program could be implemented differently so that the stream of data rather means blue, or something unrelated to photo processing altogether. Perhaps the computer can show “red” on its screen. Perhaps it is even programmed such that it plays a recording to say, “I see red.” The data that the external color was converted to is data - at what point is it represented to the computer in a tangible way, as an intentional experience about the external world, rather than as a stream of data?

Say we fit the computer into an android’s body, so it is visually indistinguishable from a human. Then we sharpen its audio processing and vocal capabilities. It then convinces someone that it “sees red,” just like it has been programmed. But what have these bells and whistles done to change whether we should regard it capable of seeing red? Nothing, it seems.

Until you can give a definition about what “thinking” means, what is “intelligence”, there is nothing to talk about. If you cannot p(name removed by moderator)oint what the machine is not supposed to be able to do, how could be even start to consider if it can or cannot do it? Can toddlers think? The certainly cannot think in abstractions. Can autistic people think? They cannot think in abstractions either. When an autistic person hears the word “dog”, she will recall all the dogs she ever encountered in her life, and not some “abstract” dog. No wonder that Turing did not want to get “bogged down” in the problem of “thinking”.

Unfortunately I do not understand the first part… and I disagree that providing a full definition is “pointless squabbling”.

The point of not giving a definition is that it is not my views that are under scrutiny. Philosophy does not get off the ground when philosophers bicker over definitions. Anyone can gerrymander a solution if they get to define everything. Laying out necessary/sufficient conditions for a coherent definition, and then making arguments to the effect that X definition leads to Y inconsistency, is a better way to go about resolving disputed definitions. Saying, “that definition doesn’t make sense!” or “I don’t accept that definition!” gets philosophy nowhere.

Oh, the infamous “kind”. Would be nice to see a coherent definition of this elusive term, too. And “ensoulment”? The church is reluctant (read: unable) to tell when the “ensoulment” is supposed to take place… By the way, would a fully biological android be considered a different “kind”, too?

Well, if you read that article about essences that I posted after you claimed that term was undefined, then perhaps you are aware of the hylemorphic view on that matter (kind is a more general term used by analytic philosophers). We aren’t really trying to get into such metaphysics here, but it does not seem to controversial to say that humans and computers are of separate kinds. (The contrary, to say that they are obviously the same kinds, seems implausible and question begging to boot.)

 

[tonyrey]

Turing lacked insight into a machine’s lack of insight. Did he really regard himself as a machine, I wonder…

 

[Bagheera]

First, I would like to see your response to the “fallacy” problem. Since we do not assert that the consequent proves the antecedent, where is the “fallacy” you speak of? After all that was the point of the whole thread.

polytropos:

But the fact is that they don’t …

Here you go again… talking about “facts”! You are the one who introduced the concepts of “intentionality” and “qualia” (in this thread). These terms need to be defined and substantiated as well. Moreover, you need to show how these properties manifest themselves in humans and how do we know that every human has them? Also how do we know that the machines do not have them?

You cannot escape the epistemology question, because you assert “knowledge” about the existence of these properties. To say that you exercised introspection and you “discovered” them in yourself is insufficient. And then you wish to extrapolate to all humans, saying that we are all the same “kind”. That will not work.

By the way, you are in error, if you think of androids as artificially grown human bodies, with electronic computers “implanted” as brains. Androids are (fictional) fully artificial, biological beings, who were grown or manufactured in a vat. Like using gene-splicing and micro-manipulation to create an artificial zygote, and grow it either in an artificial womb or in the womb of a surrogate mother. There is no epistemological () method to distinguish them from a “real” human. You may say that they lack a “soul”, but then you need to define what a soul is, and show that we have it, while they don’t. How do you decide if this android can “think” or not?

…if intentionality and qualia are necessary conditions for a being to be “thinking,” because the Turing test ignores those.

And here you use the all-important “if”. You need to substantiate that “if” – after you give a proper epistemological method of how to detect their existence or the absence of them.

polytropos:

The point of not giving a definition is that it is not my views that are under scrutiny.

But they are, since asserted that machines “cannot think”. You also asserted the existence of a fallacy, due to your misconception what the successful Turing test would be establishing. Now hopefully this misconception is out of the way, so we can concentrate on the rest, which is a very interesting question indeed.

 

[tonyrey]

Turing lacked insight into a machine’s lack of insight. Did he really regard himself as a machine, I wonder…

The acid test of any hypothesis is how we live. Anyone who behaves like a machine is regarded as a lunatic!

 

[inocente]

Agreed. But then it follows that our judging that the machine can think is limited to what we can adduce about external actions. But we think (I think that is an uncontroversial claim), so we know that thinking is not limited to external actions; so we know that whatever sense we can call a machine “thinking” by the Turing test is limited and analogous.

Sounds like you’re repeating Turing’s theological objection in a circular form: your axiom appears to be humans think, non-human’s can’t, and you’re saying we know that because humans think, non-human’s can’t.

You (and Turing) also assume we wouldn’t be able to read off a machine’s thoughts, but from an engineering angle it would be important to take diagnostics, just as with today’s computers.

Suppose we could log a machine’s thoughts in exquisite detail, would that convince you?

Depends what you mean by objective. When I think about how to solve a theorem in my head, it seems that I am using my intelligence, even though an external observer can’t tell.

You appear to be saying there’s no way to detect intelligence except by introspection. Doesn’t that make the concept of measuring intelligence super-duper-subjective?

*My calculator can compute sums and products much more quickly than I can, but it is doubtful that we should call it “intelligent,” for it could just be more effectively and quickly implementing the algorithm that an abacus uses, and an abacus doesn’t think. It behaves rather well, but it seems that any criteria designating it as intelligent lead to absurd result.

But then, when the (name removed by moderator)uts are spoken human sentences, which the machine parses in order to formulate a response, one must ask where the intelligence suddenly comes in.*

From complexity, where else?

The soul (at least in the Catholic tradition) is taken to be what accounts for intellectual capacities that cannot in principle be accounted for physically. So “being intelligent” and “being ensouled” are the same thing. But this is far afield from this topic, which isn’t about theology or hylemophic arguments for the soul.

Don’t see how that answers my question. Theism is belief in a deity, it says nothing about souls, and certainly nothing about whether only the deity can grant the capability of thought.

Erm, yes, I’m just pointing out why his comments on dualists seem kind of out of place. He seems to be saying, “Well, even if dualism were true, God could still ensoul machines.” But I doubt he felt that dualism was true… and his point is theologically implausible, so…

Theologically implausible to whom? Another theologian might argue that our true purpose is to build machines to the glory of God, and a dualist can’t argue that carbon is more in the image of God than silicon. Seems Turing stitched up the dualist pretty darn good.

You are comparing my distinction between humans and computers to others’ distinction between races and genders.

Yes, that was my point - if it is bad to discriminate according to appearances (gender, skin color, etc) then how is your distinction different?

Regarding aliens, we would need evidence that they were intelligent. We would still regard the intelligence as non-physical (following other arguments that I haven’t given here).

The scenario I gave was the aliens have Star Trek matter transporters, mathematics way beyond ours, stunning literature and so on. Doesn’t that indicate they’re intelligent? What more evidence would you need? Do all humans also have to get a pass grade in your test?

 

[polytropos]

First, I would like to see your response to the “fallacy” problem. Since we do not assert that the consequent proves the antecedent, where is the “fallacy” you speak of? After all that was the point of the whole thread.

I provided an argument as to sorts of things we can do that a machine in principle cannot. It was in the chunk of my last post that you did not respond to. I am saying that it is a fallacy to suppose that a machine, if it could act like us, has those properties (or “most likely” has those properties), simply because behavior is not dependent oon those qualities.

Here you go again… talking about “facts”! You are the one who introduced the concepts of “intentionality” and “qualia” (in this thread). These terms need to be defined and substantiated as well. Moreover, you need to show how these properties manifest themselves in humans and how do we know that every human has them? Also how do we know that the machines do not have them?

Qualia and intentionality are well-defined and established terms in philosophy of mind, ie. professional philosophers like Daniel Dennett or the Churchlands who choose to do away with them realize that the burden is on them to provide argument against them. Qualia generally refer to the qualitative aspect of sensations (as I said, pain, color, taste, etc.). How our nervous system receives painful stimuli is not confusing. How that leads to a subjective experience is so far unexplained by neuroscience and materialist philosophy of mind. Intentionality is the “aboutness” of thoughts, ie. the way that I am aware that the information I entertain is about things in the world (perhaps abstracted), and is not “raw data” (cf. my discussion of thinking about the color red in the abstract, and why it is doubtful that a computer could do so analogously).

How do these properties represent themselves? If you don’t have them, then you are a philosophical zombie.

I only know that I have qualia and intentionality, so I can’t peek into someone else’s head and see them; they are subjective experiences. I suppose that other things of my kind (human) have them as well, not because they act like they do, but because they are of the same kind as me.

I provided an argument based on qualia and intentionality (not responded to) as to why there does not seem to be any reason for machines to have them.

You cannot escape the epistemology question, because you assert “knowledge” about the existence of these properties. To say that you exercised introspection and you “discovered” them in yourself is insufficient. And then you wish to extrapolate to all humans, saying that we are all the same “kind”. That will not work.

As I said, if you deny that you have qualia and intentionality, then you are a philosophical zombie by definition. Qualia and intentionality are a part of my first-person experience. It is not necessary to my argument that we extrapolate to other humans; it is sufficient for me (or for anyone honestly following the argument) to admit that I have qualia and intentionality, which do not in principle submit to a material explanation.

Honestly, even the center of consciousness is not materially explained yet (and I am open to certain emergent theories of consciousness and qualia, though it seems that their resolution will require a metaphysical overhaul). You can’t verify that another person is conscious anymore than I can verify that another person has qualia and intentionality; yet you (presumably) are conscious.

By the way, you are in error, if you think of androids as artificially grown human bodies, with electronic computers “implanted” as brains. Androids are (fictional) fully artificial, biological beings, who were grown or manufactured in a vat. Like using gene-splicing and micro-manipulation to create an artificial zygote, and grow it either in an artificial womb or in the womb of a surrogate mother. There is no epistemological () method to distinguish them from a “real” human. You may say that they lack a “soul”, but then you need to define what a soul is, and show that we have it, while they don’t. How do you decide if this android can “think” or not?

Thank you for the clarification (although I was obviously using the term loosely, and I don’t seem to be too far off the mark). But you did not respond to what I actually said about my android. If I am programming a machine to recognize the wavelengths of light that constitute red, there is not necessity for (or reason to believe that) the machine has a subjective experience of red like you and I do. If I add software to it so that it handles (name removed by moderator)uts and outputs in a convincingly human-like way, and then couch it in a human-like synthetic body, there is no difference in what the machine is doing. There is still no reason to suppose that it is experiencing red (since that is extraneous to what I designed the machine to do).

Furthermore, even if there were a material explanation for qualitative and intentional states (and that is a big if), I would not need to account for it in programming my red-recognizer. So it could behave as though it sees red when it in fact doesn’t, even if there were a material explanation for qualia. So the inference (whether formal or just empirical) from human-like behavior to thinking is simply unfounded.

 

[polytropos]

And here you use the all-important “if”. You need to substantiate that “if” – after you give a proper epistemological method of how to detect their existence or the absence of them.

Well, I have now given arguments for intentionality and qualia. If you are not a philosophical zombie, then you have intentionality and qualia. They are not materially detectable. But what you’re missing is that that is not the problem for my theory but for your theory. The facts that there is not a material epistemological method to detect them, that there is not consistent materialistic theory that accounts for them, that they are a part of the first-person experience of every non-zombie, all count against theories that fail to make reference to them.

But they are, since asserted that machines “cannot think”. You also asserted the existence of a fallacy, due to your misconception what the successful Turing test would be establishing. Now hopefully this misconception is out of the way, so we can concentrate on the rest, which is a very interesting question indeed.

I have been clear about what a “successful Turing test” establishes. It establishes that a machine has been created which acts like a human. We can’t extrapolate further from the data.

 

[polytropos]

Sounds like you’re repeating Turing’s theological objection in a circular form: your axiom appears to be humans think, non-human’s can’t, and you’re saying we know that because humans think, non-human’s can’t.

Did you read his theological objection? Humans think, non-human’s can’t is not an axiom I have stated. I’ve pointed out that humans think and the Turing test doesn’t show that non-humans can. The test just doesn’t show what some might claim it shows.

Suppose we could log a machine’s thoughts in exquisite detail, would that convince you?

Well, yes, but I imagine there is some unclearness here as to what would constitute “a machine’s thoughts.” If I go and write a program this afternoon, I could tell it to print off each line it performs. But those aren’t “thoughts,” they’re just outputted statements. (Other issues would crop up - thoughts do not need to have any linguistic or imagistic basis, so it’s doubtful whether some of our thoughts even could be “logged” and represented in a material way. Your question seems to beg the question that “logging something’s thoughts” is even a coherent notion.)

You appear to be saying there’s no way to detect intelligence except by introspection. Doesn’t that make the concept of measuring intelligence super-duper-subjective?

You can try to measure a subject’s intelligence by external measures, because that’s all you can do. That is just an issue of epistemology; the fact that we are constrained to measure intelligence externally doesn’t show that intelligence is external. We measure it externally because we have to, but I feel like most folks wouldn’t deny that thinking is subjective. You wouldn’t test your calculator’s intelligence, because we don’t think it’s intelligent to begin with, even though it calculates sums more quickly than I can.

From complexity, where else?

Oh, right, complexity. The discipline of philosophy of mind is now resolved!

Don’t see how that answers my question. Theism is belief in a deity, it says nothing about souls, and certainly nothing about whether only the deity can grant the capability of thought.

This is not really relevant to the topic, which isn’t about souls. But it is believed that the soul is immaterial owing to its intellectual properties, and cannot be created by anything material.

Theologically implausible to whom? Another theologian might argue that our true purpose is to build machines to the glory of God, and a dualist can’t argue that carbon is more in the image of God than silicon. Seems Turing stitched up the dualist pretty darn good.

And such a theologian is free to do so. He could even be you, if you’d like to go perform some exegesis to find that our true purpose is to build intelligent machines, and Christians of all denominations have just missed the point for 2000 years. (Hence, “theologically implausible.”)

I don’t see how Turing “stitched up the dualist pretty darn good.” Dogs have complex brains. Dualists (with some exceptions) don’t hold that God ensouls them. You and Turing seem to be saying that theists are committed to thinking that if something exists then God could and would ensoul it.

Yes, that was my point - if it is bad to discriminate according to appearances (gender, skin color, etc) then how is your distinction different?

Because other genders and other races are vastly less dissimilar from us than computers, perhaps?

The scenario I gave was the aliens have Star Trek matter transporters, mathematics way beyond ours, stunning literature and so on. Doesn’t that indicate they’re intelligent? What more evidence would you need? Do all humans also have to get a pass grade in your test?

It indicates that they are likely intelligent. Probably sufficiently so to treat them as such. If we found later that they were machines from which intelligence could not in principle emerge, then we would have been wrong; nothing I’m saying claims that our judgments about intelligence are indefeasible.

 

[Bagheera]

polytropos:

I am saying that it is a fallacy to suppose that a machine, if it could act like us, has those properties (or “most likely” has those properties), simply because behavior is not dependent on those qualities.

What kind of “fallacy” are you talking about? It cannot be the “affirming the consequent”.

This is what you asserted:
polytropos in the OP:

The fallacy crops up when one tries to use the Turing test to determine that machine’s behaving like us implies that the machine thinks like us. We have an argument of the form:

(P1) If X thinks, then X can behave as though it thinks.
(P2) X behaves as though it thinks.
(C) X thinks.

(P1) seems reasonable given our experience. We think, and our behavior determines what it would mean for something to behave as though it thinks. Likewise, we can grant (P2), that a machine capable of behaving as though it thinks is conceivable. But the issue is that this line of reasoning is the straightforward fallacy of affirming the consequent.

And I already answered that we NOT assert this, so there is no fallacy involved. So can we dispense with the “fallacy” and concentrate on the rest?

 

[polytropos]

What kind of “fallacy” are you talking about? It cannot be the “affirming the consequent”.

You said:

We say that the “human like responses make the assumption that the machine thinks a likely one”.

This still seems to affirm the consequent. You are saying that human-like responses warrant the assumption that the machine thinks. So the machine’s thinking is something internal and undetectable, for which you are taking human-like responses to be an indicator. (If this is not what you mean where I’ve quoted you, please clarify.) If a machine has human-like responses, you are saying that the responses are “likely” consequent of its thinking.

So yes, you are reasoning from consequent to antecedent, but as I have argued in my other posts, this is not “likely” after all.

So can we dispense with the “fallacy” and concentrate on the rest?

It would probably be good for you to focus on the more substantive arguments.

 

[polytropos]

Recently I heard an interesting story on NPR in their science hour. There is this annual meeting on artificial intelligence and the Turing test, where they have those judges in a blind test who have to decide if they are having a conversation with a human person or a computer. In 2008 or 2009 for the first time 25 % of the judges where fooled by a computer, dangerously close to the threshold of 30 % that Turing had proposed (the discussion on NPR then also was, why not 50 or 51 % ?).

Yet then they told the story of that one computer that had been learning for many years; with the arrival of the internet it then could have conversations with millions of people instead of just a few. During that process it continuously learned, and became better and better, until one day, after that 25 % threshold had been crossed in that competition, one judge found out how to unmask it as just a machine instead of a human person. He said something outrageous and absurd, like: “Last night I did not sleep well because a meteor crashed into my bedroom.” The computer than replied: “I woke up at 1 in the afternoon.”

The computer obviously was thrown off by the absurd and unexpected remark – it lay outside of what it had encountered and learned during those millions of other conversations – and it could not reply like a human being would. Why? The incident exposed that while a computer can learn and learn how to react – after all, a computer is really good at computing – it does not have the mental judgment needed to reply to an absurd remark in an appropriate manner. In other words, it cannot think.

After this judge had learned to ‘crack the code’ and informed his fellow judges, the next year none of the judges were fooled on the competition, not a single one. So much for the Turing test and ‘thinking machines’.

I have been thinking about this. It is rather interesting that imitation is taken to be the standard for a machine’s intelligence. This paradigm seems to lead to some weirdness.

In this case, the machine failed because it was asked a question it had not been “prepared” for. So ostensibly, this weakness needs to be patched up if the machine is to be reprogrammed to pass the Turing test. But of what does the “patching up” consist? It consists of exposing the machine to circumstances in which it will “learn” to handle absurd questions.

But why should we suspect that this incremental correction is making the machine more intelligent? Certainly it would make it more difficult to distinguish the machine’s responses from those of a human. But it still indicates that the machine has no understanding of real semantic content; it “learns” from conversations by learning to match responses to token (name removed by moderator)uts. But the way in which it is corrected indicates that it just handles another case - it does not actually know that what was uttered to it was absurd. It just now has a way to respond to a token absurdity.

The standard of imitation seems dubious altogether. If humans are biological computers, then what is special about human social intelligence? Nothing. But to emulate it is the criteria for an AI. But why? Humans are taken to be intelligent for their ability to solve, say, abstract problems; the ability to convey those problems to other people is not fundamental to someone’s intelligence (just fundamental to knowing that they are intelligent). But then we should be impressed not with machines that pass the Turing test, but with machines that solve problems whether they manage to act like us or not.

Take a computer that can solve a tough math problem. It is already incredibly complex. It is computationally powerful. Then add the “bells and whistles” sufficient to get it to pass the Turing test. Why have the bells and whistles made the difference in its intelligence? What is the difference in the machine knowing how to parse my words and formulate a socially acceptable response, and the machine knowing how to parse mathematical functions and compute the result? For the computer, running through its lines of code, what is the difference? (And why is its computations in each case judged against the backdrop of human society, if humans are just one of any number of possibly intelligent entities?)

 

[Al_Moritz]

But why should we suspect that this incremental correction is making the machine more intelligent? Certainly it would make it more difficult to distinguish the machine’s responses from those of a human. But it still indicates that the machine has no understanding of real semantic content; it “learns” from conversations by learning to match responses to token (name removed by moderator)uts. But the way in which it is corrected indicates that it just handles another case - it does not actually know that what was uttered to it was absurd. It just now has a way to respond to a token absurdity.

(Emphasis aded.)

Yes, indeed. As I pointed out, the episode shows the machine has no mental judgement, which indicates that it cannot understand and think. ‘Correcting’ the machine in this case would not miraculously instill mental judgement, just as its previous learning how to lead conversations had not done so.

 

[Bagheera]

polytropos:

This still seems to affirm the consequent. You are saying that human-like responses warrant the assumption that the machine thinks. So the machine’s thinking is something internal and undetectable, for which you are taking human-like responses to be an indicator. (If this is not what you mean where I’ve quoted you, please clarify.) If a machine has human-like responses, you are saying that the responses are “likely” consequent of its thinking.

So yes, you are reasoning from consequent to antecedent, but as I have argued in my other posts, this is not “likely” after all.

I wish we could communicate with simple line drawing, by creating Venn diagrams. So I will suggest you “imagine” (or draw) the following. Draw a small circle and write “P” into it. Then draw a larger circle around the smaller one (so that the whole smaller one is fully included in the larger one) and write “Q” into it.

The first line of the syllogism of “P => Q” means that whenever “P” is true, then it follows that “Q” is true. In other words, if you place a dot into the “inner circle”, it will also be inside the “larger circle” Therefore the valid syllogism would continue as: since “P” is true, and therefore “Q” is also true.

The incorrect way to reason: “if “Q” is true, then “P” is also true. (That would be the fallacy you speak of.) If we place a dot into the larger circle, then it may or may not be inside the smaller circle. So from “Q” it may or may not follow that “P” is true.

Since we do not assert this, the fallacy does not apply. Period.

Now we can examine the size of the “circles”. It is possible that the inner circle is almost as large as the outer one. Still it is possible to place a dot into the outer circle which does not fall into the inner circle. However, the likelihood of this event decreases as the area outside “P” is getting smaller. That is what the Turing test is all about. The more frequently we get a correct response, the more likely it is that we deal with the “real McCoy”. And there is nothing fallacious about this.

In science we never deal with “absolute”, logical certainties. We present a null-hypothesis, and perform experiments, which will either support, or deny the validity of the null-hypothesis. Just one negative result will refute the hypothesis, but zillions of “positive” results will not “PROVE” it. Those zillions of positive results will “only” substantiate that the hypothesis was correct. Yet, if just ONE negative outcome comes after those zillions of positive ones, it will invalidate the hypothesis regardless of those zillions of positive ones.

**Tell me, where is the “fallacy” here? **

I would love to go one and talk about the “philosophical zombie” and the rest. But not until we can reach an agreement here.

 

[Bagheera]

polytropos:

In this case, the machine failed because it was asked a question it had not been “prepared” for. So ostensibly, this weakness needs to be patched up if the machine is to be reprogrammed to pass the Turing test. But of what does the “patching up” consist? It consists of exposing the machine to circumstances in which it will “learn” to handle absurd questions.

This is **exactly **how humans learn.

(But you can safely disregard this for the time being. I will be happy to go into the details once we have that “fallacy” out the way.)

 

[inocente]

Did you read his theological objection? Humans think, non-human’s can’t is not an axiom I have stated. I’ve pointed out that humans think and the Turing test doesn’t show that non-humans can. The test just doesn’t show what some might claim it shows.

Read back what you said originally - you ruled out any possibility of thinking machines by fiat, by saying there is no way to know it isn’t just analogous to what you reckon is the real deal.

Well, yes, but I imagine there is some unclearness here as to what would constitute “a machine’s thoughts.” If I go and write a program this afternoon, I could tell it to print off each line it performs. But those aren’t “thoughts,” they’re just outputted statements. (Other issues would crop up - thoughts do not need to have any linguistic or imagistic basis, so it’s doubtful whether some of our thoughts even could be “logged” and represented in a material way. Your question seems to beg the question that “logging something’s thoughts” is even a coherent notion.)

Recent technology has overtaken your skepticism. Although reading human and animal minds is beset with practical difficulties, is in its infancy, and currently very low resolution, it’s becoming a reality:

technologyreview.com/news/409705/mind-reading-with-functional-mri/
technologyreview.com/view/513326/scientists-use-mri-to-glimpse-the-dreaming-mind/
articles.latimes.com/2012/may/08/business/la-fi-tn-scientists-give-dogs-mri-20120507

You can try to measure a subject’s intelligence by external measures, because that’s all you can do. That is just an issue of epistemology; the fact that we are constrained to measure intelligence externally doesn’t show that intelligence is external. We measure it externally because we have to, but I feel like most folks wouldn’t deny that thinking is subjective. You wouldn’t test your calculator’s intelligence, because we don’t think it’s intelligent to begin with, even though it calculates sums more quickly than I can.

We can correlate intelligence with physical factors such as number of neurons, we have the IQ measure for humans, and so on. Intelligence may be a somewhat fuzzy notion but it’s far from the mystery some may have claimed in the past.

Oh, right, complexity. The discipline of philosophy of mind is now resolved!

Surely at least one philosopher must have been bright enough to see that answer? If not then they may as well all go home.

This is not really relevant to the topic, which isn’t about souls. But it is believed that the soul is immaterial owing to its intellectual properties, and cannot be created by anything material.

It’s relevant for that very reason.

*And such a theologian is free to do so. He could even be you, if you’d like to go perform some exegesis to find that our true purpose is to build intelligent machines, and Christians of all denominations have just missed the point for 2000 years. (Hence, “theologically implausible.”)

I don’t see how Turing “stitched up the dualist pretty darn good.” Dogs have complex brains. Dualists (with some exceptions) don’t hold that God ensouls them. You and Turing seem to be saying that theists are committed to thinking that if something exists then God could and would ensoul it.*

No, Turing’s argument is that, even if thinking does require an interface to some undetectable immaterial realm, there’s no logical reason why an all powerful God shouldn’t be able to provide it to machines. (Although Turing politely passes over such impenetrable enigmas as why an undetectable realm is needed in the first place, how it ever got discovered if it is undetectable, and how come the physical side of the interface to that realm has never been found).

Because other genders and other races are vastly less dissimilar from us than computers, perhaps?

In a bygone age the name computer referred to a person with the job of doing what digital computers now do, so no, not dissimilar. If there were machines walking around which could think and feel and go to temple, it would be derogatory to call them mere computers.

It indicates that they are likely intelligent. Probably sufficiently so to treat them as such. If we found later that they were machines from which intelligence could not in principle emerge, then we would have been wrong; nothing I’m saying claims that our judgments about intelligence are indefeasible.

That’s Turing’s theological objection! How come you get to tell God that intelligence could not in principle emerge!

 

[inocente]

Yes, indeed. As I pointed out, the episode shows the machine has no mental judgement, which indicates that it cannot understand and think. ‘Correcting’ the machine in this case would not miraculously instill mental judgement, just as its previous learning how to lead conversations had not done so.

I tried to find the competition you mentioned. The only one I found which is run annually is the Loebner Prize:

“Within the field of artificial intelligence, the Loebner Prize is somewhat controversial; the most prominent critic, Marvin Minsky, has called it a publicity stunt that does not help the field along. In addition, the time limit of 5 minutes and the use of untrained and unsophisticated judges has resulted in some wins that may be due to trickery rather than to plausible intelligence.” - en.wikipedia.org/wiki/Loebner_Prize

Minsky is quoted here as saying “I do hope that someone will volunteer to violate this proscription so that Mr. Loebner will indeed revoke his stupid prize, save himself some money, and spare us the horror of this obnoxious and unproductive annual publicity campaign.”

 

[polytropos]

Read back what you said originally - you ruled out any possibility of thinking machines by fiat, by saying there is no way to know it isn’t just analogous to what you reckon is the real deal.

That’s not fiat. It’s a fact that if thinking is an internal subjective activity, then when external activity is all we have to go off of, we can only speak of the machines thinking “analogously.” That is not my fiat; it’s just an in-principle limitation of the Turing test.

Recent technology has overtaken your skepticism. Although reading human and animal minds is beset with practical difficulties, is in its infancy, and currently very low resolution, it’s becoming a reality:

technologyreview.com/news/409705/mind-reading-with-functional-mri/
technologyreview.com/view/513326/scientists-use-mri-to-glimpse-the-dreaming-mind/
articles.latimes.com/2012/may/08/business/la-fi-tn-scientists-give-dogs-mri-20120507

I will take a closer look at these in the next few days (I have a couple exams on wednesday so I can’t take a close look at them now). However, my short answer is that they do not seem to meet what I am claiming. I think some of my remarks in post #22, as to whether a computer can “think” about red, are relevant. Certainly there are particular neurons that fire when I see the color red. Those neurons are going to be roughly the same across my disparate experiences of seeing red, and they should roughly correlate to which neurons fires in other members of my biological species fire when they see red. Brain scans can track that. They can also use it to predict what data is entering my perception.

But I am committed to denying none of that, and none of it shows that the presence of data corresponding regularly to the experience of seeing red is a sufficient explanation of my subjective, qualitative experience of seeing red. And so, such studies have nothing to say as to whether a computer can be a “subject” which can represent the experience of red to itself, even if the appropriate data is flowing through it, as I allude to in post #22.

We can correlate intelligence with physical factors such as number of neurons, we have the IQ measure for humans, and so on. Intelligence may be a somewhat fuzzy notion but it’s far from the mystery some may have claimed in the past.

I do not think most psychologists would hold that intelligence is strongly correlated with number of neurons (although, correct me if I’m wrong, as I’m not an expert), and I know the question of what IQ actually measures is hotly debated.

In any case, I should say that I’m not a substance dualist, and I regard the brain to be necessary for all mental function, even if not sufficient for a few mental functions.

Surely at least one philosopher must have been bright enough to see that answer? If not then they may as well all go home.

Oh, quite a few see it (and few dualists, even, would deny that the brain’s complexity is important). But materialist theories have been cycling in and out of favor constantly since materialists first started attempting them; there is more to it to complexity, but if materialism is true, then no one is quite sure what.

It’s relevant for that very reason.

I am quite willing to discuss the hylemorphic view of the soul, but this topic is meant to be a critique of the Turing test, not a defense of any form of dualism. (And one does not need to be a dualist to find the Turing test or functionalism lacking - cf. Searle, Tallis, Dreyfus, Nagel.) Perhaps at some point in the future, when I’m less busy, I will start some discussions on it.

No, Turing’s argument is that, even if thinking does require an interface to some undetectable immaterial realm, there’s no logical reason why an all powerful God shouldn’t be able to provide it to machines.

Which I think we’re mostly agreed on. (I would make some qualifying arguments, but they are not really relevant to the point at hand.) But I don’t know why anyone would be impressed with that objection. The objection is that physical facts can’t beget intelligence; so the theist would be claiming that the computing power is not sufficient for the intelligence. There is certainly more that both sides would add here, but I don’t know why even Turing would be satisfied by his response. (Actually, he probably didn’t care about formulating a theological objection if he saw theology as implausible, and the theological objection is probably in the paper more for the sake of his atheist readers than for theological ones. If he were interested in really disputing theological claims, he probably would have addressed traditional arguments in philosophy of mind, rather than the layman’s assertion that only humans have souls.)

I’ll also add that I don’t believe in an “undetectable immaterial realm,” since I’m not a substance dualist.

In a bygone age the name computer referred to a person with the job of doing what digital computers now do, so no, not dissimilar.

This is a huge equivocation. In your “bygone age,” the moral value of the “computer” (the person with the job of computing) had moral value by virtue of being a person, not a person who computes.

If there were machines walking around which could think and feel and go to temple, it would be derogatory to call them mere computers.

I agree. But that is because our creating them with a particular circuitboard would not be what accounts for their “thinking” and “feeling.”

 

[polytropos]

That’s Turing’s theological objection! How come you get to tell God that intelligence could not in principle emerge!

I don’t tell God anything. It is what I claim based on what intelligence is and what the physical facts about a computer are.

Suppose a materialist claimed that intelligence could emerge from a single electron. I think we would have to regard his claim as vacuous, because intelligence is not the sort of thing that would emerge from a single electron. Some theists might claim that it must still be possible, but I would disagree with them. That is another topic, but the short reason is that I would regard such impossibilities as falling under the idea that God cannot perform the logically impossible, and what an electron is may just not be the sort of thing that can be intelligent.

But then there are limits in principle on what intelligence can emerge from. The dispute is just over what they are.

A further example: Take Quine’s underdetermination hypothesis. The argument is that the physical facts cannot absolutely determine one scientific theory to the exclusion of other incompossible scientific theories. It is impossible for the physical fact to do so, owing to the sort of things that physical facts are. God’s omnipotence does not entail that He can arrange physical facts in such a way so as to exclude all but one scientific theory; the idea is simply vacuous. No one is detracting from God’s power to say He can’t, anymore than no one is detracting from God’s power to say He can’t make a square circle.

 

[inocente]

That’s not fiat. It’s a fact that if thinking is an internal subjective activity, then when external activity is all we have to go off of, we can only speak of the machines thinking “analogously.” That is not my fiat; it’s just an in-principle limitation of the Turing test.

If you reject physical evidence then I see no way by which to define what you mean by “we” - have you not ruled out DNA and every other measure of what it means to be “we”?

But I am committed to denying none of that, and none of it shows that the presence of data corresponding regularly to the experience of seeing red is a sufficient explanation of my subjective, qualitative experience of seeing red. And so, such studies have nothing to say as to whether a computer can be a “subject” which can represent the experience of red to itself, even if the appropriate data is flowing through it, as I allude to in post #22.

I don’t understand this supposed hard problem of qualia.

Suppose that, a la Terminator, all sense data was reported in the style of a heads-up display. So instead of sensing red, your visual system displays “light at 650 nm”, instead of hunger, a list of required proteins and so on flashes up.

That surely wouldn’t be thought inexplicable, so how does it change anything if the brain instead uses “user friendly” qualia to enable prioritization and avoid reams of unnecessary detail flashing up?

I do not think most psychologists would hold that intelligence is strongly correlated with number of neurons (although, correct me if I’m wrong, as I’m not an expert), and I know the question of what IQ actually measures is hotly debated.

I think average intelligence of a species correlates with the number of neurons (excluding those needed for “body management”), and individual intelligence in humans correlates with number of synapses.

In any case, I should say that I’m not a substance dualist, and I regard the brain to be necessary for all mental function, even if not sufficient for a few mental functions.

Apologies then, I made an unwarranted assumption, you’re the first Catholic I’ve met here to say that.

Oh, quite a few see it (and few dualists, even, would deny that the brain’s complexity is important). But materialist theories have been cycling in and out of favor constantly since materialists first started attempting them; there is more to it to complexity, but if materialism is true, then no one is quite sure what.

Science will sort it. Philosophers do very poorly at this kind of problem: they have no way to break it down, no way to check progress, and they compete rather than cooperate.

I am quite willing to discuss the hylemorphic view of the soul, but this topic is meant to be a critique of the Turing test, not a defense of any form of dualism. (And one does not need to be a dualist to find the Turing test or functionalism lacking - cf. Searle, Tallis, Dreyfus, Nagel.) Perhaps at some point in the future, when I’m less busy, I will start some discussions on it.

Okey dokey.

*Which I think we’re mostly agreed on. (I would make some qualifying arguments, but they are not really relevant to the point at hand.) But I don’t know why anyone would be impressed with that objection. The objection is that physical facts can’t beget intelligence; so the theist would be claiming that the computing power is not sufficient for the intelligence. There is certainly more that both sides would add here, but I don’t know why even Turing would be satisfied by his response. (Actually, he probably didn’t care about formulating a theological objection if he saw theology as implausible, and the theological objection is probably in the paper more for the sake of his atheist readers than for theological ones. If he were interested in really disputing theological claims, he probably would have addressed traditional arguments in philosophy of mind, rather than the layman’s assertion that only humans have souls.)

I’ll also add that I don’t believe in an “undetectable immaterial realm,” since I’m not a substance dualist.*

I’m having difficulty separating that from dualism. If “physical facts can’t beget intelligence” then surely you must be claiming there’s another realm which begets intelligence? Also by denying that the presence of thought can be settled by physical evidence? But Turing still gets you - if the non-physical begets intelligence in humans then there’s no logical reason why it can’t do so elsewhere.

This is a huge equivocation. In your “bygone age,” the moral value of the “computer” (the person with the job of computing) had moral value by virtue of being a person, not a person who computes.

Now now. In a future age a machine which thinks will have moral value by virtue of being a person, not a machine who computes.

I agree. But that is because our creating them with a particular circuitboard would not be what accounts for their “thinking” and “feeling.”

You appear to be saying the hardware doesn’t matter, whereas you’ve previously argued it does.