Searle through Turing on Intentionality

Ryder Dain

Oct 30, 2012 19:45

Hm. No one's here, but hello all the same. Hopefully my time zone and opportunities to chat will come into grand alignment at some point.

Here's a brief question: this entire chat interface is a little bewildering. It's a chat -- do we really need to be able to flag or edit a real-time statement?

EDIT: I mean, it's great for saving face and all...

Joseph Weissman

Oct 30, 2012 22:34

@RyderDain hello! Chat definitely takes a little getting used to

I think editing and some of the bells and whistles (one-boxing, starring, etc.) are really just nice-to-haves

(Editing does make it a lot easier to correct minor typoes that would otherwise clutter the log with *-style corrections!)

Ryder Dain

Nov 1, 2012 22:59

@JosephWeissman Since this is the less formal forum, can I ask you: why do you pursue philosophy? Not to ask why you think it's relevant or useful- even if that's the case -but instead, why is it important to you?

Joseph Weissman

Nov 1, 2012 23:23

@RyderDain I am tempted to answer you in Nietzshean terms here -- who but philosophy is interested in the enterprise of thought or critique, at their most affirmative and aggressive?

Who else is interested in generalized demystification, the exposure of all fictions, whatever their source or aim?

Joseph Weissman

Nov 1, 2012 23:40

I study philosophy to grow.

I'm a writer; that's my primary self-identification. The life of the mind means a lot to me, and a healthy diet of theory keeps me embarking regularly on critical explorations of problems.

Ryder Dain

Nov 2, 2012 23:11

That was excellently said.

Beautifully written, as well. You've gone and reminded me that I haven't been reading enough.

Joseph Weissman

@RyderDain thanks!

Ryder Dain

So, Mr. Weissman - what's the current nut you've been trying to crack? I read a little of your post on ontology.io, and that made me wonder where you stand on Searle's Chinese Room experiment.

Maybe a leap, but I'm always curious about different takes on that one. A little poetry in the answer never hurt, either.

Joseph Weissman

Nov 4, 2012 03:40

@RyderDain working through the same stuff as always, really -- lately it's been Bachelard and Serres again, but I'm planning on ordering some new books soon; a lot of great stuff has come out pretty recently! :)

Thanks for the reading. I'm not sure exactly how the Searle would be connected, but I think it's an evocative thought experiment which maybe doesn't really demonstrate what it thinks it does

Joseph Weissman

Nov 4, 2012 21:31

Anyway, would love to explore it with you. Again, my sense is the Chinese room wants to say "substrate matters" but then ends up effectively demonstrating the opposite.

In other words, I don't think the evidence he attempts to present really do much to argue the case that the machine (which in his hypothetical instance can successfully pass a Turing trial in a natural language) is somehow metaphysically incapable of supporting sentience.

I may be misreading the experiment, or overlooking some nuance on the argument.

In passing it does seem strange to me this strange resurgence of dualism as the technical possibility of artificial "general" intelligence begins looming into view.

Again, I am not sure Searle would subscribe to matter-mind dualism; my guess is he's trying to make an argument about embodied consciousness under sort of neutral, monist metaphysical givens.

But we come back to a rather Spinozist question -- after all, what are the limits of what a body can do? Do we really know?

I don't think there's any particularly convincing argument that sentience is dependent on a particular substrate.

It was already Turing back in the 1950s who noted rather pointedly that the biological processes at work in the brain could be simulated with a computer.

Ryder Dain

Nov 6, 2012 12:10

Having listened to Searle's own lectures on the topic for Spring 2012 (they're free on iTunes as an 'iTunes U' podcast), I can reliably report that he rejects any notion of dualism - in fact, goes out of his way to fight the label of even property dualism so strongly that I have to wonder if he's succeeded or not. I think he's on the side of "emergentism" for the most part, if that's even a coherent stand-alone position yet.

I've often felt what he was trying to get at with the CR was to demonstrate that our understanding of what understanding actually is was flawed, and leads to yet another "homunculi all the way down" sort of explanation. That is, that automated processes can't explain the intentionality behind our normal use of language, even if perfect mimicry was possible.

...it's a little funny that he would both take that position, yet find even the dandruff of dualist notions so abhorrent.

Joseph Weissman

Nov 7, 2012 00:42

Yeah, I guess that irony is really all I'm reacting to. It's a little symptomatic to me, anyway.

To unpack it a bit: analytics want logical empiricism, but they want the transcendental too -- and they're not willing to consider continental philosophy (or transcendental empiricism.)

For the most part unwilling. I'm hoping the next generation of thinkers will be more inclined to careful reading and less viciousness.

I think one difficulty here is that AI has been simultaneously thought in theological and scientific registers.

At any rate one question to explore might be: what is really at stake in this problem of a machine which can pass a Turing trial? Why does it have metaphysical import at all?

I think we're talking about spirit, at some point, with Turing's strange game.

The open-ended way social interaction unfolds -- after all, a conversation cannot be exhaustively planned.

Anything whatsoever -- the conversation can become a world, a thought, a life, a feeling.

It has to at least to pretend to have a world, a life, thoughts and feelings.

Just in passing, I think this points to an important connection between the Turing test and the Turing machine -- the universal machine, the machine which can imitate any machine we can imagine.

I might tentatively suggest the problem is precisely that dualism has definitively collapsed, and presumably we're going to be creating (and exploiting) artificial sentience with techno-science.

The ethical implications tend to be poorly thought through.

Joseph Weissman

Nov 7, 2012 01:14

It seems to me :)

Joseph Weissman

Nov 7, 2012 01:37

At any rate, I do find myself wondering about this idea that a thinking machine could be a metaphysical impossibility.

It would perhaps seem to ignore the power of Turing's innovation -- the pure plastic capacity of programmable computers, and in particular, to simulate processes of arbitrary complexity...

Joseph Weissman

Nov 7, 2012 01:53

(The meaning of a simulation, maybe, is what is "really" at question.)

Joseph Weissman

Nov 9, 2012 00:05

Anyway, @RyderDain, would love to discuss further if you had any particular concerns or questions in mind.

Ryder Dain

Nov 10, 2012 00:01

Nothing really particular, no -- but as my mother used to say, if you throw enough poop at the wall, something's gotta stick. I'm interested in the nature of intentionality, so the CR and Turing are generally good places to start.

As for your question, "What's at stake" ... It's fairly trivial to point to any reading of the Turing test as a claim that we should accept passing it as sufficient evidence that thinking is going on. I strongly doubt that Turing himself had any notions of that processing of language as *intentional*, though I'd appreciate it if someone could provide contradictor…

However, that doesn't seem to be the case for people that have come after him. Rather than seeing Turing's claim as one about the relative simplicity of calculating syntactical rules well enough to mimic the kind of processing our brain does when we speak in response to others,

I imagine what's at stake for some people is predicated on their assumption that the thesis of the Turing Test in fact presumes something about intentional states - or reveals these same people's commitments to seeing language as instantiating intentional states.

That's not to say it's impossible that intentional states couldn't be realized on different architectures other than the squishy, pink meat we use to think- but presuming that passing a speech-test is sufficient to demonstrate intentional states is a bit of a dramatic leap.

Apologies for the late reply - I ran off to a conference and I've been away from my computer for the last couple of days.

and now to bed :)

oh - @JosephWeissman - I should probably include your name to signal your browser, anyway

Joseph Weissman

Thanks! Yes, the @'s are definitely helpful for facilitating dialogue :)

I wonder if the "intentionality" we're talking about isn't already consciousness/sentience.

I think Turing's trial is pretty elegant for determining whether there is a thinking agency under examination.

Ryder Dain

Nov 10, 2012 00:20

How so?

That is to say - how do you get from speech to conscious thoughts driving, directing, or even just "being about" the words being applied as a response?

i really need to get to bed, sadly :( conference continues at 9:30 in the morning and it's already after 1am

but I'll pick up if you wouldn't mind leaving a response

Joseph Weissman

Nov 10, 2012 01:03

@RyderDain well, again, the short answer is spirit -- "passing" for a legitimate partner in the open-ended process of a conversation

It's precisely a means of making sentience legible

A conversation can become (about) anything -- it can even turn towards itself, towards philosophy

Turing's test is elegant because it includes all this 'by proxy' -- that is to say, through this pure "imitation" that takes place in the trial

(I think this infinity through simulation has an interesting resonance in the Turing machine itself; the machine which can become any conceivable machine.)

Turing gives us as experimental proxy for consciousness the property of being capable of fooling another sentient creature into believing it's sentient

Simply by engaging in convincing conversation, it has demonstrated itself at least potentially sentient. The point would be that in a conversation, you can go in-depth into anything at all -- it really demands the system under examination have some robust sense of what a world is, what thinking and feeling mean, and so on.

It means it has to have a sense of the holy, or sacred, for instance.

Kurzweil talks about spiritual machines, in a somewhat accelerationist vein, but I think it's appropriate at least in some sense to think about sharing 'spirit' with machines.

Spirit means 'breath', right? So again I guess the point I would emphasize would be about the nature of the conversation as such -- the infinity and future that conversation inherently opens onto.

Ryder Dain

Nov 11, 2012 11:07

I tend to avoid words like "spirit" or "soul" myself, since they're so heavily loaded terms with so many definitions depending on the context. Jewish philosophy goes so far as to outline different categories of spirit (judaism.stackexchange.com/questions/8921/…) for identifying what aspect of the soul we're talking about.

One view I didn't see at that link, which I was taught in elementary school, was that there is first "chai", which applies to anything we think of as "alive", inclusive of plants - "ruach", or "spirit" (rough translation) is taken to mean the animating force of life, inclusive of insects or some nominal class of autonomous but otherwise unthinking creatures.

"Nefesh" (also called the spirit, or sometimes soul) seems to indicate rudimentary consciousness or awareness, and finally "Neshama", also sometimes translated as "spirit" but more often as "soul" captures the agency ascribed to human beings, which carries with it the requisite moral responsibility common to the kind of explicit, God-directed structured ethics that comes with most religions.

In that respect, "Neshama" is the root of immortality that most people would ascribe to most notions of the soul -- but again, this is just the Judaic take on the notion. Hindu conceptions would necessarily be different, mixing different parts of these concepts to support the constancy of reincarnation as different non-human living creatures.

Not to mention the christian notions, most of which I'm given to understand lump all aspects of vitality, spirit, and agency into a single force of "soul" which is instantiated by a fetal zygote at conception - whereas a more nuanced view would say only one or two of the more prototypical forces common to life are present at conception, and "spirit" and "soul" only enter at a later date.

In the Judaic esoteric view, agency alone is given the label of "breath", despite the literal translation of "ruach" as wind - taken from the Sefer Yetzira:
1:9 Ten Sefirot of Nothingness: One is the Breath of the Living God, blessed and benedicted be the Name of the Life of worlds. Voice, Breath and Speech. This is the Holy Breath (Ruach HaKodesh)."

That's more of a confusion in hebrew translation, where "ruach" typically means "wind" when it stands alone, but "breath" when its conjoined with a proper noun. It's a muddle.

In any case, I suppose what I'm getting at is, if you're going to use "spirit" here to refer to some component which is constitutive of consciousness, could you elaborate on what you mean by the term a little more?

Joseph Weissman

Nov 11, 2012 16:57

Well, I guess one less "overloaded" way to put would be to frame it in terms of poetic imagination.

Just for motivation of all this -- a much weaker but very general point about the Turing trial is that the system under examination would have to be capable of learning

Which already means giving the appearance of reflection and analysis and so on

Perhaps more provocative, it is tempting to give a Nietzshean reading of the Turing trial

In other words -- what is the meaning of the simulation of intelligent conversation?

Turing wants to say, well -- they're effectively the same (simulation and "reality"). But there's a strange possibility left open, right?

Stanislaw Lem writes some interesting fiction about artificial intelligence. There's one in particular I am thinking of where a higher-order intelligence doesn't actually possess subjectivity or identity in any human sense

But can discourse eloquently on the subject (explaining at length why it claims it is "no one.")

It passes the Turing trial, but through the construction of a "mask" to interface with us

Any system passing a Turing trial, imitating an intelligent conversational partner successfully, has to demonstrate at least a certain degree of creativity and imagination.

Spirit may be inappropriate for other metaphysical reasons; but insofar as breath is the medium of the infinite movements that attend speech and to which any convincing intelligent conversational partner would have to be attuned...

In light of the Nietzschean reading, I guess I would underscore that my point about the system grasping the holy or sacred, having a spirit, etc. -- this should all be said at least as much culturally-biologically as metaphysically or theologically

S.M.

Nov 11, 2012 17:16

@JosephWeissman you seem to be alon here.

Joseph Weissman

(@S.M., we have kind of been sending telegrams back and forth.)

S.M.

I see.

Good day.

Joseph Weissman

In other words it would have to be able to imitate what a human conversational partner might understand as having a "soul" or "spirit"

@RyderDain The bottom line might be that the Turing test is indexed to the species and culture and conversational expectations of those performing the trial

@S.M. Sorry! Didn't intend to run you off. Just wanted to make sure I had gotten that last thought out :)

Joseph Weissman

Nov 13, 2012 21:01

Joseph Weissman

Nov 13, 2012 23:33

Alan Turing (June 1912 – 7 June 1954)

QuietThud

Nov 15, 2012 20:43

Turing!<3

Joseph Weissman

Nov 15, 2012 20:53

@QuietThud I know! What a mind...

Joseph Weissman

Nov 15, 2012 22:52

Alan Turing

Alan Mathison Turing, OBE, FRS ( ; 23 June 1912 – 7 June 1954), was a British mathematician, logician, cryptanalyst, and computer scientist. He was highly influential in the development of computer science, giving a formalisation of the concepts of "algorithm" and "computation" with the Turing machine, which can be considered a model of a general purpose computer. Turing is widely considered to be the father of computer science and artificial intelligence. During World War II, Turing worked for the Government Code and Cypher School (GC&CS) at Bletchley Park, Britain's codebreaking centre...

Ryder Dain

Nov 16, 2012 22:03

@JosephWeissman Thanks for the Turing photo :) -i hadn't meant to leave this thread hanging, but I'm chewing over your answer and will have to wait until I'm in front of my larger Turing-complete machine to respond.

The Symposium

Participants