[ragavpayne]

Hello,

I just purchased the book and I'm just leafing through the pages. so, I don't know if the answers to the questions i'm going to ask are already in the book. But, anyway, let me give it a try. The questions i'm going to ask are about the chinese room thought experiment.


1)
The experiment says:-

"The book has instructions dictate ways to manupulate, sort, and compare chinese characters. Mind you, the the directions say nothing about the meanings of the chinese characters; they only deal with how the characters are to be copied erased, recorded, transcribed, and so forth."

The experiment concludes that the answer offered by the room is correct but the room isn't intelligent.

I think that the experiment is flawed and has a paradox. Let me elucidate :-

To begin with, can you imagine an "answering-book" that has instructions only on how to copy, record, erase transcribe etc., chinese characters but doesn't tell/know the meaning of the question/character?

And if there's any part of a thought experiment that you cannot imagine or replicate in real world, then its the missing link. The missing link can connect any unconnected chain of logical reasoning.

The only thing that could lead to the correct answer is the book being itelligent, which is impossible. So, I conclude that the thought experiment wouldn't end the way its creator wanted it to. It would have ended with gibberish on paper.


The "instructions book" itself was the biggest "flaw-indicator" of the experiment. For me, the experiment is like the one aristrotle came up with. The one that let him conclude that lighter objects fall slower and heavier objects fall faster.

Am i the only person who see's this as a paradox?

2)
The experiment concludes that computers can't be intelligent because they don't understand the story. why can't it be looked at this way? If the computer can reach its objective somehow, then why do we care if it understands or not? It doesn't matter until the work given to the compuer is done. Anyway it might matter if the work given to the computer requires it to come up with its own objectives which ofcourse requires prejudice.


Anyone who thinks that the experiment is genuine, then please clear my doubts.


Regards.
--
rAgAv.

[thesaturnine]

Hi all,

The Chinese Room analogy tries to reach its conclusion by arguing that 'human understanding' is more than just the ability to swap symbols. Actually, I am not sure if this is actually argued for, or if it is simply assumed. In any event, some of my research motivates me to set out towards the complete destruction of that position.

(1.) Naturally, lay-people tend to assume that there is more to human experience than a systematic exchange of images. (1.1.) They feel that there is more to life - indeed, most lay-people will believe that there is a significant portion of their daily human experience which cannot be quantified and/or rationalised. (1.2.) Personally, however, I dare say that all human experience IS quantifiable.

(2.) Nevertheless, not all human experience is quantifiable by every human being. (2.1.) A person with superior quantifying abilities would be able to consider any human experience, and compeltely articulate its structure in mathematical terms, as a distribution of data. (2.1.1.) I believe that I have acquired at least a bit of this ability in recent years. (2.2.) But lay-people simply do not have this ability. (2.2.1.)However, I have begun to wonder if it is possible to train people to have such abilities.

(3.) The ability to study raw human experience, and to systematically quantify using the language of mathematical distributions - perhaps it is best called 'informatical phenomenology'. (3.1.) I suggest this because it seems to me that the mind is an information system - surely the brain is an information system. (3.2.) Alternatively, if one feels the need for funkier vocabulary, it might be called 'atomic phenomenology', 'quantitative phenomenology', or even 'quantum phenomenology'. (3.2.1.) I say this because (speaking from experience) the first step to understanding any human experience as a mathematical object, is to subject the raw experience to a simple analysis - a concrete delineation of its parts.

(4.) Such a study then begins with the question, 'what are the atoms of experience?' where atom here is taken in the traditional sense, of being a small discrete unit. (4.1.) In an attempt to answer that question, I suggest that we take into consideration the word 'pixel' which is commonplace in the field of information technology. (4.1.1.) 'Pixel' stands for 'picture element', that is to say, a pixel is a small part of a picture. (4.2.) In this way begin to discuss experience in terms of pictures. (4.2.1.) To begin, we might discuss each of our sensory modalities, as providing us with different types of pictures - a sensation/experience of sight is then an image in the optic mode, a sensations/experiences of sound are images in the acoustic mode, sensations/experiences of touch are images in the haptic mode, (and the thermal mode, etc.), s/e's of taste are images in the gustatory modalities, and s/e's of smell are images in the olfactory mode, and so-on-and-so-forth. (4.2.1.1.) Perhaps there are more basic categories of data, in fact I think that there are - but for now, this provides a fairly intuitive introduction to the subject of quantitative phenomenological analysis.

(5.) A major point of information *sic* is that there are two major realms of human experience which are simultaneous, and superimposed upon each other. (5.1.) The first realm of experience is the dataset from the nerves in our body (from the PNS - the peripheral nervous system). (5.2.) The second realm of experience is the dataset from the imagination (from the CNS - the central nervous system). (5.2.1.) I want to clarify that this second dataset, the 'imaginary dataset', includes any sights, sounds, tastes, feelings, etc, which we recall from memory, think about, or otherwise simulate in the mind. (5.2.2.) To be absolutely precise on where I stand on this matter, I would like to suggest that in humans there is no conscious cognition whatsoever, without an imaginary component. For example: (5.2.2.1.) when we think of apples we actually refer to the images of apple-sights, apple-sounds, apple-tastes, and apple-smells which are stored in our memory - and to no such thing as a purely abstract concept of apples(5.2.2.2.) when we recognise apples in the world it is simply that the imagination has traced out the gestalt of apple-images from an otherwise general dataset-from-the-PNS, (5.2.2.3.) i.e. imagination IS consciousness, and so programs like NuPIC are technically conscious in a simple way. (5.2.3.) Jeff has a possible (and probable) biological mapping for this function in On Intelligence. (5.3.) This second major consideration, in this method of analysis which I am proposing, has more to do with the 'informatical analysis' rather than the 'mathematical anaysis' of raw experience.

OK. That's enough for now. If you understood and agreed with all of the above on the first reading, I would be very happy to have thought-company. I think for some people, this understanding will come quite easily, whereas others will simply not get it for quite some time.

Because it is fundamental phenomenological model - a way of seeing the world - a worldview - a view of the world as the sum of many Lego-like components of experience. And some people have already lodged themselves in certain worldviews from which they dare not dislodge their consciousness.

But I like this model because I cannot find anything in my experience which contradicts it - yet. I am subjecting it to constant testing, of course. Anyway... if there are any replies to this post I may not respond until 15-June when I get back from some training program that I will be attending abroad.

Till then... I hope you enjoy thinking about how I think... haha.

Cheers.

-yj

(I read On Intelligence some time last year, and thought that Jeff's ideas are quite close to my understanding of how the mind processes data. Recently I noticed some of Jeff's more recent work vis-a-vis @ Numenta and in the Foleo, so I return here, as a fan. I have studied the structure of my own experience in much more detail than I have studied the structure of the brain, so I will say I am a specialist in studying raw experience, not a specialist in studying the brain. People like Jeff are specialists in studying the brain - and I am here because think that there is a niche for complimentary interactions between informatical neurologists and informatical phenomenologists.)

[iammisc]

The whole point of the Chinese Room Analogy is that the current way of doing AI (giving the computer a whole list of instructions) doesn't result in understanding but as the OP said, in gibberish.

[Lloyd]

This argument has generated an incredible volume of heat over the years since it was promulgated. There are many wise rejoinders, but here I will simply respond to the first reply. In his instructions that the "operators" in the room would know only how to shuffle symbols does not do justice to the obvious fact that "the room" had to have learned how to read Chinese, even though the operators inside were only shuffling.

Still, in the long run, I suppose this whole business has been valuable for the AI and philosophical communities.