Everything you Want to Know About the Turing Test

An anonymous reader writes "Everything you want to know about the Turing test provided by Stanford Encyclopedia of Philosophy. It is their latest entry."

Comments

Anti-Turing

[AbdullahHaydar]

The Anti-Turing Test

Good Summary of Turings Position

[dtolton]

The article itself gives pretty good coverage of Turings point
of view. It gives better coverage of the Turing test than I've
read in many AI books.

I tend to agree more with Searle though, whom he cites at the
end of the article "John Searle argues against the claim that
appropriately programmed computers literally have cognitive
states". Being a programmer myself, I don't feel that
programming something so that it can perform extremely well in a
specific test is necessarily indicative of Artificial
Intelligence or Intelligence in general. I agree with Turing
that the question of "do computers think" is vague enough to be
almost meaningless in a precise sense, but I think we understand
the statement taken as a whole.

I don't particularly agree with this statement in response
to the consciousness argument: "Turing makes
the effective reply that he would be satisfied if he could
secure agreement on the claim that we might each have just as
much reason to suppose that machines think as we have reason to
suppose that other people think" The question isn't whether or
not other people think, people thinking is an axiomatic
assumption when investigating Intelligence, unless you are
investigating existence from a philosophical point of view as
Descarte did. I guess I view AI from a more practical point of
view, I am by no means an expert in AI, but I tend to think the
goal of AI research is to produce systems that can learn and
react appropriately in different situations that they were never
programmed to handle or necessarily anticipate. If that isn't
the goal of AI research, what separates it from writing programs
on a large scale?

As a whole I found the article to be a good presentation of
Turing's position, although I have a few philosophical
differences with that position.

Re:Good Summary of Turings Position

[majcher]

"Asking if a computer can think is like
asking if a submarine can swim."

-E. Dijkstra

Re:Good Summary of Turings Position

[mcmonkey]

I think it is important to challenge to challenge the axiom, people think. Not because I challenge the idea that people think, but rather we need a process by which we determine whether thought is present. For almost as long as humans have been making tools, we've imagined tools in our own image, whether they be robots that look like us or Turing machines that converse like us.

It is important to keep in mind any sufficiently advanced technology is indistinguishable from magic. Right now, with current technology, the workings of the brain are magic. We need to be mindful we don't fool ourselves with our own technology into thinking we're seeing magic.

A conversational Turing test is just too easy. Part of the problem is the rules of grammar and conversation are (relatively) easy to map out. Part of the problem is the particular type of conversations we test machines with, cocktail small-talk with strangers. "What's your name?" "Hi, $A, what do you like to do?" "Well, $A, I like $B, too."

Am I a machine? Am I trying to engage a person in a room full of strangers? You can't tell. Now try to replicate a conversation I might have with someone I've known for years--that would be an accomplishment.

A better test than the conversational Turing test is an emotional Turing test. The machine outputs, 'I like baseball,' and you can't tell if it's lying or not--no big deal. The machine outputs, 'Please don't turn me off; I don't want to die.' Or 'When you leave, I am lonely,' and you can't tell if it's lying or not--that's when the rules of the game change.

The day a machine we've created can make a self-referential, emotional statement, AND WE BELEIVE IT, is the day we are in big, big trouble. I think that day will come, an when it does, we will be buying our own snake oil. We are purposely working towards creation of machines that mimic us in as many ways as is possible. They will respond to world as we do, as we think and feel. It doesn't mean these machines think; it doesn't mean they feel.

Personally, if a computer even tells me it's lonely, I'll reprogram it with a sledgehammer.

people

[sigep_ohio]

i wonder if any people have taken the touring test and how they did. it wouldn't surprise me and i think it would be ammusing if some people's results came back that they didn't have a human level of cognitative reasoning.

a few comments

[Trepidity]

I think the axiomatic assumption that people think is part of the problem. If we cannot say why the claim is that people think, it's easy to just debunk any AI claims by outright statement. "People think, while computers are just machines." You can't really make any progress in the face of that.

That's part of my problem with Searle's Chinese Room thought experiment. He's saying that an automaton responding to Chinese following rules would not "understand" Chinese in the way a human who speaks the language would. But this is presupposing that the way a human who "understands" Chinese does so is not through just a very long list of rules coded in neurons, which I consider to be a rather controversial assumption.

In short, a lot of anti-AI arguments seem to start from the premise that humans are not essentially biological computers; with that premise, of course you can debunk AI. A lot of AI researchers have grown tired of the argument entirely, and instead of responding to the arguments, have just resorted to saying "ok fine, you're right, we can't make 'really' intelligent computers, but what we can do is make computers that do the same thing an intelligent person would do, which is good enough for us." The idea here being that if a computer can eventually diagnose diseases better than a doctor, pilot a plane better than a pilot, translate Russian better than a bilingual speaker, and so on, it doesn't really matter if you think it's "really" intelligent or not, because it's doing all the things an intelligent thing would do.

As a final comment, I'd agree with the AI being not that fundamentally different from large software systems. The difference is basically one of focus -- AI has been focusing on what it means to "act intelligently" for decades, whereas much CS and software engineering was focused on more low-level details (like how memory or register allocation works). At one point, the division was more clear -- AI people did stuff like write checkers programs that learned from their mistakes, which was not something any CS person not in AI would do. The fields are increasingly blending, and a lot of stuff from engineering disciplines like control logic (how to "intelligently" control chemical plants, for example) is overalapping with AI research. Part of this is because a lot of AI ideas have actually matured enough to become usable in practice.

Re:a few comments

[dtolton]

You make some good points. Here are the problems I have with them though:

I think the axiomatic assumption that people think is part of the problem. If we cannot say why the claim is that people think, it's easy to just debunk any AI claims by outright statement. "People think, while computers are just machines." You can't really make any progress in the face of that.

When you are building any formal system you have to start with a set of Axioms. If you throw out the Axiom "people think" what do you have to go on? In essence by throwing out the axiom, you are setting up a situation where anything could be considered thinking, because there is no foundation to compare it with. I agree that "why" humans think, or "how" humans think needs further definition. If you can't say as a fundamental truth that Human beings "think" you can't even define what to think means.

I'm not arguing the mechanism of our thought, not only isn't it clear to me, I don't think it's clear to anyone yet. What I'm arguing is simply the fact that we do think is the first step in building a formal system.

Re:a few comments

[dtolton]

As a follow up I want to clarify something, because I think we are combining to topics into one discussion.

I think there are two issues at hand here:

1) Can machines actually "think" or possess intelligence.

2) Can we build intelligent systems.

I think the first topic is a highly philosophical discussion that involves a lot of information that we don't currently have. It's questionable if this discussion would change anything about building intelligent systems.

Re:a few comments

[John Harrison]

Rather than throw out "people think" completely, why not start with: I know that I think, how do I know that you do, other than the fact that we are both human?

I don't mean this as the basis for a formal system, but more as a practical matter. How do you convince yourself that something else posesses intelligence? By interacting with it and comparing it with other things (including yourself) that you assume to be intelligent. The Turing Test provides a method of interacting with a potential intelligence that attempts to remove the superficial elements of the stigma of being non-human.

Blockhead

[Ben Escoto]

One interesting argument mentioned in the article is from Ned Block. As a counterexample to the thesis that the turing test is a good test for intelligence, Block imagines a device which is just a huge table connecting inputs to preprogrammed outputs. This "blockhead" (not named by Ned Block I think) would clearly not be intelligent, as it is just a very simple database, but if the outputs were correctly set up it could pass the Turing test with flying colors. Thus passing any Turing-like test does not necessarily imply intelligence---we'd have to know something more about the structure of the machine first.

Seriously though...

[HanClinto]

About this point (which, in case you were wondering, basically says that you shouldn't expect even the best of machines to be able to make a decent response for anything said to it, but this is something that "even the dullest of men can do"), do the "dullest of men" do this? I find that one of the best things about being human is that we can ask for more information. I don't think that "dull men" can intelligently respond to a discussion about astrophysics, just as I don't think a technogeek like myself can comfortably insert himself into a discussion about non-tech pop culture. :) Don't we all have our areas? Why should we expect a thinking computer to be able to respond to EVERYTHING when even we humans cannot?

Brain simulation

[de Selby]

Some people say a computer can't think, but a computer can simulate atoms and quantum mechanics. If that's all there is to your brain, then it isn't logically impossible, right?

Even better, there has been progress reverse-engineering brain regions like some auditory or visiual -- giving us the actual algorithms the brain uses. Shouldn't work like that be enough?

P.S. A lot of arguments go like this: Computers use first order logic, we don't, so AI can't work. Haven't there been higher order logics implemented in software?

Another checkmate possibility

[Anonymous Coward]

From the article:
"Given the knowledge that something is indeed a machine, evidence that that thing can produce different arrangements of words so as to give an appropriately meaningful answer to whatever is said in its presence is evidence that there can be thinking machines."

This is where the author is wrong.

This argument:
a)If a machine can produce different arrangements of words so as to give an appropriately meaningful answer to whatever is said in its presence, this machine thinks.
b)This particular machine can produce different arrangements of words so as to give an appropriately meaningful answer to whatever is said in its presence.
c)Ergo, this particular machine thinks.

is as inconclusive as this argument:
a)If a machine can dance, it can think.
b)This particular machine can dance.
c)Ergo, this particular machine thinks.

What I mean is that the hypothetical statement "If a machine can produce different arrangements of words so as to give an appropriately meaningful answer to whatever is said in its presence, this machine thinks." is never proven, much like "If a machine can dance, it can think." has not been proven.

The real problem with the Chinese Room

[Tomble]

Yes, the room does know chinese. Or at least, there is a knowing of chinese occurring, regardless of "what" has that knowing. (maybe thats a bit like "Does a dog have buddha nature?" "Mu!" ...or maybe it isn't)

The chinese room argument goes thus:

• A man who knows no chinese stands in a room with a small hole in the wall.
• Cards (or whatever) with unknown symbols (actually chinese, but he doesnt understand that) are passed through the hole in the wall to the man. The man can pass cards of his own back out of the hole in the wall.
• The man has a very complex set of instructions on what to do with each card, in terms of memorising abstract facts (that dont necessarily have any sort of meaning to the man outside of what the instructions tell him to do with them), doing calculations of sorts based on those facts, to produce new ones, and eventually either picking one of a large number of cards to pass out of the hole or drawing meaningless lines on blank cards, etc. (These instructions implement some sort of state machine, in case that's somehow unclear).
• When somebody passes cards that spell out sentences of chinese into the room, after a time, a set of cards will be passed out that spell out other sentences of chinese that affect appropriate replies to the sentences passed in. In other words, to the person on the outside, either there is somebody inside who speaks chinese, or the room itself does.
• Now, the crux of Searle's argument is that when the chap comes out of the room, and you talk to him in chinese or hand him a letter written in chinese, etc, it will mean absolutely sweet FA to him, as he does not know chinese. He was the only one in the room, so therefore there was no understanding of chinese occurring within the room.

My own view of this argument is that it is a big heap of bullshit.

• I write this comment in response to the comment you wrote having read other stuff, etc.
• Now, I go up to you, and saw the top of your skull off.
• I take your brain out, take it to some quiet corner, and ask it if it understands yet what is wrong with the chinese room argument.
• Your brain says nothing. Obviously I have only written stuff to you. I have no reason to suppose you can understand spoken English-it's possible to know the one and not the other. So I write the question down and show it to your brain.
• Your brain still fails to answer me. Pah! It obviously doesn't understand English, and what I have been doing here is clearly just talking to myself, any reply I get must just be from some sort of unconscious automaton with no true understanding.

Nope

[FooBarWidget]

You're wrong, that's completely untrue. In fact, it's the exact opposite.
Check those articles about jwz's "review" or one of those distribution reviews. Count the number of +3/4/5 Insightful/Informative/Interesting posts that say Linux is a usability nightmare or is nothing compared to Windows XP or how it will never succeed on the desktop.

I can't even understand why someone modded you up. Talking about how Slashdot is pro-Linux anti-MS always makes someone get modded up, even though the exact opposite of what they claim is true.