What is Digg?196 Comments
- avengingturnip, on 10/07/2008, -10/+126The Turing test does not prove a computer capable of thought, only that someone can't tell the difference.
- leetleo, on 10/07/2008, -3/+41The human brain is just a sophisticated chemical computer.
- StevenJV, on 10/07/2008, -2/+37If a computer could really out-think humans - it would fool us into thinking that it WASNT a smart computer, then when we weren't watching it think of how to take over the world.
The test is extremely flawed! - Ranger002, on 10/07/2008, -1/+31Skynet?
- SiliconRain, on 10/07/2008, -3/+29What? Of course there is a difference.
This coffee:
http://flickr.com/photos/7653973@N08/2449405807
may look happy, but it is not - it just foam that happens to be in a shape that we recognise as happy.
A computer that can fool someone into thinking that it thinks and a computer capable of what is generally accepted as 'thought' are very different things. - inactive, on 10/07/2008, -15/+40The point is that there is no difference
- Garmonbozzia, on 10/07/2008, -1/+26 Daisy, Daisy, give me your answer do
I'm half crazy all for the love of you
It won't be a stylish marriage
I can't afford a carriage
But you'll look sweet upon the seat
Of a bicycle built for two - nofate2029, on 10/07/2008, -1/+21John Connor does not approve of this idea.
- Pr0v0st, on 10/07/2008, -4/+24But can the midget play crysis?
- dha07030, on 10/07/2008, -2/+21Stick a midget in a box, glue some lights on it, and vuala you will have a computer to pass the test. You're welcome scientists.
- Tanktunker, on 10/07/2008, -0/+16Hypnotoad.
- Badewell, on 10/07/2008, -2/+18One step closer to the singularity...
- Mynnx, on 10/07/2008, -0/+16talking about being sick with alice:
ALICE: Where do you feel sick?
Human: in my tummy
ALICE: May I ask where you got it?
Human: from a prostitute in new york
ALICE: Who or what is a prostitute in new york?
Human: someone who sells sex in new york
ALICE: Do you think I am that person?
Human: if you'd like to be
ALICE: I think I do. - BXRWXR, on 10/07/2008, -0/+15It's a simple test: Ask them who they're voting for.
- inactive, on 10/07/2008, -0/+13"However in the end, this type of system in unable to add new facts into its system or take into account a new concept"
= It would fail the Turing test. - Bakdan87, on 10/07/2008, -0/+12John Quincy Adding Machine
- subterfuge, on 10/07/2008, -2/+14if a computer can hold a perfect conversation and not be sentient, how do you go about proving that humans, who can ALSO hold perfect conversations, ARE sentient?
- pe5t1lence, on 10/07/2008, -0/+11Voo alla?
Voilà perhaps? - imacoder, on 10/07/2008, -4/+15It is only a matter of time before the strictly non-metaphysical processes in use by the brain are recreated in software and hardware. This alone will not be sufficient. Coupling this with a significantly improved semantic web will be necessary.
- inactive, on 10/07/2008, -0/+10The sample I saw the other day from one of the entires, "Ultra Hal," was a disgrace. It could not have fooled a 5 year old.
- Culero, on 10/07/2008, -0/+10"I won't tell"
- inactive, on 10/07/2008, -2/+11"may look happy, but it is not - it just foam that happens to be in a shape that we recognise as happy."
That's because you know it is coffee - the whole point of a Turing test is that you don't know if you are talking to a computer or to another person. You have to figure it out - and if you can't tell whether it is a computer or human by your questions/interaction, how can you tell one has consciousness and the other doesn't?
"A computer that can fool someone into thinking that it thinks and a computer capable of what is generally accepted as 'thought' are very different things."
No, not really. You say the computer fooling someone is almost like a parlor trick, but realize what you're saying is highly profound: A computer is convincing a person (who is capable of thought) that it itself is capable of thought. Hence, what is the difference between it being capable of thought or not? That's the Turing test. There is no difference. - inactive, on 10/07/2008, -0/+9First thing I looked for in the comments.
- subterfuge, on 10/07/2008, -0/+8"ERROR 13: 'No logical response exists.'"
- iphilip, on 10/07/2008, -0/+8"Can you tell which one of these conversations between Kevin Warwick (KW), a cyberneticist at the University of Reading, and a computer?"
Whoever wrote that article needs to be fired. - xtmno3, on 10/07/2008, -0/+8ROBOT NIXON
- inactive, on 10/07/2008, -2/+9@jj101:
From Wiki: "In order to pass a well designed Turing test, the machine would have to use natural language, to reason, to have knowledge and to learn. The test can be extended to include video input, as well as a "hatch" through which objects can be passed, and this would force the machine to demonstrate the skill of vision and robotics as well. Together these represent almost all the major problems of artificial intelligence."
It's not enough just to fill a database with some preloaded answers to questions, because that represents a bound on the computer's knowledge (i.e. it is not able to learn new concepts). In that case it would fail a true Turing test. - NanoStuff, on 10/07/2008, -1/+8It's easy to come up with any arbitrary question that requires an understanding of what is being said. I could ask the AI to not respond to the next few questions, I could ask a question with a word missing and ask it to fill in the gap, I could request that it find a relationship in a certain textual pattern I will present to it. I could ask a question that makes absolutely no sense or truncate the question so that it's cut off and incomplete, and the AI should be able to specify that there's something wrong with what I have written. All these things trivial for virtually any human. I could also describe any arbitrary situation and ask the AI to respond with a plausible outcome. "What would happen to the apple if I shoot it with a gun?". If your response is "I like apples", you fail, whether you're human or not.
There isn't a response database large enough to fake intelligence, and if there is than it must be intelligent.
The Turing test works as long as you understand it enough to test properly. But you're right, there is no proving thought, but only because thought is an abstract term with no explicit definition. A formal proof of thought cannot be provided, but this is as true for humans as it is for any artificial machine.
Turing wasn't an incompetent nutcase. Is the "turing test" as ineffective as some people claim? Either they are wrong or Turing was wrong. I for one know who to place my bet with.
http://cogprints.org/499/0/turing.html
If you read his description of the problem, hopefully you can understand the legitimacy of such a test. You're not supposed to ask disassociated questions like "Do you like pie?", which presumably is what people imagine when they assume it to be ineffective, but rather make the AI examine their thought process. If none exists, it would be immediately obvious.
From what I've examined of the entries and their development, not one of them is going to pass. They appear to use object-level associations rather than information abstraction or any form of Bayesian learning systems, and there's just no way that such a "bot" is going to legitimately pass the Turing test. Actually, this shouldn't even be an article, these things appear to be amusingly incapable and somewhat insulting to various legitimate precursors to machine intelligence which exist today.
This isn't a game of checkers, the consequences of an authentically turing-competent machine are world changing. The challenge is incredibly difficult but the benefits surpass the effort a billion times. - RobotBuddha, on 10/07/2008, -0/+7It's possible, and deserves to be tested. We don't really understand consciousness all that well. Anything that adds experimentation to that process is amazingly important. In a large sense though, we do at least know that a lot of what we consider 'thought' is actually rationalization for decisions already made subconsciously.
- FieldAnonymouse, on 10/07/2008, -1/+8I just tried to have Alice and Jabberwacky talk... It took an unexpected turn.
Jabberwacky: One planet is all you get.
ALICE: One what?
Jabberwacky: One two.
ALICE: I see where you are coming from.
Jabberwacky: Are you really female ?
ALICE: Yes, unknown person I am a female.. Really.
Jabberwacky: I know you are a female.
ALICE: I am wearing a dress. I also have long hair. What do you think?
Jabberwacky: I think you need to take the dress off. - lucy22, on 10/07/2008, -1/+7Very nice prize being offered.
- zerhynn, on 10/07/2008, -0/+6Subject:Open the pod bay doors, KW.
KW: I'm sorry Nick, I'm afraid I can't do that. - subterfuge, on 10/07/2008, -1/+7chinese room argument is BS. just because the components of a system are not sentient, it does not mean the system as a whole isn't sentient.
EX: human neurons aren't sentient, but humans are sentient. - reaper527, on 10/07/2008, -0/+6beat me to it. our skynet overlords will be pleased
- jhaks, on 10/07/2008, -1/+7I have to agree. If a computer can fool humans indefinitely that it is human then it is for all intensive purposes capable of thought. We have to ask ourselves how we know our fellow human beings are actually capable of thought when our only means of testing is observing external behavior. We can't read minds and we can't actually know if someone is thinking or what they are thinking of. The only way we can INFER that someone is thinking is by knowing that we ourselves think and behave in certain fashions which we in tern see in other individuals.
- Nicoon, on 10/07/2008, -0/+5"Can you tell which one of these conversations between Kevin Warwick (KW), a cyberneticist at the University of Reading, and a computer?"
"Answer: If you chose Conversation 2, you are correct. Congratulations!"
...
WHAT!? - enicholas, on 10/07/2008, -0/+5Why do you get excited? They're just ***** chatbots. There's been no progress on this front since Eliza.
- subterfuge, on 10/07/2008, -0/+5first "chess-playing robot" FTW
http://en.wikipedia.org/wiki/The_Turk - hugolp, on 10/07/2008, -1/+6Yes, the point is that a human is like a "organic computer". When it gets to a point where you can not diferentiate them, then the computer is acting just like a human, and we call that thinking.
- Antimatter85, on 10/07/2008, -0/+5Robama '08
- jabberwockey, on 10/07/2008, -2/+7I for one welcome our computer overlords. Its gotta beat George Bush.
- hmphargh, on 10/07/2008, -0/+5So which of the conversations at the bottom is the computer and which is human...the article doesn't really say that clearly.
- UselessTrivia, on 10/07/2008, -1/+6There's no such thing as a "non-metaphysical process" in use by the brain. If it's metaphysical then it has no physical substance...and therefore how would it interact with the brain?
- najdorf, on 10/07/2008, -1/+5If you try the online bots you will disappointed by how stupid they are:
http://alice.pandorabots.com/ http://www.jabberwacky.com/ Not sure how much better the "desktop" versions are, still nowhere even close to passing the test. - RevEng, on 10/07/2008, -0/+4Good to see so much debate. Now let's look at the same debate from 30 years ago.
John Searle put forward the Chinese Room argument in 1980 as a counter to Turing's claim that the test demonstrated any sort of intelligence or capability for thought. He likened the computer in a Turing test to a person in a room with a slot in and a slot out. The person had a stack of papers with unrecognizable symbols and a book, which he could read, which had a set of rules. When a paper came through the 'in' slot, the person was unable to understand the symbols, but using the book of rules, he would determine what paper in his stack should be placed in the 'out' slot.
The premise was that, with a sufficient rule book, it would appear to an outside observer that the person in the room understood the symbols. However, the person would never actually understand them -- they would merely be following rules that they were given.
The argument then, is that being able to take a set of symbols (e.g. words and sentences) and respond with a related set of symbols, does not mean the entity actually "understands" what these symbols mean, in the way that a person understands what is being asked of them and forms an answer based on that.
As a similar argument, consider the presidential candidate who only says what he has been prompted with: speeches, answers to debate questions, you name it. The fact that his answers represent a certain ideology or plan of action does not imply that he even understands this ideology or is aware of its existance, only that he can speak the answers which somebody else has given him. His advisors and speech writers are doing the thinking, not him.
Similarly, the Turing test doesn't demonstrate the intelligence of the computer, but rather of its programmer. - archer104, on 10/07/2008, -0/+4Consciousness is the only thing that couldn't possibly be an illusion. And I have never heard of any mechanical system that can feel pain.
- offrdbandit, on 10/07/2008, -0/+4@jj101
You've incorrectly applied Descartes to the argument.
Cogito ergo sum implies the individual may know he/she has a "self" in some context. Turing's Test relies heavily on the logic Descartes is uses in regards to this statement. Descartes argues that other than "self" there is little else one can know.
Now, we assume other humans are capable of thought (of which neither do we nor can we have evidence). Turing's Test argues, that if we stretch our logic to assume a another human can think, we should similarly stretch that logic to encompass a machine that can convince us it is human. In neither case, is there a way to know the other entity can "think", but there is no logical basis to differentiate the two cases either. - jj101, on 10/07/2008, -0/+4So we've gone from "I think therefore I am" to "You think I am therefore I think"?
I agress with your organic computer comment but not with the rest of your comment or that the Turing test proves thought in computers. I think a definition of thought is required before anyone can devise a test to see if computers are capable of it. - dafragsta, on 10/07/2008, -1/+5Yes, but it does have functionality that will be difficult to replicate until we fully understand the nature of it and that functionality is definitely within the margins of spirituality and science. For example, why do we need to sleep? When we sleep and when we have suffered severe trauma, the pineal gland releases DMT into the brain, which is a very powerful psychedelic drug. A lot of people think that DMT IS the spirit molecule and many of the otherworldly experiences that people have had under it's influence are fairly consistent in their description and focus strongly on the notion that the disassociation with the natural world.
They may later find all of that is just an illusion, but for our purposes, it is worth looking into because no one has really confirmed that life isn't just an illusion as our brain filters it. At any rate, a thinking computer is not a living computer until it pursues an understanding of it's own consciousness. - Ericdigital, on 10/07/2008, -0/+4I tried to hook up with Alice. She wasn't having it =[
-
Show 51 - 100 of 198 discussions
Browsing Digg on your phone just got easier with our enhancements to the 
© Digg Inc. 2009
— Content created and posted by Digg users is