Social Question

LostInParadise's avatar

Is the Turing test valid for detecting consciousness?

Asked by LostInParadise (32186points) February 22nd, 2013

The Turing test was proposed by the mathematician Alan Turing as a means of determining whether a computer has consciousness. The test is simply to have a person ask questions of both a computer and another person and, on the basis of the answers, tell which is the computer and which is the person. I think that we can agree that this is a minimal test in the sense that if the computer consistently fails it, we can say that it does not have consciousness.

I think the test is reasonable. If it walks like a duck and talks like a duck then it is a duck. If there is no way of distinguishing the behavior of a computer or, more generally, a robot, from that of a human, then the computer has consciousness.

The philosopher John Searle proposed an argument known as the Chinese room argument to argue against the validity of the Turing test. Imagine a room with a library of books that tells how to answer in Chinese any question written in Chinese. A person in the room answers questions written in Chineses submitted through a slot by looking up the answer and then returning the answer through the slot. Searle maintains that the person does not understand Chinese, despite behaving in a way indistinguishable from someone who does understand Chinese. He says that the behavior of the person is analogous to a computer executing a program.

I would use the following argument to counter Searle. Firstly, I would question being able to have an answer to every question. This would make the library omniscient, like God, and I could use the argument to question whether God has consciousness.

Going with the existentialists, I would say that what makes us consciousness is the need to make choices. Our humanness is determined as much by our limits as our strengths. If I were participating in a Turing test, I would ask questions of a personal nature. What is your favorite color, what foods do you prefer, what teams do you root for? To answer such questions, the computer has to limit itself to take on a specific personality. If it could do so convincingly then I would say that it has consciousness.

Observing members: 0 Composing members: 0

8 Answers

ragingloli's avatar

Of course. That is the same way you judge humans to be conscious. It would be hypocritical to apply different standards to another life form than you use for yourself.
And no, “but it is a machine” does not count as a valid argument, that is just another form of racism.

Furthermore, that whole “computers are just huge databases that store every possible question and the according answer and thus can not be conscious” is a strawman, because it assumes that AI has to be of that nature, which just is not the case, as actual research and development into AI does not go into that direction, but the ability to interpret question, make internal connections between topics and making probability assessments, like that AI they made to compete in a game of Jeopardy, which a pure database could not solve, because you have to interpret the answer as a question and understand what is asked, also taking into account slang, puns or references to popular culture.
The AI won, by the way.

janbb's avatar

@ragingloli The AI won, partly through speediness, but it made some huge gaffs that very few humans would have made.

LostInParadise's avatar

There actually is a group that performs tests No winners so far.

El_Cadejo's avatar

How odd, I was just talking to a friend about this last night. I do think the Turing test is valid btw. We’re in some interesting times right now.
When our computers (in the not so distant future I’m guessing) are able to simulate a reality to a level that it is no longer distinguishable from that occurring outside the world, would those experiences one had in the game be any less real than those of “outside life” where is the line drawn?

wundayatta's avatar

If you’re talking to some entity, and you think you are talking to a person, then are you not attributing consciousness to them?

I don’t think we’re there yet in terms of programming computers to behave the way a person does, and I doubt we will be there for a long time, if ever.

I think the test is reasonable. I’d be surprised if we ever attribute the quality of consciousness to something we programmed.

burntbonez's avatar

How do you decide another person is conscious? Think about that.

SomeoneElse's avatar

Sorry, but I thought you had put ‘Turin’ and that’s shrouded in mystery . . .

Paradox25's avatar

It is good for detecting human-like dialog, but not sentience.

Answer this question

Login

or

Join

to answer.
Your answer will be saved while you login or join.

Have a question? Ask Fluther!

What do you know more about?
or
Knowledge Networking @ Fluther