When does technology become artificial intelligence?: Philosophy Forums

Philosophy Forums » General Philosophy » When does technology become artificial intelligence? » When does technology become artificial intelligence?

When does technology become artificial intelligence?

Page: 1 2 3 4 5

When does technology become artificial intelligence?

OnceFlewOver
Aspirant

Usergroup: Members
Joined: May 01, 2004

Total Topics: 4
Total Posts: 32

Posted 08/13/04 - 11:26 AM:	#11
zephos wrote: Not sure exactly sure what questions your looking to have answered. When does technology become A.I.? When it passes the Turing test. a.k.a. When it can successfully fool the average human in conversation for an indefinite amount of time. I think the bigger question that you bring up deals more with whether or not one can prove that human thought is not bounded by the very same 'prerecorded actions.' What is the Turing test exactly?

AKG
Tenured Poster

Usergroup: Members
Joined: Jan 15, 2004
Location: Berkeley, CA

Total Topics: 87
Total Posts: 3329

Posted 08/13/04 - 03:19 PM:	#12
OnceFlewOver wrote: What is the Turing test exactly? I'm sure it can be set up in different ways, but basically it involves having a human have something like an online chat with a few "people," some of those people being human and some of them being machines. If the human can't distinguish the human chatters from the machine chatter(s) then the machines have passed the Turing Test. I think the Turing Test is rather bad test of intelligence. Human conversation is strongly influenced by culture. Human conversation is a small part of our intelligence. It would demonstrate some intelligence, as the ability to communicate in a relativly complex language (natural language is more complicated than current machine language I would imagine) would require certain intelligence. But the Turing Test relies on one thing that has nothing to do with intelligence - how well adapted the machine is to modern culture. Moreover, the Turing Test doesn't test a number of important things, like creativity, the ability to adapt to new situations, etc. Conversation is a human "problem" and intelligence shouldn't be measured as the ability to solve human problems only, but any given problem where the solution is relatively complex. I suppose I'm repeating myself but to clarify and summarize, I think the Turing Test not only misses on a large number of skills relevant to intelligence, it even tests the machine on somethings irrelevant to intelligence.
"The only reason we die... is because we accept it as an inevitability." -- Stewie "To enslave nuance to dogma is folly." -- Lord Hillyer

zephos
Assistant Professor

Usergroup: Members
Joined: Jul 13, 2004

Total Topics: 5
Total Posts: 406

Posted 08/13/04 - 04:32 PM:	#13
AKG wrote: I'm sure it can be set up in different ways, but basically it involves having a human have something like an online chat with a few "people," some of those people being human and some of them being machines. If the human can't distinguish the human chatters from the machine chatter(s) then the machines have passed the Turing Test. I think the Turing Test is rather bad test of intelligence. Human conversation is strongly influenced by culture. Human conversation is a small part of our intelligence. It would demonstrate some intelligence, as the ability to communicate in a relativly complex language (natural language is more complicated than current machine language I would imagine) would require certain intelligence. But the Turing Test relies on one thing that has nothing to do with intelligence - how well adapted the machine is to modern culture. Moreover, the Turing Test doesn't test a number of important things, like creativity, the ability to adapt to new situations, etc. Conversation is a human "problem" and intelligence shouldn't be measured as the ability to solve human problems only, but any given problem where the solution is relatively complex. I suppose I'm repeating myself but to clarify and summarize, I think the Turing Test not only misses on a large number of skills relevant to intelligence, it even tests the machine on somethings irrelevant to intelligence. The problem with statements like that above means that perhaps some humans will fail to pass the intelligence test when humanity agrees the 'average' human is indeed intelligent. That is to say, not everyone is skilled at creativity or adapting to new situations, so to demand that a computer do these tasks seems a bit unfair to prove intelligence. On the other hand the Turing test does include these things, for a machine to pass this test it must adapt to each new individuals conversation to pass well. It must be able to react "humanly" to a wide range of personalities. Creativity can also be tested in a Turing test as well. If you ask it to make up a story it would have to do so, without a database of pre-built stories this would require creativity. Ultimately I feat other tests of intelligence not because machines will someday maybe too easily pass the Turing Test, but rather that any other test humans think of might not include all humans.
spelling, grammar, and incoherent expression of ideas brought to you by the public education system...

AKG
Tenured Poster

Usergroup: Members
Joined: Jan 15, 2004
Location: Berkeley, CA

Total Topics: 87
Total Posts: 3329

Posted 08/14/04 - 06:57 AM:	#14
zephos wrote: The problem with statements like that above means that perhaps some humans will fail to pass the intelligence test when humanity agrees the 'average' human is indeed intelligent. That is to say, not everyone is skilled at creativity or adapting to new situations, so to demand that a computer do these tasks seems a bit unfair to prove intelligence. So, some humans won't pass the Turing Test. A machine shouldn't have to be able to pass all the tests, and "creativity" is a vague concept, so I don't know how you can know for sure that not all humans can pass a creativity test. I didn't necessarily mean artistic creativity, but more of a problem solving creativity. And not all humans should be considered intelligent anyways. On the other hand the Turing test does include these things, for a machine to pass this test it must adapt to each new individuals conversation to pass well. It must be able to react "humanly" to a wide range of personalities. Creativity can also be tested in a Turing test as well. If you ask it to make up a story it would have to do so, without a database of pre-built stories this would require creativity. This is the problem. Reacting "humanly" doesn't mean reacting intelligently. The Borg from Star Trek would be considered a very intelligent species with very un-human, very machine-like speech. Ultimately I feat other tests of intelligence not because machines will someday maybe too easily pass the Turing Test, but rather that any other test humans think of might not include all humans. I don't see why you would fear such a thing, but that's a personal emotional reaction -- not very relevant. Moreover, there are people who couldn't pass the Turing Test today I'm sure.
"The only reason we die... is because we accept it as an inevitability." -- Stewie "To enslave nuance to dogma is folly." -- Lord Hillyer

YadaYada
fool

Usergroup: Members
Joined: Jun 12, 2004
Location: in the past

Total Topics: 5
Total Posts: 316

Posted 08/15/04 - 07:18 AM:	#15
OnceFlewOver wrote: The intelligence of technology is based on compilatiion, while human intelligence unfortunately is not. Collectively, as technology becomes more and more advanced, humans' understanding stays at the same level. (500 years ago you could easily understand the everyday tools around you, while now, how many people know how a microwave works? They know how to use it, yet do they know HOW it works?) 1. Technology. How many people could explain 500 years ago HOW a hammer works? Probably none. But today we probably have over a million people who can explain the microwave to an acceptable level. 2. Intelligence has two components. One is inherited and unchanging, the other is cultural and depends on progress and education. Anyone on this board is a genius by medieval standards. Free will also comes in to play with artificial intelligence. ... Computers could choose randomly, yet it's never been proven, and that's really just probability, not choice. We do it exactly that way. The only difference is in labelling one "choice", the other "just probability".
{}{}{}{ . . . }{}{}{ . . . }{}{}{}{}{ . . . We may assume the superiority ceteris paribus of the demonstration which derives from fewer postulates ~ Aristotle

xian
Student

Usergroup: Members
Joined: Apr 25, 2004
Location: NW England

Total Topics: 1
Total Posts: 50

Posted 08/15/04 - 03:47 PM:	#16
AKG wrote: Moreover, the Turing Test doesn't test a number of important things, like creativity, the ability to adapt to new situations, etc. Why do you think that? You could easily ask the chat partner to write a poem, for instance. Or to interpret one. Or to elaborate on one. You could ask for a short essay on a complicated topic... And then ask questions about it. And that process back and forth. Sure, the Turing Test has its flaws... One being that a program could use a lookup table to react to any conversation that might come up in a given finite time... And we would certainly (?) not call such a program "intelligent"... However, since no current program comes even close to pass the Turing Test, and no program is even remotely conceivable to do so at the moment, I personally think that the Turing Test is not so bad a test after all... Of course, it would only account for specifically human intelligence, it gives us no means to define a more general concept. xian

Neutralist
Initiate

Usergroup: Members
Joined: Aug 15, 2004
Location: East Coast, USA

Total Topics: 1
Total Posts: 7

Posted 08/15/04 - 04:24 PM:	#17
A computer as of now cannot learn but simple variables handed to it. The simple test of intelligence is to ask "why" and to find the answer, even if it takes the rest of the existance of the object (and maybe beyond existance). Actions that are prerecorded cannot come to this conclusion. It must break the cycle. A neural net would have to find a way to absorb knowledge on its own, and then it becomes A.I., or then what is to stop it from becoming inorganic actual intelligence?
"Truth is one. Sages call it by different names." - Rig Veda

sponge
banned

Usergroup: Members
Joined: Jun 04, 2004

Total Topics: 17
Total Posts: 184

Posted 08/16/04 - 07:08 AM:	#18
Personally I think that AI is a geek wet dream. It presumes that humans are no more than computers, however this is extremely doubtful. We don't know how we work, how can we make a computer think ?

Reformed Nihilist
Oblong

Usergroup: Members
Joined: Jul 15, 2004
Location: The mighty fortress of the north

Total Topics: 68
Total Posts: 7921

Posted 08/16/04 - 07:24 AM:	#19
I think the biggest problem is, their is no clear and accepted definition for intelligence. By some definitions computers are already intelligence, by others they would be excluded from ever achieving it. The big question that AI brings up, is what is actual (non-atrificial) intellegence.
Nobody ever became a famous philosopher by being a champion of ecumenical hybridism Daniel Dennett Freedom Evolves

mj_cge
Agnostic

Usergroup: Sponsors
Joined: Jun 17, 2004
Location: Stockholm

Total Topics: 7
Total Posts: 172

Posted 08/16/04 - 07:53 AM:	#20
Consider the scenario that we in the future can build a complete, fully functioning human body, including arranging the brain cells and their connections so that this artificially built person starts of with a life time of memories. Is this person's intelligence artifical intelligence?
"I shall no longer tinker with the machinery of death." - US supreme court justice

Page: 1 2 3 4 5

Sorry, you don't have permission to post. Log in, or register if you haven't yet.