The what is AI question
Locked
 
|
Steve, Josh and I revisited this old(?) chestnut at today's FRIAM
meeting, so I had to look it up... The American Association for Artificial Intelligence defines AI as "the scientific understanding of the mechanisms underlying thought and intelligent behavior and their embodiment in machines." John McCarthy authored an easy to read discussion on the subject at: http://www-formal.stanford.edu/jmc/whatisai/node1.html Connections to human intelligence seem very strong. Robert C. www.cirrillian.com |
|
|
I'm a little confused. If AI is the art of replicating the mechanisms
of human intelligence with machines, doesn't that assume that brain function is digital? I don?t think that's been demonstrated as yet. The sources I looked up a few months ago all pointed to the information content of synaptic pulses (other than their chemical content) being in the spacing of pulses, not their shape, size, number or pattern. With all the pulses virtually the same and the information content being contained in the length of the gaps between them, thought processes would look more like music than strings of yes/no calculations wouldn't they? If we really don't know how thought works, perhaps it be better to say that AI is the use of machines in an attempt to imitate thought? Phil Henshaw ????.?? ? `?.???? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 680 Ft. Washington Ave NY NY 10040 tel: 212-795-4844 e-mail: pfh at synapse9.com explorations: www.synapse9.com > -----Original Message----- > From: friam-bounces at redfish.com > [mailto:friam-bounces at redfish.com] On Behalf Of Robert Cordingley > Sent: Friday, December 22, 2006 9:30 PM > To: The Friday Morning Applied Complexity Coffee Group > Subject: [FRIAM] The what is AI question > > > Steve, Josh and I revisited this old(?) chestnut at today's FRIAM > meeting, so I had to look it up... > > The American Association for Artificial Intelligence defines > AI as "the > scientific understanding of the mechanisms underlying thought and > intelligent behavior and their embodiment in machines." John > McCarthy > authored an easy to read discussion on the subject at: > Connections to human intelligence seem very strong. Robert C. www.cirrillian.com ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
On Dec 24, 2006, at 4:24 PM, phil henshaw wrote: > I'm a little confused. If AI is the art of replicating the > mechanisms > of human intelligence with machines, doesn't that assume that brain > function is digital? As I understand it (from nearly three decades of hanging around the AI people) AI has one major purpose, to understand intelligent behavior. Whether that behavior is instantiated in humans (one instance of symbol-processors) or computers (a second instance) is not the point--the point is to understand what intelligence really is, or to put it another way, what the two instances of intelligence have in common so that a general scientific theory of intelligence-- any and all intelligence--can be known. From the beginning, human intelligence has been used as an example of intelligent behavior, and therefore well worth understanding. But human cognitive psychology and AI aren't identical, though they share many assumptions and techniques. In early AI efforts, "imitating" human thought ("simulating" the scientists prefer to say) was a reasonable way to begin. In fifty years, some aspects of human thought have been surpassed by computers. However, when computers think, we generally tie them to human intelligence in some way (mathematical proofs or other means of verification) because we humans need to understand what they're doing. The dance is intricate. Pamela "My idea of good company, Mr. Elliot, is the company of clever, well- informed people, who have a great deal of conversation; that is what I call good company." "You are mistaken," said he gently, "that is not good company, that is the best." Jane Austen, Persuasion -------------- next part -------------- An HTML attachment was scrubbed... URL: /pipermail/friam_redfish.com/attachments/20061224/49c2eacf/attachment.html |
|
|
That's a wonderful expression of it.
The quest for what 'intelligence' is, as an ideal independent of either the present human or computational models is nicely vague and inspiring. Perhaps the growing number of known profound disconnects from reality displayed by human 'intelligence', such as the global consensus that doubling the size and complexity of economic activity every 20 years forever is a good idea, does make it best we place 'intelligence' somewhere beyond our present reach! I like comparing humans and computers as 'symbol processors' too, which both certainly are. How both also rely on a core framework of ideas to accomplish their confined abilities in that area, and how all results are therefore direct images of that core framework, is maybe something to look at. Phil Henshaw ????.?? ? `?.???? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 680 Ft. Washington Ave NY NY 10040 tel: 212-795-4844 e-mail: pfh at synapse9.com explorations: www.synapse9.com <http://www.synapse9.com/> -----Original Message----- From: [hidden email] [mailto:[hidden email]] On Behalf Of Pamela McCorduck Sent: Sunday, December 24, 2006 12:41 PM To: The Friday Morning Applied Complexity Coffee Group Subject: Re: [FRIAM] The what is AI question On Dec 24, 2006, at 4:24 PM, phil henshaw wrote: I'm a little confused. If AI is the art of replicating the mechanisms of human intelligence with machines, doesn't that assume that brain function is digital? As I understand it (from nearly three decades of hanging around the AI people) AI has one major purpose, to understand intelligent behavior. Whether that behavior is instantiated in humans (one instance of symbol-processors) or computers (a second instance) is not the point--the point is to understand what intelligence really is, or to put it another way, what the two instances of intelligence have in common so that a general scientific theory of intelligence--any and all intelligence--can be known. >From the beginning, human intelligence has been used as an example of intelligent behavior, and therefore well worth understanding. But human cognitive psychology and AI aren't identical, though they share many assumptions and techniques. In early AI efforts, "imitating" human thought ("simulating" the scientists prefer to say) was a reasonable way to begin. In fifty years, some aspects of human thought have been surpassed by computers. However, when computers think, we generally tie them to human intelligence in some way (mathematical proofs or other means of verification) because we humans need to understand what they're doing. The dance is intricate. Pamela "My idea of good company, Mr. Elliot, is the company of clever, well-informed people, who have a great deal of conversation; that is what I call good company." "You are mistaken," said he gently, "that is not good company, that is the best." Jane Austen, Persuasion -------------- next part -------------- An HTML attachment was scrubbed... URL: /pipermail/friam_redfish.com/attachments/20061224/5bb503b5/attachment.html |
|
|
In reply to this post by Phil Henshaw-2
On 12/24/06, phil henshaw <sy at synapse9.com> wrote:
> I'm a little confused. If AI is the art of replicating the mechanisms > of human intelligence with machines, doesn't that assume that brain > function is digital? I don? think that's been demonstrated as yet. The metaphor makes sense, but the thing is, we really don't have enough there to generalize from. In practical terms, most implementations of AI tend to be very targeted. Like the techniques which emulate inference and causality are very, very different from the techniques which emulate language and grammar. (Just as an example.) What you really have is not a grand unified theory of human consciousness so much as a grab-bag of techniques that sorta work. Some techniques are effective enough to offer insight into the individual processes they emulate, but there really isn't anything consistent enough to offer general insight into intelligence itself. -- Giles Bowkett http://www.gilesgoatboy.org http://gilesbowkett.blogspot.com http://gilesgoatboy.blogspot.com |
|
|
Those interested in a "grand theory of human consciousness" will enjoy
Gerald Edelman's "Bright Air, Brilliant Fire." From a perceptive essay: "It is a grand masterpiece laid out before us of how the matter of the brain is organized and assembles itself into recursive, intertwining loops of systems of neuronal groups as it bootstraps itself into perception, primary consciousness, and higher consciousness." db ----- Original Message ----- From: "Giles Bowkett" <[hidden email]> To: "The Friday Morning Applied Complexity Coffee Group" <friam at redfish.com> Sent: Sunday, December 24, 2006 2:47 PM Subject: Re: [FRIAM] The what is AI question > On 12/24/06, phil henshaw <sy at synapse9.com> wrote: >> I'm a little confused. If AI is the art of replicating the mechanisms >> of human intelligence with machines, doesn't that assume that brain >> function is digital? I don? think that's been demonstrated as yet. > > The metaphor makes sense, but the thing is, we really don't have > enough there to generalize from. In practical terms, most > implementations of AI tend to be very targeted. Like the techniques > which emulate inference and causality are very, very different from > the techniques which emulate language and grammar. (Just as an > example.) What you really have is not a grand unified theory of human > consciousness so much as a grab-bag of techniques that sorta work. > Some techniques are effective enough to offer insight into the > individual processes they emulate, but there really isn't anything > consistent enough to offer general insight into intelligence itself. > > -- > Giles Bowkett > http://www.gilesgoatboy.org > http://gilesbowkett.blogspot.com > http://gilesgoatboy.blogspot.com > ============================================================ > FRIAM Applied Complexity Group listserv > Meets Fridays 9a-11:30 at cafe at St. John's College > lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
In reply to this post by Giles Bowkett
On Dec 24, 2006, at 2:47 PM, Giles Bowkett wrote: > On 12/24/06, phil henshaw <sy at synapse9.com> wrote: >> I'm a little confused. If AI is the art of replicating the >> mechanisms >> of human intelligence with machines, doesn't that assume that brain >> function is digital? I don? think that's been demonstrated as >> yet. > > The metaphor makes sense, but the thing is, we really don't have > enough there to generalize from. In practical terms, most > implementations of AI tend to be very targeted. Like the techniques > which emulate inference and causality are very, very different from > the techniques which emulate language and grammar. (Just as an > example.) What you really have is not a grand unified theory of human > consciousness so much as a grab-bag of techniques that sorta work. > Some techniques are effective enough to offer insight into the > individual processes they emulate, but there really isn't anything > consistent enough to offer general insight into intelligence itself. Perhaps. Newell and Simon might disagree, and say that at a certain level of abstraction, the ability to create and manipulate symbols is the sign. But I agree that AI has been targeted (to Minsky's loud regret) and we cannot yet draw from that a grand unified theory. I'm serene; physics has been at it for a lot longer, and they're having trouble with grand unified theories too. P. > "My idea of good company, Mr. Elliot, is the company of clever, well- informed people, who have a great deal of conversation; that is what I call good company." "You are mistaken," said he gently, "that is not good company, that is the best." Jane Austen, Persuasion -------------- next part -------------- An HTML attachment was scrubbed... URL: /pipermail/friam_redfish.com/attachments/20061224/6d2a6b00/attachment.html |
|
|
What if the analogy of intelligence is unexpected predictability? I can
roll a pair of dice, and that is unpredictable; but it??s not unexpected. I expect a Gaussian curve of totals. A few thousand years ago, the states of the moon were unpredictable (eclipses, elevation, and to some extent, phases). Humans consequently animated it with intelligence by calling it Luna-the moon goddess. All deities have intelligence. The same occurred with the planets, weather; and even social conditions like love and war. Only when these things became expectedly predictable did they loose their intelligence. You all remember ELIZA! At least for the first five minutes of play, the game did take on intelligence. However, after review of the actual code did the game instantly lose it mystery. Kasparov bestowed intelligence on Deep Blue, which I??m sure the programmers did not. In this sense, intelligence is not a property that external things have. It??s something that we bestow upon, or perceive in external things. Is not one of the all time greatest insults on one??s intelligence the accusation of being predictable? I suspect that any measure of intelligence will be relative to the observer??s ability to predict expected causal effects and be pleasantly surprised-not too unlike the Turing Test. Robert Howard Phoenix, Arizona _____ From: [hidden email] [mailto:[hidden email]] On Behalf Of Pamela McCorduck Sent: Sunday, December 24, 2006 3:55 PM To: The Friday Morning Applied Complexity Coffee Group Subject: Re: [FRIAM] The what is AI question On Dec 24, 2006, at 2:47 PM, Giles Bowkett wrote: On 12/24/06, phil henshaw <sy at synapse9.com> wrote: I'm a little confused. If AI is the art of replicating the mechanisms of human intelligence with machines, doesn't that assume that brain function is digital? I don?t think that's been demonstrated as yet. The metaphor makes sense, but the thing is, we really don't have enough there to generalize from. In practical terms, most implementations of AI tend to be very targeted. Like the techniques which emulate inference and causality are very, very different from the techniques which emulate language and grammar. (Just as an example.) What you really have is not a grand unified theory of human consciousness so much as a grab-bag of techniques that sorta work. Some techniques are effective enough to offer insight into the individual processes they emulate, but there really isn't anything consistent enough to offer general insight into intelligence itself. Perhaps. Newell and Simon might disagree, and say that at a certain level of abstraction, the ability to create and manipulate symbols is the sign. But I agree that AI has been targeted (to Minsky's loud regret) and we cannot yet draw from that a grand unified theory. I'm serene; physics has been at it for a lot longer, and they're having trouble with grand unified theories too. P. "My idea of good company, Mr. Elliot, is the company of clever, well- informed people, who have a great deal of conversation; that is what I call good company." "You are mistaken," said he gently, "that is not good company, that is the best." Jane Austen, Persuasion -------------- next part -------------- An HTML attachment was scrubbed... URL: /pipermail/friam_redfish.com/attachments/20061225/edf71e2f/attachment.html |
«
Return to Friam
|
1 view|%1 views
| Free forum by Nabble | Edit this page |