On 12/01/2008, Richard Loosemore <[EMAIL PROTECTED]> wrote: > > Every time a dispute erupts about what the real definition of > "intelligence" is, all we really get is noise, because nobody is clear > about the role that the definition is supposed to play. > > If the role is to distinguish Narrow AI from AGI, Ben's definition is > fine. If the role is to define a class of (arbitrary) systems, any > definition whatsoever is fine so long as there is no circularity in it > (although the result will not necessarly have any relationship to the > commonsense meaning of "intelligence"). If the role is to act as a > loose organizing principle for a field of inquiry, it needs to have some > power to act as an organizing principle. > > With this in mind, Shane Legg's paper is not "the canonical reference", > it is a trivial reference, being nothing more than a naive list of > definitions collected from elsewhere, with only the shallowest > understanding of their context, relationships or roles. >
Point taken. For the record, I considered the definition I gave as *one of* the necessary preconditions for a system to be brain-like. Human brain-like systems obviously has a lot more conditions built on top of that. Feel free to ignore the definition I gave if you wish. I still think the word problem is misleading when trying to build an intelligence. The world isn't parceled in to well defined problems. It is just there, and you have to deal with it best you can. If you are offended by geek humour with a kernel of truth, please look away now If anyone had a full, precise, makeable definition of human level intelligence, all they would have to do is declare it somewhere and then call it in main(), then compile it and we would have an AI. ;) Will ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244&id_secret=85318384-4b147d
