Steve:They programmed and trained on one corpus, and then competed on another corpus that none of the competitors had previously seen. Each corpus was carefully analyzed by hand before turning the computers loose on it, so scoring was simply a measure of how close the computers came to the hand analysis. No one came close to achieving a perfect score.This seems like a pretty good "measure" to me.
Nope. Wrong kind and culture of intelligence. Rationality. Production of the old. The attempt is to find things that are variations on a given formula. AGI isn't about that. It is about creativity. Production of the new. Can the computer - well actually it will only be a robot - draw in - and recognize - an endless range of NEW styles of handwriting? (If that's the example). You can measure the old, you can't measure the new - and there are billions of examples existing in our culture. Nowhere is creativity measured, only crudely graded. How would you measure a NEW program that s.o. has just written in terms of value/intelligence? Or a new political/economic/business "program" ? Silly question. Wherever you see people talk about measuring AGI/creativity - you will see people who simply don't *understand* AGI/creativity. Reread Deutsch - he is saying that AGI is creativity because it is still *news* within the AGI community. On 27 November 2013 18:30, Steve Richfield <[email protected]>wrote: > Michael, > > The Hobbs paper was about one system that competed with other systems to > analyze a corpus of real-world terrorist threat reports, complete with > contorted and ambiguous statements, to determine particular things. The > Hobbs system was SECOND best among the contestants. > > They programmed and trained on one corpus, and then competed on another > corpus that none of the competitors had previously seen. Each corpus was > carefully analyzed by hand before turning the computers loose on it, so > scoring was simply a measure of how close the computers came to the hand > analysis. No one came close to achieving a perfect score. > > This seems like a pretty good "measure" to me. > > Given the combination of this paper and the *60 Minutes* report about the > NSA bugging the Internet backbone, it seems inconceivable (to me) that this > posting isn't being analyzed by a similar system. > > Steve > =================== > On Wed, Nov 27, 2013 at 2:06 AM, tintner michael <[email protected] > > wrote: > >> Russ:PDF: >> http://louisville.edu/speed/computer/tr/UL_CECS_02_2011.pdf/at_download/file >> ) >> is a good read on that question. >> >> The paper concludes: >> >> "Progress in the field of artificial intelligence requires access to well >> defined problems of measurable >> complexity." >> >> ...and all AGI problems including language use and vision are >> ILL-defined, creative and not measurable as opposed to well-defined, >> rational and measurable. Think just of essays, papers and projects which >> compose well over 50% of education as distinct from IQ, SAT, knowledge >> tests and the like - they cannot be measured, only graded. >> >> Creative/AGI intelligence is a whole different world and level of >> problemsolving/intelligence from rational/narrow AI intelligence. >> High-level as opposed to low-level intelligence. >> >> (At least this paper has a few glimmers of the breadth of human >> problemsolving rather than being purely mathematical/logical). >> >> >> >> On 27 November 2013 04:05, Russ Hurlbut <[email protected]> wrote: >> >>> It is good practice to find truth in statements such as these before >>> dismissing them. This often requires adopting one or more contexts. >>> >>> In this case, if one assumes a traditional definition of "AI-complete" >>> by extending Hobbs statement to imply actually creating an artificial >>> intelligence, then anything short of AI-Complete would be fall under Hobb's >>> definition of "computer science." If one chooses to apply the dual process >>> theory ( http://en.wikipedia.org/wiki/Dual_process_theory#Systems ), >>> then one could argue that an Expert System would fit Hobbs definition of >>> fast, computer science. Conversely, the massively parallel unconscious >>> processing that humans regularly perform (e.g. in speech, vision) would >>> require enormous computing resources and considerable time - even more so >>> using resources available twenty years ago. >>> >>> Does solving syntactic ambiguity really result in creating an artificial >>> intelligence? Yampolskiy's paper AI-Complete, AI-Hard, or AI-Easy: >>> Classification of Problems in Artificial Intelligence (PDF: >>> http://louisville.edu/speed/computer/tr/UL_CECS_02_2011.pdf/at_download/file) >>> is a good read on that question. >>> >>> >>> On Tue, Nov 26, 2013 at 3:19 PM, Piaget Modeler < >>> [email protected]> wrote: >>> >>>> Hobbs' statement: >>>> >>>> >>>> *Q: What is the difference between computer science and artificial >>>> intelligence? * >>>> *A: In computer science you write programs to do quickly what people >>>> do slowly. In artificial intelligence, it is just the opposite.* >>>> >>>> In AI we don't write programs to do slowly what people do quickly. In >>>> Expert Systems in particular, once it is known what people >>>> do symbolically, an expert system often does the symbol manipulations >>>> faster >>>> that a person. Also, Expert Systems can perform >>>> those symbol manipulations 24 x 7 x 365. Thereby bringing >>>> consistency, accuracy, and endurance to the formerly human task. >>>> >>>> This statement is clearly false. >>>> >>>> ~PM >>>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >>>> <https://www.listbox.com/member/archive/rss/303/18488709-8cf25195> | >>>> Modify <https://www.listbox.com/member/?&;> Your Subscription >>>> <http://www.listbox.com> >>>> >>> >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >>> <https://www.listbox.com/member/archive/rss/303/6952829-59a2eca5> | >>> Modify <https://www.listbox.com/member/?&;> Your Subscription >>> <http://www.listbox.com> >>> >> >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >> <https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac> | >> Modify <https://www.listbox.com/member/?&;> Your Subscription >> <http://www.listbox.com> >> > > > > -- > Full employment can be had with the stoke of a pen. Simply institute a six > hour workday. That will easily create enough new jobs to bring back full > employment. > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/6952829-59a2eca5> | > Modify<https://www.listbox.com/member/?&;>Your Subscription > <http://www.listbox.com> > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
