Path: csiph.com!newsfeed.hal-mli.net!feeder3.hal-mli.net!newsfeed.hal-mli.net!feeder1.hal-mli.net!npeer01.iad.highwinds-media.com!news.highwinds-media.com!feed-me.highwinds-media.com!post01.iad.highwinds-media.com!newsfe14.iad.POSTED!00000000!not-for-mail From: seeWebInstead@rem.intarweb.org (Robert Maas, http://tinyurl.com/uh3t) Errors-To: ErrorsToHere@YahooGroups.Com X-Spam-This: SpamCopies@YahooGroups.Com X-Twitter: CalRobert Newsgroups: comp.ai.philosophy,comp.programming Subject: Re: Towards true A.I. References: <1jwvj1x0ayc05.ytcz87k1p1x4.dlg@40tude.net> <1qwcrz02cn5hx.1sj169ai6zp0r$.dlg@40tude.net> <4FC623E4.5C69FE90@bytecraft.com> Message-ID: Lines: 42 X-Complaints-To: abuse@rawbandwidth.com NNTP-Posting-Date: Mon, 18 Jun 2012 18:09:49 UTC Date: Mon, 18 Jun 2012 11:08:53 -0700 X-Received-Bytes: 2874 Xref: csiph.com comp.ai.philosophy:4613 comp.programming:1812 > From: Walter Banks > Your tiny url is broken or it may be as intended and I don't have > enough intelligence or information to know the difference. I use lots of tinyURLs. Which one is giving you trouble? > I have done quite a bit of AI over the years. The most > important comment anyone has ever made to me about > AI is. We have spent so much effort parsing external > image sources (text, speech, image) and so little > effort in the extraction of information. I think I agree with you, but it would be helpful (to this discussion) if you give 3-5 examples of a situation as follows: - Overall general situation - Specific data being observed with respect to that situation - How far the A.I. work has gone in parsing the data - What next step is missing, what *should* be done next, what actual information should be gleaned from the parse-tree that was already computed. > Some of the current AI successes have been brute force. > There is a project at the University of Waterloo on speech > response systems that basically does a lot of brute force > pattern matching in parallel and decides from degree of > matches and context the most likely meaning and responds > appropriately. That sounds vaguely like the high level part of the methodology used by the moderately-successful "Watson" system used to play "Jeopardy" (TV show), with different low-level pattern matching tools due to different type of data being analyzed. Based on what I've learned about recent research in natural intelligence, such as reported on the "Brain" series of "Charlie Rose" and various reports on other science programs, I'm leaning toward believing that part of natural intelligence actually works that way, with various neurons competing for attention, with their signal amplitudes proportional to their confidences in their respective proposals, such that most confident answer wins the debate and is passed on to the next level of decision making. Google-groups-search-key: imtrgfdi