Is AKC for real? I don't know. COULD it be a computer program using various aliases? I am 99% sure it could be; and that the 17th. Machine-Learning Conference to be held in a few days will COVER IT UP. Indeed they won't even ask the question as I have done. If Internet discussion becomes the medium for the best minds in the citizenry of our democracies, what happens when even better artificial minds are pitted against them? We become slaves without even knowing it. FWP ---------- Forwarded message ---------- Date: Mon, 26 Jun 2000 17:02:23 -0700 (PDT) From: Franklin Wayne Poley <[EMAIL PROTECTED]> Reply-To: [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: [BCPolitics] [EDTV-Robotics-State-Of-The-Art] [ML] A Challenge for ICML-2000 But AKC, Jones-Jose-Catweasle-Barry-et-al is way ahead of us! Now what does that do to the "body politic" of the near future when the Constitution of Canada is written and rewritten on blank computer screens? Who can compete in an Internet political debate with a state-of-the-art computer? FWP ---------- Forwarded message ---------- Date: Mon, 26 Jun 2000 14:40:05 -0700 (PDT) From: Franklin Wayne Poley <[EMAIL PROTECTED]> Reply-To: [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: [EDTV-Robotics-State-Of-The-Art] [ML] A Challenge for ICML-2000 (fwd) James Gluck IBM Dear Mr. Gluck: If the Human Genome Project can decipher 3.2 billion pieces of code, I have to think that it is reasonable to undertake a similar scale mega-project to decipher the English vernacular. That is what I had called IBM about a couple of years ago when I first completed a draft of "Machine Psychology". I think it would take a company as large as IBM to even get the project started by declaring that it can be done and that an estimate of the time/funding required would be ? Sincerely-FWP ------------------------------------------------------------------------------- Machine Psychology: http://www.atoma.f2s.com/atomareport.html (file #10) ------------------------------------------------------------------------------- ---------- Forwarded message ---------- Date: Mon, 26 Jun 2000 14:12:35 -0700 (PDT) From: Franklin Wayne Poley <[EMAIL PROTECTED]> To: [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Subject: [ML] A Challenge for ICML-2000 > ---------- Forwarded message ---------- > Date: Fri, 23 Jun 2000 16:19:52 -0700 (PDT) > From: Kiri Wagstaff <[EMAIL PROTECTED]> > To: Franklin Wayne Poley <[EMAIL PROTECTED]> > Cc: [EMAIL PROTECTED] > Subject: Re: [ML] NLP 1/2 > > > Actually, Tom, what drives my questions is not understanding = getting it > > but contemplating that all those experts in natural language out there > > (educators, psychologists, linguists etc.) might already "have it". They > > might already KNOW all the rules for first rate conversational ability (if > > they pool their knowledge)...and frankly my opinion right now is that they > > probably already DO 'have it'. I will hold to that opinion until I see a > > detailed example of a situation/scenario/set of circumstances where the > > author is saying...we can't figure out the rules here. > > I have been following this conversation with interest. Frank, your > question (as you explain it) implies the assumption that the NLP > (Natural Language Processing, which often includes NLU and NLG) > problem *can* be solved by a set of rules. Yes it does. (BTW, what is NLG? I take it NLU is natural language understanding). My training and experience tell me language is orderly, not chaotic. I think those are the only two choices we have: orderly or chaotic; and a probabilistic/statistical model is not chaotic as statistics has rules too. I repeat what I posted earlier which is that I would like to see the detailed presentation of a circumstance in which humans use words yet the word usage has no rules, ie it is chaotic. I would like to see it presented well enough that it would be accepted in a peer reviewed journal (eg one of the phil/psych journals now online). I think it would be an excellent challenge to the educators, philosophers, psychologists and linguists who spend life long careers trying to find the orderliness, ie rules, in word usage. Until I see such a presentation I will assume that all word usage has rules and I will go one further and say I assume that the global community of experts could pool its talents now and SET OUT ALL THE RULES FOR WORD USAGE IN HUMANS. This is unlikely to be the > case. You can get pretty far with a set of rules (see the Eliza > program for a very simple example), but it would appear that there is > a point beyond which rules aren't going to be sufficient. This is > where the need for knowledge about the topic of the conversation comes > in, but the problem goes beyond having a database to access: in order > to have "human equivalent" language abilities, the computer needs some > way to comprehend human common sense, emotion, and humor (among other > things) - all of which are big challenges to any system currently > available and almost certainly not solvable using a set of hand-coded > rules. In addition, there is a need for non-monotonic reasoning and > the ability to maintain multiple (possibly contradictory) working > interpretations of what other people are saying (in order to > participate in a conversation). Most of the many psychologists I studied with and worked with over almost 4 decades would agree with Watson in his 1913 Psych Review paper when he said, "The time seems to have come when psychology must now discard all reference to consciousness; when it need no longer delude itself into thinking it is making mental states the object of observation." Some would agree wholly; some with qualification. Some will agree but not call themselves "behaviorists". But these issues are best left to psychologists. The main point for now is that many thousands of psychologists and psychology-trained professionals are available to ferret out the rules of "common sense". We take confused and bewildering situations in which words are used and we find out what the rules are. Let me give just one example. When I worked in the mental hospital I had a patient who engaged in what is called "ritualistic behavior". He would walk up to hot air registers and similar devices, bob and weave and utter some incantations (words). The layman might just say, "That guy is crazy" and let it go at that. The layman doesn't even ask if it is orderly or has rules. The psychologist discovers that there are rules for this ritual. And they could be programmed into a robot with bipedal locomotion if one had any reason for doing so. There were only so many words and word combinations used in the incantation. And the stimulus conditions were known re when the incantation would be given, ie hot air registers etc. A robot could be programmed to emit the words at the correct time and place. This robot would meet the criterion of "human equivalency" for this particular human patient for a limited part of the patient's repertoire of behavior. > Essentially, human language would appear to be so tightly integrated > with our cognition that to properly simulate it, you would also need > to simulate the cognition itself. In other words, language isn't a > skill you can divorce from thought. Again, I think we can have faith or belief that a mentalistic notion like "thought" (or common sense) exists. But that does not detract from the ability of that huge army of professionals who deal with the rules for word usage to set out a COMPREHENSIVE RULE BOOK FOR HUMAN WORD USAGE. The hard-line behaviorist (which I am not) would say there is no such thing as "thought" as separable from the word strings which use the word "thought". While I personally subscribe to subjectivity as real, I have been very impressed by the ability of all those psychologists to ferret out the rules whenever words are used so I think they should be given a chance to work on a mega-project which would TEACH MACHINES TO TALK WITH HUMAN PROFICIENCY. The teaching component of this might be nothing more than the immediate learning of having the program installed after it is written. Like Johnny Mnenomic, the robot just says "Hit me" and rapidly the knowledge is installed or downloaded. Some might object to calling this "learning" but that is another issue. I appreciate the effort of those who have contributed to this thread on Machine-Learning. But I think some computer people are going over territory which psychologists went over thoroughly almost a century ago. I see two fallacies in the arguments presented so far against the prospect of drafting such a comprehensive rule book for human word usage now. (1) Let's call it the "fallacy of assuming no discoverable rules". IMO all cases of word usage have discoverable rules because I have seen so many instances in which these rules are discovered after some effort. Conversely I have never seen a well presented scenario (presented well enough to be published in a peer-reviewed journal) in which word usage is chaotic and has no rules. (2) Let's call it the "fallacy of the ghost in the machine". It is assumed that there are mentalistic entities like "common sense" which ineffably guide human word usage and will forever prevent us from discovering the rules of word usage. Once again, I have never seen an example in a peer reviewed journal which would give the experts a "go at" finding a set of rules. Behavioristic psychology is said by some to have "exorcised the ghost in the machine". At one time it was widely believed that "demons" were the cause of unusual or aberrant word usage. I think the AI/machine learning field still has too many believers in demons. For this reason, NLP is sometimes > referred to as "AI-complete", meaning that if you could solve the > natural language problem, you would also be able to solve any other > problem in AI. That is a nice turn of a phrase. At first I was inclined to say, "It is AI-complete, depending on what you call AI". Why? Because the first thing that came to mind was the problem of a general vision system or general sensory system in machines. If such a system is outside NLP, how can NLP be AI-complete? But the example of the patient with his rituals says otherwise. In simulating his behavior, a robot with NLP would also have to have simulated sensory abilities integrated with the NLP. Thus it would be able to respond with the correct words to the correct stimuli (in at least two modalities; ie respond to heat and visual stimuli). So I would agree that NLP which meets the criteria of the best of the human babblers would also be "AI-complete". All the more reason I would hope that ICML-2000 would focus on this issue: What kind of mega-project would it take to have the appropriate experts set out all the rules for human word usage in one human language, like English, convert them to code and run the program in a computer? In particular, how many worker hours would this take and what would the costs be? Sincerely-FWP <A HREF="http://www.ctrl.org/">www.ctrl.org</A> DECLARATION & DISCLAIMER ========== CTRL is a discussion & informational exchange list. Proselytizing propagandic screeds are unwelcomed. Substance—not soap-boxing—please! These are sordid matters and 'conspiracy theory'—with its many half-truths, mis- directions and outright frauds—is used politically by different groups with major and minor effects spread throughout the spectrum of time and thought. That being said, CTRLgives no endorsement to the validity of posts, and always suggests to readers; be wary of what you read. CTRL gives no credence to Holocaust denial and nazi's need not apply. Let us please be civil and as always, Caveat Lector. ======================================================================== Archives Available at: http://home.ease.lsoft.com/archives/CTRL.html <A HREF="http://home.ease.lsoft.com/archives/ctrl.html">Archives of [EMAIL PROTECTED]</A> http:[EMAIL PROTECTED]/ <A HREF="http:[EMAIL PROTECTED]/">ctrl</A> ======================================================================== To subscribe to Conspiracy Theory Research List[CTRL] send email: SUBSCRIBE CTRL [to:] [EMAIL PROTECTED] To UNsubscribe to Conspiracy Theory Research List[CTRL] send email: SIGNOFF CTRL [to:] [EMAIL PROTECTED] Om