[email protected] said: Yes, but as a public forum, in addition to its entertainment and social value, it obliges us to uphold high standards of work and homework, and it is doubtful the most frequent posters have been keeping these high standards. And those few here who have simple solutions and simple refutations would be better off delivering them but not too often, obviously repetition after a point is not a virtue, rather a low standard. Me: The value that I get from these discussions is that every once in a while I get an idea that might be useful. I want to talk about programming as it might relate to an AGI project. I personally think that petty remarks are the low water marks in these groups, but I have to admit that I am amazed by conjectures that seem to be based on unattainable computational achievements as premises for their projects. It does not make a lot of sense. sokratis.dk: Now, can I "see" a mind without statistics (or equivalent generative code)? No, at least not until someone provides a deep and surprising explication of the normal distribution. Even if you start life with a statistics-free mind and just go out looking for resources, let's say looking for apples on apple trees, or water in water holes, you will stumble upon the normal distribution, first you walk into a couple of apple trees here and there, then into a cluster, you reach a plateau and then decline again. What could be simpler! It should come right after me Tarzan, you Jane! Me: I may be misinterpreting what you said, but it looks like you said that since you can see distributions everywhere they should be used as the basis for AGI. That kind of reasoning sounds like it might really go down with the statistics-based-AGIers but it does not seem particularly persuasive to me. sokratis.dk: Now, we do have a substantial statistical toolbox that we have not really deployed towards AGI, the main reason being its difficulty capturing generative, recursive, self-similar models: the reason the word "models" appears in these two specific locations in this paragraph is not to be explained by a Gaussian mix or any other probabilistic tool, it is all grammar and syntax. Me: I think you are saying that statistics has difficulty capturing computational syntax of generative, recursive and self-similar models. You cannot use statistics to capture everything there is about a resource (in spite of what I am guessing is your unspoken fantasy that such a thing is theoretically attainable - given the propensity of people in this group to believe that unattainable computational achievements can stand as the sound premises for their AGI conjectures). So if you actually wanted to use statistics to capture something about the generative, recursive and self-similar models you could do so if you just went ahead and did it. But, even if the result of the effort was interesting it would not make your AGI project attainable. There are many things that cannot be measured and the idea that by forcing a bunch of different kinds of measurements in with a bunch of estimates your going to get be able to use the smush to produce insight seems a little far-fetched to me. My current theory in Simple AGI is that knowledge is based on structure and reasons. Using the form of your argument I could say that, since you can see structure and reasons in everything it stands that this can be used as the basis for AGI. (What could be simpler?) Well that argument is not very strong, but the thing is that I can see structure and reasons in thought and that is why I believe that these components are so important in AGI. You cannot readily find the reasons and the structures in every system of ideas just as you can't measure every part of an idea or even determine what a reasonable domain of measurement might be for every idea. But you can find structure and reasons in thought even when you can't find statistics. Jim Bromer
Subject: Re: [agi] A Very Simple AGI Project From: [email protected] Date: Wed, 31 Jul 2013 09:41:51 +0200 To: [email protected] On 30.07.2013, at 02:18, Jim Bromer <[email protected]> wrote: I am interested in discussing AGI, not because I have it all figured out, but because I don't. You started to say: "Simple learning, statistics transcending..." Yes, but as a public forum, in addition to its entertainment and social value, it obliges us to uphold high standards of work and homework, and it is doubtful the most frequent posters have been keeping these high standards. And those few here who have simple solutions and simple refutations would be better off delivering them but not too often, obviously repetition after a point is not a virtue, rather a low standard. Now, can I "see" a mind without statistics (or equivalent generative code)? No, at least not until someone provides a deep and surprising explication of the normal distribution. Even if you start life with a statistics-free mind and just go out looking for resources, let's say looking for apples on apple trees, or water in water holes, you will stumble upon the normal distribution, first you walk into a couple of apple trees here and there, then into a cluster, you reach a plateau and then decline again. What could be simpler! It should come right after me Tarzan, you Jane! It is telling how indispensable random number generators are in most kinds of scientific code. I have previously remarked that unpredictability is probably a desirable feature in complex cognitive and social systems, while someone here kind of missed the point and replied that unpredictability is merely the result of complexity. Now, we do have a substantial statistical toolbox that we have not really deployed towards AGI, the main reason being its difficulty capturing generative, recursive, self-similar models: the reason the word "models" appears in these two specific locations in this paragraph is not to be explained by a Gaussian mix or any other probabilistic tool, it is all grammar and syntax. Bring it on! AT AGI | Archives | Modify Your Subscription ------------------------------------------- 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
