Loosemore used to talk about simple Complex Adaptive Systems (CAS). The step by step actions of a simple CAS might be described in mathematical terms but there was no way to use mathematics to effectively fast forward those steps (or to solve the more interesting questions). The search problem in AI/AGI is similar. The search will find information that needs to be compared to the situation at hand (for example) but then the continuing search will often be dependent on those previous comparisons. This means that the effectiveness of mathematics will often be blunted by AI problems because you can't use the math to derive highly effective search procedures that gets to the problem quickly. My view here is that Information Theoretic primitives and Generative primitives are probably not powerful enough to make the search problem efficient enough. However, there may be more sophisticated generative methods which might be designed to produce more effective searches. Grammatical Generators are not really designed to compress data precisely but to encode and decode relational information about the objects in the grammar using a standardized system. It seems like this does make for some effective compressions but they are not direct compressions. A grammatical string does not decompress into an encyclopedic store of knowledge related to the subject. On the other hand a well written article can provide a reader who is prepared to understand the article with a lot of information. I contend that there must be some more sophisticated mechanisms in the grammar which are computational in action but which are composed of something more sophisticated than what I am calling generative grammar primitives. I realize that this would be a lot more interesting with some simple applications that demonstrated my point but it is what it is. Jim Bromer
On Wed, Nov 12, 2014 at 5:04 PM, Mike Archbold via AGI <[email protected]> wrote: > I respect Ben's work quite a bit, but I have to say I have never > understood the connection between mathematics and intelligence. > Mathematics does seem to represent the best and most successful form > of Platonism, or more generally, some type of certainty which exists > in some realm that we can count with seemingly rock solid certainty. > > So we can use it to compute with. I read with great interest AGI > approaches based on geometry, as one paper posted recently here > described. Still, I admit to having significant doubts as to how you > get from geometric forms to thinking, and what justification there is > for such an approach. Some of us, not me, are really born > Pythagoreans -- that somehow everything boils down to numbers in the > end. > > Mike A > > On 11/11/14, Ben Goertzel via AGI <[email protected]> wrote: >> Jim, >> >> Hmmm... really, the blog post to which I linked in the email starting >> this thread, is not very mathematical at all; it's more psychological >> in nature really.... In that post, I'm more trying to show that >> mind-related math can be rooted in psych, rather than vice versa... >> >> But yeah, I do think that the mind's pattern formation and recognition >> processes can be formalized mathematically in terms of a fairly >> elementary-looking framework. Though there is a lot of >> specialization within this framework in the brain or in a >> very-finite-resources AGI system like OpenCog, which creates a lot of >> implementation complexity... >> >> -- Ben G >> >> On Tue, Nov 11, 2014 at 9:37 PM, Jim Bromer <[email protected]> wrote: >>> You persist in your delusion that you could express your ideas >>> mathematically and that they would constitute some kind of innovation >>> for AGI. The problem is that information theory, algorithmic >>> information theory and, lets call it, Goertzel's pattern information >>> theory are all primitives. Even though these primitives can hold (or >>> represent) more than one referent they are primitive forms that are >>> just too unsophisticated for any significant growth in intelligence. >>> (That is I don't think they would be able to gain enough traction to >>> be used to grow intelligence because the application of multiple >>> instances of information primitives inevitably lead to lossy and noisy >>> implementations.) I am leaving this comment as a primitive criticism >>> because I don't think you actually understand what I am trying to get >>> at. The complex patterns that could be used to generate true >>> intelligence may not be manifestations of these kinds of primitives >>> because the generation of multiple patterns using these kinds of >>> primitives may be effective or efficient. So, even if you guys realize >>> that you have to achieve some higher insight you inevitably end up >>> referring to the primitive forms (of information theoretic based AI >>> conjectures) as if they are somehow going to end up becoming more >>> sophisticated. >>> Jim Bromer >>> >>> >>> On Sat, Nov 1, 2014 at 7:40 AM, Ben Goertzel via AGI <[email protected]> >>> wrote: >>>> Some fairly analytic-philosophical thoughts on the underpinnings of >>>> intelligence, >>>> >>>> http://multiverseaccordingtoben.blogspot.hk/2014/11/grounding-representation-and-pattern-in.html >>>> >>>> -- >>>> Ben Goertzel, PhD >>>> http://goertzel.org >>>> >>>> "The reasonable man adapts himself to the world: the unreasonable one >>>> persists in trying to adapt the world to himself. Therefore all >>>> progress depends on the unreasonable man." -- George Bernard Shaw >>>> >>>> >>>> ------------------------------------------- >>>> AGI >>>> Archives: https://www.listbox.com/member/archive/303/=now >>>> RSS Feed: >>>> https://www.listbox.com/member/archive/rss/303/24379807-653794b5 >>>> Modify Your Subscription: https://www.listbox.com/member/?&; >>>> Powered by Listbox: http://www.listbox.com >> >> >> >> -- >> Ben Goertzel, PhD >> http://goertzel.org >> >> "The reasonable man adapts himself to the world: the unreasonable one >> persists in trying to adapt the world to himself. Therefore all >> progress depends on the unreasonable man." -- George Bernard Shaw >> >> >> ------------------------------------------- >> AGI >> Archives: https://www.listbox.com/member/archive/303/=now >> RSS Feed: https://www.listbox.com/member/archive/rss/303/11943661-d9279dae >> Modify Your Subscription: >> https://www.listbox.com/member/?&; >> Powered by Listbox: http://www.listbox.com >> > > > ------------------------------------------- > AGI > Archives: https://www.listbox.com/member/archive/303/=now > RSS Feed: https://www.listbox.com/member/archive/rss/303/24379807-653794b5 > Modify Your Subscription: https://www.listbox.com/member/?&; > Powered by Listbox: 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
