Re: [agi] The Missing Piece

Hi John, Re your idea that there should be an "intermediate-level" representation: 1. Obviously, we do not currently know how the brain stores that representation. Things get insanely complex as neuroscientists go higher up the visual pathways from the primary visual cortex. 2. I advocate us

Re: [agi] The Missing Piece

On 3/8/07, Matt Mahoney <[EMAIL PROTECTED]> wrote: [re: logical abduction for interpretation of natural language] One disadvantage of this approach is that you have to hand code lots of language knowledge. They don't seem to have solved the problem of acquiring such knowledge from training da

[agi] My proposal for an AGI agenda

This is the agenda for an *integrated* AGI system (vs Minsky's distributed one): 1. Fix a knowledge representation scheme (eg CycL, or Novamentese? etc) 2. Work out an uncertain logic (ie some form of logic + probability / fuzziness). 3. Develop an *efficient* deductive algorithm for said lo

Re: [agi] My proposal for an AGI agenda

On 3/11/07, YKY (Yan King Yin) <[EMAIL PROTECTED]> wrote: This is the agenda for an *integrated* AGI system (vs Minsky's distributed one): 1. Fix a knowledge representation scheme (eg CycL, or Novamentese? etc) 2. Work out an uncertain logic (ie some form of logic + probability / fuzziness).

Re: [agi] My proposal for an AGI agenda

On 3/11/07, Jey Kottalam <[EMAIL PROTECTED]> wrote: I'm stuck at steps 1 and 2. Even if we assume that the remaining steps can be implemented once a knowledge representation is chosen, how do I evaluate and judge a knowledge representation scheme's appropriateness for AGI? All the steps are int

Re: [agi] The Missing Piece

YKY (Yan King Yin) wrote: Hi John, Re your idea that there should be an "intermediate-level" representation: 1. Obviously, we do not currently know how the brain stores that representation. Things get insanely complex as neuroscientists go higher up the visual pathways from the primary v

Re: [agi] My proposal for an AGI agenda

YKY (Yan King Yin) wrote: This is the agenda for an *integrated* AGI system (vs Minsky's distributed one): 1. Fix a knowledge representation scheme (eg CycL, or Novamentese? etc) 2. Work out an uncertain logic (ie some form of logic + probability / fuzziness). 3. Develop an *efficient

Re: [agi] general weak ai

On Saturday 10 March 2007 14:36, Andrew Babian wrote: > I can't speak for Minsky, but I would wonder what advantage would there be > for having only one agent? An arbitrator. You have only one body, and it would be counterproductive for it to try to do different things at the same time. (It's c

Re: [agi] My proposal for an AGI agenda

On 3/11/07, Ben Goertzel <[EMAIL PROTECTED]> wrote: All this is perfectly useful stuff, but IMO is not in itself sufficient for an AGI design. The basic problem is that there are many tasks important for intelligence, for which is it apparently not possible to create adequately efficient deducti

Re: [agi] My proposal for an AGI agenda

YKY (Yan King Yin) wrote: On 3/11/07, Ben Goertzel <[EMAIL PROTECTED] > wrote: > All this is perfectly useful stuff, but IMO is not in itself sufficient > for an AGI design. The basic problem is that there are many tasks > important for intelligence, for which is it ap

Re: [agi] My proposal for an AGI agenda

On Sunday 11 March 2007 06:58, YKY (Yan King Yin) wrote: > 1. Fix a knowledge representation scheme (eg CycL, or Novamentese? etc) The main problem with this is that it seems to assume that there is One True Knowledge Representation in the system. In the automatic microprocessor design stuff I

Re: [agi] My proposal for an AGI agenda

3. Give an example of a task where logical inference is inefficient? ;) I mentioned your question to my wife and she responded: "Well, how about mathematical theorem-proving?" ;-) Quite apropos... In fact we may refine the retort as: "Well, how about proving nontrivial theorems in MAT

Re: [agi] My proposal for an AGI agenda

On 3/12/07, Ben Goertzel <[EMAIL PROTECTED]> wrote: "Natural concepts" in the mind are ones for which inductively learned feature-combination-based classifiers and logical classifiers give roughly the same answers... 1. The feature-combination-based classifiers CAN be encoded in the probabilist

Re: [agi] My proposal for an AGI agenda

YKY (Yan King Yin) wrote: On 3/12/07, Ben Goertzel <[EMAIL PROTECTED] > wrote: > "Natural concepts" in the mind are ones for which inductively learned > feature-combination-based classifiers and logical classifiers give > roughly the same answers... 1. The feature-com

Re: [agi] My proposal for an AGI agenda

Hi Josh, You touched on a lot of deep issues; I'll give a *tentative* answer here. Let's see what happens... My main point is: a unified KR allows people to *work together*. On 3/12/07, J. Storrs Hall, PhD. <[EMAIL PROTECTED]> wrote: > 1. Fix a knowledge representation scheme (eg CycL, or N

Re: [agi] My proposal for an AGI agenda

On 3/12/07, Ben Goertzel <[EMAIL PROTECTED]> wrote: Yeah, and what you will find is that these "more efficient algorithms" are more efficient only if you let them work with non-logical knowledge representations ;-p In NM, we do in fact have a unified logical representation -- but we also have a

Re: [agi] My proposal for an AGI agenda

On 3/11/07, J. Storrs Hall, PhD. <[EMAIL PROTECTED]> wrote: The main problem with this is that it seems to assume that there is One True Knowledge Representation in the system. In the automatic microprocessor design stuff I was doing in the 90s, there were up to 50 different levels of representa

[agi] Logical representation

On 3/11/07, Ben Goertzel <[EMAIL PROTECTED]> wrote: YES -- anything can be represented in logic. The question is whether this is a useful representational style, in the sense that it matches up with effective learning algorithms!!! In some domains it is, in others not. "Represented in logic

Re: [agi] My proposal for an AGI agenda

The 50 got reached by the Aquarius project (Berkeley and later ISI) which ran the gamut from Prolog through optimizing Prolog compilers to Prolog hardware to a fairly hairy 2-level shared-memory multiprocessor system. My own system was on the simple end, with a fairly vanilla instruction set spe

[agi] Off-topic: The Shtinkularity

Hi all, If you have 2.5 minutes or so to spare, my 13-year-old son Zebulon has made another Singularity-focused mini-movie: http://www.zebradillo.com/AnimPages/The%20Shtinkularity.html This one is not as deep as RoboTurtle II, his 14-minute Singularity-meets-Elvis epic from a year ago or so .

Re: [agi] My proposal for an AGI agenda

On Sunday 11 March 2007 15:07, YKY (Yan King Yin) wrote: > My main point is: a unified KR allows people to *work together*. That would certainly be nice, but I have yet to be convinced that it's possible :-) > Let's look at the alternative, which is even more dismal: you have many > represent