Re: [agi] Progress in the way Godel's work impacts AGI

Somehow I'm not sure if this is relevant in any way, shape or form. I certainly can't interpret it. On Thu, 10 Jan 2019 at 00:46, Colin Hales wrote: > A little bit of progress. > > Machine learning leads to unsolvable problem >

[agi] Progress in the way Godel's work impacts AGI

A little bit of progress. Machine learning leads to unsolvable problem A relatively simple machine-learning problem has led researchers to a question that is mathematically unanswerable

Re: [agi] AIRIS - causality based AGI project

As I said, very interesting. So you use Python? Are there learning algorithms that could be exported as a standalone code snippet? Always looking for good learning algoithms. On Wed, 9 Jan 2019 at 23:07, wrote: > Good question. Knowledge is currently stored as follows: > > Action > -Input da

Re: [agi] AIRIS - causality based AGI project

Good question. Knowledge is currently stored as follows: Action     -Input data prior to action     -Changes that occured in the inputs after action Which then effectively gets used to fill in: IF action && input state THEN expected change. A rough raw data example would be: 'Move Up'[     [2,

Re: [agi] AIRIS - causality based AGI project

OK, my question to clarify the question would be this: How are the rules that your system learns stored internally? My home page is BotCompany.de, the language is called JavaX , there is a bunch of videos and I can run rules like these

Re: [agi] AIRIS - causality based AGI project

Thanks Stefan, Yeah, I believe Ben Goertzel started this group. At least, it was his website that directed me here. Do you have a link to info on your project? I'd love to check it out so we can compare approaches. I hear you about labels... I've been trying to teach myself all kinds of termi

Re: [agi] AIRIS - causality based AGI project

Very interesting. What's "Ben's list"? This one? I'm not 100% sure after seeing the video and the blog post, but... maybe your approach is comparable to mine? I want to have humans and AI collaborate in order to produce and test natural-language rules. So that would be "GOFAI + learning" if you w

[agi] AIRIS - causality based AGI project

Hello everyone! For the last few years I've been working on a causality focused approach to AGI that I call Autonomous Intelligent Reinforcement Interpreted Symbolism (AIRIS). It's based on the idea of creating a GOFAI that learns the "IF > THEN" rules of a given environment through observation an