Richard:
>> [...]
>
> This is too simple by a long way. I can design a KR easily enough, but
> KRs are not libraries, they are used by something. The "using part" is
> what counts: it might take five days to design the KR, five (or fifty)
> years to design and build the system that makes use of the KR.
> KRs are not libraries, they are used by something. The "using part" is
> what counts: it might take five days to design the KR, five (or fifty)
> years to design and build the system that makes use of the KR.
This is the revised agenda:
1. design an AGI architecture (I've proposed one)
2. design a KR
3. design algorithms for learning, inference, query-answering, memory management etc
4. fill the AGI with knowledge
5. design behavior generation algorithms
As I said, I'm trying to focus on 1-4 first, ie building a general passive knowledge maintenance system.
Re your criticism:
1) I don't think an AGI-adequate KR can be designed in 5 days. It's a nontrivial problem.
2) What you call the "using part" would be a bunch of algorithms. You seem to think there's a myriad of complex mechanisms to be programmed for 50 years, but I think that's an overestimation. I think all AGI functions (as far as passive AGI is concerned) can be reduced to a handful of core algorithms (learning / inference / query-answering / memory management etc) and require ~5 years.
You proposal for a software development environment for AGI is interesting, especially that we should strive to establish a common KR standard. After that, the task is to design the various algorithms and to fill the AGI with knowledge.
YKY
To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
