> On Jun 24, 2015, at 7:30 PM, Jim Bromer <[email protected]> wrote:
> 
> I really still do not understand what you are getting at. Are you
> talking about 3-D cubes? Do you mean that there is some cool formula
> that uses some simple representation of the cube or something. I
> really cannot see what you are getting at.


Yes, I limited it to fixed dimension cubes to make it as simple as possible.

If you do not have the computer science to solve that problem then you also do 
not have the computer science to solve AGI. The general solution to this 
problem is necessary, but not sufficient, to reason about the physical world. 
This is useful because you can prove the capabilities of an AGI design without 
proving that the design is actually AGI.

This gets to a big problem with AGI: how do you invalidate someone’s proposed 
AGI design given that it is virtually impossible for anyone to agree on what an 
AGI architecture needs to look like? The popular definition of AGI is so fuzzy 
that design propositions are not falsifiable. So everyone claims their ideas 
for AGI are the One True Way.


Whether or not a design is AGI, you can show concrete things about the computer 
science expressiveness of the design as a generic piece of software. There is a 
laundry list of basic capabilities that most people can agree must be tractably 
expressible in AGI even if they can’t agree on the definition of AGI. Any 
discussion of e.g. graph cutting algorithms does not have much room for magical 
thinking. You propose a design, I immediately go through a list of computer 
science problems that restrict capabilities and expressiveness to see which 
ones apply. It is hard to ignore the basic computer science problems even if 
you are in love with your design.

I do the same thing for software that has nothing to do with AGI. It is the 
fastest way to get at the heart of the matter without attacking the larger 
claims.

-------------------------------------------
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

Reply via email to