What kinds of reasoning about the physical world are you aiming at? Simple predictions? Motion of objects? The beautiful thing about the physical world is that all the information is there, we just have to read it? We don't have to represent the whole thing or run an entire physical simulation in our minds. Abstraction of unimportant details is key. You can escape the frame problem by not trying to model reality in its entirety. Use reality as a databaselook things up when needed, only track what you're interested in. Use open and close world assumptionsadvantageously. ~PM
From: [email protected] Subject: Re: [agi] Re: Could Brain Emulation be NP-Hard? Date: Thu, 25 Jun 2015 08:45:18 -0700 To: [email protected] On Jun 24, 2015, at 7:36 PM, Piaget Modeler <[email protected]> wrote:Why is this at all important? The short version: if you cannot solve these computer science problems then it is not possible to reason about the physical world at a scale that matters. Given that AGI requires the ability to reason about relationships in the physical world, you can prove that most proposed AGI architectures are not able to tractably express physical world relationships. You can invalidate and exclude AGI designs at a computer science level yet another way. This computer science has nothing to do with AI. In fact, it was discovered during the (failed) design of analytical database engines for Google Earth circa 2006. Up to that point, everyone assumed reasoning about the physical world was just very difficult. We proved it was actually impossible within the computer science at the time. It is enormously helpful for designing an AGI. The easiest way to design any novel software architecture is to exclude as much of the design phase space as possible. AGI | Archives | Modify Your Subscription ------------------------------------------- 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
