Greetings SteveYou noted that we have difficulty discussing AGI because we don't know where the “top” of this subject is. If you mean to ask where is the peak of this mountain of AGI development, then I think it is relevant to speak of the “most” difficult aspect of AGI. I have an opinion based on my efforts to get to the coding stage of an AGI unit. If I were to coin a word for this difficulty, it is “valutizing.”
If you condense the essence of an AGI into an ability to make good decisions, then it surely depends on what one values. I prefer to think of AGI in terms of units because there “moments” of time and “places” where the unit will find itself. It is in these situations that a specific unit may act, and it is the units understanding of the situation that will determine which rules are activated. How does that relate to values?
Knowing what is valuable is essential to understanding where the motivation will come from and what is likely to happen (how you write the rules.) My suggestion is that the “top” is a difficult problem of placing value on specific things at a specific moment. Attached is a “thought” problem that illustrates how difficult it is to find the “value” of a thing. The attachment is called 100 Trees.
I have more to say about getting around this problem, but I'll wait to see if anyone wants to discuss.
Stan On 01/04/2015 10:17 PM, Steve Richfield via AGI wrote:
Hi all,I was about to respond to Jim's latest thread regarding conceptual structure, but then I realized that the reason everyone here is talking at/past each other is that nearly everyone has a different idea as to where the "top" of this subject is located. First, some prospective examples:1. Some believe that the "top" is an ability to acquire, store, access, and act on information to provide a text-based interface. 2. Some believe that the "top" is an ability to self-organize to form an intelligent system. 3. Some believe that since intelligence apparently evolved from a primitive process control system, that paralleling this development might start with a better understanding of self-organizing process control systems. 4. Some believe that in the process of learning how to do MUCH better compression, that we will learn how to self-organize the process of processing intelligent communications. 5. There are almost as many of these as there are members on this forum. I could easily attach names (including my own) to the above, but I prefer to avoid having this devolve into an argument as to exactly what the various members believe.OK, so just WHERE IS the real "top"? Can a system be considered to be "intelligent" without being self-organizing? Can an approach be considered to be valid without being extensible to ALL of our functions?Myself, I think self-organization is essential, and if a system can't even self-organize to perform simplistic process control, e.g. like a hydra, then what hope is there to ever be "intelligent" (#3 above)? However, I seem to be alone in this view, yet I can't fathom how others ever expect success without these basics.I would be interested in seeing crafted replacements or additions to my above descriptions of various views of the "top", that embody SOME reasonable rationale as to how they might lead to AGI success.Can anyone shine light in this very dark corner? Thanks. Steve.*AGI* | Archives <https://www.listbox.com/member/archive/303/=now> <https://www.listbox.com/member/archive/rss/303/9320387-ea529a81> | Modify <https://www.listbox.com/member/?&;> Your Subscription [Powered by Listbox] <http://www.listbox.com>
------------------------------------------- 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
One Hundred Trees.odt
Description: application/vnd.oasis.opendocument.text
