I think PM is alluding to or otherwise working with something derived from situation calculus. Looking at the Wikipedia entry and recalling some other similar kinds of things, you can see just how difficult it would be to get an AGI program to be able to understand different situations. Because initial understanding is essentially at the same level of difficulty as creating a useful or insightful response that means that the situation calculus was not half a solution to AGI kind of knowledge. The glass wasn't even quarter full as it turns out.
Most of the time when we generate insightful thoughts around a problem we are drawing on knowledge about many different things and much of this knowledge is beneficial even if it does not solve the main problems that we wish we could solve. Drawing on experience or the memory of experience is something that many people think requires highly sophisticated sensorimotor interactions with the world. I disagree and my disagreement leads to many implications that I have to wonder about. I think that sophisticated knowledge can be encoded into text. Then, according to this point of view, in order to answer questions (or to otherwise derive insight) about a situation the AGI program would have to be able to derive that information from its knowledge as was derived from textual interactions with the world. Much of this information would be composed of different smaller insights that had been previously derived. Some of these previously acquired insights might be refined and expressed as generalizations and so the combination of simpler insights might be *generated*, in the computational-theory sense of the term, not just mushed together individually and refined. However, this leads to certain questions which are related to some of this group's predilections. Since generalization is a kind of compression then am I only talking about distributed compressions? Well, since generalizations could be combined -by form- and -by role- then that means that I am talking about a special kind of compression in which the output could be generated without first decompressing the individuals components. Or more precisely, that means that I am talking about a special kind of generalization in which the output of the combinations of generalized components do not need to first be fully decompressed to be used. The potential in this method, which uses both old AI theories and relates directly to the potential of distributed compression methods seems obvious. But that does not mean that it is easy to figure out how to get a computer program to do something like this. Jim Bromer On Thu, May 1, 2014 at 4:15 PM, Mike Archbold via AGI <[email protected]>wrote: > On 5/1/14, Piaget Modeler via AGI <[email protected]> wrote: > > Okay, > > Now that we have a fuzzy definition of situations, what do the words > > "situation induction" mean to you? > > Please advise. > > ~PM > > > > > > Did you acquaint yourself with "situation calculus"? I think Ben > alluded to this. > Mike > > > ------------------------------------------- > > AGI > > Archives: https://www.listbox.com/member/archive/303/=now > > RSS Feed: > https://www.listbox.com/member/archive/rss/303/11943661-d9279dae > > Modify Your Subscription: > > https://www.listbox.com/member/?&; > > 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/303/24379807-f5817f28 > Modify Your Subscription: > https://www.listbox.com/member/?&; > 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
