I said I was going to say something about the problems I have with the terms Jim Bromer made up, procedural knowledge and procedural inference. I kind of see where the confusion came from. There is a perfectly reasonable concept that we have procedural _memory_, which contrasts with declarative memory. And in a vague sense, you might call that knowledge. But the term "knowledge" is better reserved for things that can be true or false, and procedures cannot be true or false. We might evaluate them in some way, but not with truth.
Also, there is a notion of implicit knowledge that is incorporated into procedural memory. It's kind of how where computer program acts as if it knew something. But I have a problem with that, because it is the programmers who had the knowledge usually, not the computer. That's what makes it implicit knowledge. Flor a computer to be intelligent, it needs to use knowledge it has acquired and not rely on the programmer to have the knowledge and figure things out. And that's generally a problem I see with AI--so much of the figuring out is done in advance by the people involved. AI people have a kind of fudge they use, and say, well, the computer acts as if it's intelligent and that's good enough to call it intelligent. Maybe that's fine for some people, but I don't buy it. I'm not particularly sure what Jim even means by "procedural inference", but I'm pretty sure it's not inference in the technical sense. I'm not super OK with misusing technical words. I'm guessing he just means procedures triggered by some operation. That would hardly be inference. The whole idea of procedures being triggered by processing of declarative knowledge is fin in itself, but it raises questions. Can the procedures be learned or modified by the computer? Is Jim going to hand-code all the procedures? Will the computer have a way to evaluate the results, and if so, what will that be? Maybe he has a plan for that, but that's just a beginning step, not a developed plan. andi > Jim whinged: > But at least you are starting to talk about something. However, arguments > based on authority (Hofstadter says) and arguments based on popular > opinion (you are not on the same page as us) are two of Aristotle's > fallacies of reasoning. If you are unable to explain why Hofstadter's > position on analogy means that text-based AGI is not feasible then perhaps > you might take the time to find some quotes from Hofstadter to support > your opinion. > > me (andi): > I am not going to explain Hofstadter. If you want to know more, listen to > his talks, read his books, or remain in ignorance. I was not trying to > appeal to authority, just trying to point you to a source I find valuable. > > You have invented two terms I find suspect: "procedure knowledge" and > "procedure inference". When I get back to a keyboard and don't have to > type on my phone, I might explain my problems with them in more detail, > but for now, I will say I don't think they are sufficiently defined to be > used and I will concede that I don't understand what you mean by them. > Andi ------------------------------------------- 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
