> While the details vary widely, Mike and I were addressing the very concept > of writing code to perform functions (e.g. "thinking") that apparently > develop on their own as emergent properties, and in the process foreclosing > on many opportunities, e.g. developing in variant ways to address problems > in new paradigms. Direct programming would seem to lead to lesser rather > than greater "intelligence". Am I correct that this is indeed a central > thread in all of the different systems that you had in mind?
Different AGI systems rely on emergence to varying extents ... No one knows which brain functions rely on emergence to which extents ... we're still puzzling this out even in relatively well-understood brain regions like visual cortex. (Feedforward connections in visual cortex are sorta well understood, but feedback connections, which is where emergence might play in, are very poorly understood as yet.) For instance, the presence of a hierarchy of progressively more abstract feature detectors in visual cortex clearly does NOT emerge in a strong sense... it may emerge during fetal and early-childhood neural self-organization, but in a way that is carefully genetically preprogrammed. But, the neural structures that carry out object-recognition may well emerge as a result of complex nonlinear dynamics involving learning in both the feedback and feedforward connections... so my point is, the brain is a mix of wired-in and emergent stuff, and we don't know where the boundary lies... as with vision, similarly e.g. for language understanding. Read Jackendoff's book Jackendoff, Ray (2002). Foundations of Language: Brain, Meaning, Grammar, Evolution. and the multi-author book mitpress.mit.edu/book-home.tcl?isbn=0262050528 for thoughtful treatments of the subtle relations btw programmed-in and learned aspects of human intelligence ... much of the discussion pertains implicitly to emergence too, though they don't use that word much ... because emergence is key to learning... In the Novamente design we've made some particular choices about what to build in versus what to allow to emerge. But, for sure, the notion of emergence from complex self-organizing dynamics has been a key part of our thinking in making the design... Neural net AGI approaches tend to leave more to emerge, whereas logic based approaches tend to leave less... but that's just a broad generalization In short there is a huge spectrum of choices in the AGi field regarding what to build in versus what to allow to emerge ... not a herd mentality at all... -- Ben ------------------------------------------- agi Archives: http://www.listbox.com/member/archive/303/=now RSS Feed: http://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: http://www.listbox.com/member/?member_id=8660244&id_secret=103754539-40ed26 Powered by Listbox: http://www.listbox.com
