YKY: I've talked to John Weng many times before, and I found that his AGI has some problems but he wasn't very eager to talk about them.
MT: Is anyone in AGI eager to talk about their problems? My impression is it's a universal failing. YKY: For example, it could only recognize pre-trained objects (eg, a certain doll) but not "general" object classes like "dolls", "cups" or "cars". Another problem is that IDHR cannot recognize relations between objects such as: on(apple,table) on(pillow,bed) etc. This second problem, I have pointed out many times before but nobody seems to register it. MT: "On" is what I would call a "superconcept". Others are verbs like "go [to/ towards]" "make," "do," "give," "take," "have.." They're superconcepts because unlike your object classes, they embrace infinitely open-ended classes of action or relationship, in which the members have no obvious physical form that, however loosely, unites them. So it's asking a great deal of any AGI to understand such superconcepts. An infant can do it, but that's another matter. John Weng says on his site: "The concept of a developmental program does not mean just to make machines grow from small to big and from simple to complex. It must enable the machine to learn new tasks that a human programmer does not know about at the time of programming." Have his robots even come close to learning new tasks? ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=231415&user_secret=e9e40a7e
