Paul:It is my understanding that the basic problem in Friendly AI is that it is possible for the AI to interpret the command "help humanity" etc wrong, and then destroy humanity (what we don't want it to do). The whole problem is to find some way to make it more probable to not destroy us all. It is correct that a simple sentence can be interpreted to mean something that we don't really mean, even though the interpretation is logical for the AI.
Yes - and essentially this is a replay of the problem that has plagued philosophy and linguistics for hundreds if not thousands of years - the dream of producing a language with precise meanings - the "perfect language." (Has this not been discussed here?) Eco wrote a book about it. I think it's now generally recognised that it is a pure fantasy. I'm not so sure though whether it has been fully recognised that the whole function of language and any symbolic system is to be general and abstract and NOT pin down meaning or reference precisely. You obviously don't want numbers, for example, like "1" to refer to only one particular object. But you don't even want apparently particular names, like "Paul Horsmalahti" to refer to one particular object at one particular point in time. There are many "Paul Horsmalahti's", for every human has a rich, varied and developing personality - and,usually, physique.. ----- 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=4007604&id_secret=20469632-016a92
