From: nuzz604 (firstname.lastname@example.org)
Date: Fri Dec 09 2005 - 11:44:23 MST
Perhaps a good compromise would be to develop a system using a narrow-AI approach that is capable of processing knowledge and answering questions, but not generally intelligent enough to improve itself or reshape the world. When you have a working system, even narrow AI, it will be easier to acquire funding. The question is not about drawing the line between having a goal structure and not having a goal structure (The program's goal is to execute every line of code in it's proper manner). The question is about drawing the line between narrow and general AI.
Make it narrow enough to -look good-, even if you can't make it intelligent enough to do any real work. You should have a good enough friendliness theory and a good enough "demonstration" by then to attract funding.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:54 MDT