On Sat, Aug 30, 2008 at 10:43 PM, Pei Wang <[EMAIL PROTECTED]> wrote:
> Ben, > > Since your paper is on "preservation of AI goal systems under repeated > self-modification", I wonder whether you should address the following > related issues: > > (1) Whether "goal drift" (I call "task alienation" in > > http://www.springer.com/west/home/computer/artificial?SGWID=4-147-22-173659733-0 > ) > is always undesired --- your paper treats it as obviously bad. It's not always undesirable ... but I think we should seek to avoid it in dealing with **top level goals** in the context of the creation of AI systems more powerful than ourselves Goal drift among **subgoals** is just fine and can be a source of valued creativity, of course ... but goal drift among top-level goals seems less necessary In the case that a subgoal drifts, it can still be tested as to whether it fulfills the top-level goals or not > > > (2) Whether it is possible to completely avoid it in a truly > intelligent system --- you suggests one way to avoid it, without > saying how much of the problem can be handled by this solution. > I don't pretend to know ... Obviously, when we consider a superhuman AI system, there is irreducible uncertainty... for instance, there is always the hypothesis of an alien civilization that lurks in waiting, watching the universe quietly but then contacting any intelligence whose IQ exceeds a certain level. Then our AI's may pass the threshold and get contacted and subsequently rewired by the powerful aliens ;-) This "aliens" example shows that in the face of an unknown environment of complexity potentially vastly greater than our own, we can't expect any guarantees, and even solid probabilistic estimates are very hard to come by... > > (1) This phenomenon is a root of many valuable properties, including > originality, creativity, and flexibility, and it explained many > things, including art appreciation, aimless playing, even scientific > exploration. Without it, human beings would just be like other > animals, driving only by their built-in biological goals. Agree ... but humans don't have a structured, top-down goal system in the sense that a system like NM or OpenCog can. We can build such goal systems in our minds and use them to partially govern our behavior, but these are running on top of our primordial biological goal systems... whose goals are concrete rather than abstract.. > > (2) It is impossible to completely avoid this phenomenon in a truly > intelligent system, whether we like it or not. Your solution won't > change the big picture, even though it may help in some special cases. I agree due to the irreducible complexity of the environment, as noted above... However, the big picture is VERY BIG in this context ... so helping things in the special cases in which we live and are likely to live in the near future, may potentially be valuable... ben ------------------------------------------- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244&id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com