Agreed, but the Turing test is annoying from the perspective of a system like Novamente that explicitly does NOT try to be humanlike...
I don't know whether the Turing test or the University of Phoenix test will wind up being harder for a semi-mature Novamente system... ben On 4/25/07, J. Storrs Hall, PhD. <[EMAIL PROTECTED]> wrote:
On Tuesday 24 April 2007 18:06, Eliezer S. Yudkowsky wrote: > Aside from that, it [the U. of Phoenix test] sounds fair enough to me, > and unlike the Turing Test it might not require strongly superhuman > intelligence. The Turing Test doesn't require superhuman intelligence, strong or otherwise. I could pass it, for example. It would require abilities beyond what could be called legitimately intelligent, given that the intelligence was implemented in such a way as to be optimized to the capabilities of a serial von Neumann computer instead of an evolved human brain, since it would have to hide some of its abilities and pretend feelings and so forth. On the other hand, an upload would pass the Turing test effortlessly, and would be about as close to a definition of an exactly human-equivalent AI as you're going to get. An AI that was the product of a Kurzweil-style development effort where the functional structure of the brain was copied in general, but not one specific individual, could pass it as well without any more effort than an actor or spy uses in assuming a fictitious identity. Josh ----- 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/?&;
----- 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=fabd7936
