What I'm asking is the objective of the test. Is it to convince the judge that you are a computer?
On Fri, Mar 6, 2015 at 6:24 PM, Steve Richfield <[email protected]> wrote: > Matt, > > Just because there are no (known) non-biological AGIs doesn't mean that we > can't run a competition for the biological variety. Just set it up so that > all participants are welcomed, regardless of their technology. > > Continuing... > On Fri, Mar 6, 2015 at 9:57 AM, Matt Mahoney via AGI <[email protected]> > wrote: > >> Could you explain the rules? >> > > The "rules" don't yet exist and would be established by the those > operating the future competition. Of course, an AGI would view "rules" to > be advisory and perceive itself free to "violate" them in creative ways > that work to the AGI's perceived interests. > > My thought was that the judge(s) would present a small number (2-3) > present social or scientific challenges in textual form, that might be > reasonable for a super-duper AGI to resolve, and see what the participating > people and/or teams come back with regarding their individual selections of > a smaller number (1-2) of those challenges. > > Scoring might be for the best of the poorest answers, but I suspect that > there may be other opinions about this. > > Any "ties" would be resolved in favor of the AGI that best followed the > rules. > > There would probably have to be a length limit, say 10 pages total, > because of the need to explain concepts that are NOT now commonly known. > > Early AGIs would be compute-bound, so several days would be allowed to > answer such world-changing questions. > > Thoughts? > > Steve > ====================== > On Fri, Mar 6, 2015 at 12:54 PM, Steve Richfield via AGI <[email protected]> > wrote: > >> It seems obvious (to) me that any envisioned super duper AGI of the >> future would be easily able to win a reverse Turing competition - >> demonstrating with advanced logical solutions to difficult problems that it >> is a machine and NOT merely human. >> >> To see how such an AGI might function, and how its responses might be >> perceived by mere humans, it seems (to me) VERY interesting to see what >> might come from such a competition, even though (for now) it only includes >> teams of mere humans. >> >> I suspect that heidenbugs (correct functionality that is seen to be >> erroneous) and incorrectly perceived sinister intent would make it nearly >> impossible for mere humans to accurately judge such a competition. If so, >> this would seem to doom the future utility of AGIs. >> >> As with Winograd schemas, the test is in the doing. Every your or so I >> post looking for others interesting in operating and/or participating in a >> reverse Turing competition. >> >> Any interest? >> >> Steve >> >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >> <https://www.listbox.com/member/archive/rss/303/3701026-786a0853> | >> Modify <https://www.listbox.com/member/?&;> Your Subscription >> <http://www.listbox.com> >> > > > > -- > -- Matt Mahoney, [email protected] > >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >> <https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac> | >> Modify >> <https://www.listbox.com/member/?&;> >> Your Subscription <http://www.listbox.com> >> > > -- -- Matt Mahoney, [email protected] ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
