Bob Mottram <[email protected]> wrote: To create a good test environment for AGIs there needs to be no be no single optimal or correct solution to the problem of existing in that environment. Being intelligent involves continuous learning in a dynamic environment where there are multiple possible trade-offs and strategies (aka "complex goals in complex environments").
But already you have taken a proposal for a test of simple (would-be) AGI programs and turned it into a test for an AGI program which would be capable of, "continuous learning in a dynamic environment where there are...'complex goals in complex environments'. Since we know that this is impossible at this time, we can also know that this kind of test is useless at this time. If you had an AGI program that had these capabilities (in the near future) you would not care about the dinky test. We need tests that are specifically designed to test for very simple learning and which can then be modified slightly in order to show generality. I do not think that all AI programs that work are narrow but I do think that they are all limited at this time. AGI means that the learning algorithms could be applied to various kinds of situations. However, it might take a programmer some time to adapt his program his program to show that it could work with different IO modalities so we would need initial variations to challenge the programmer with. I am wary of AI programs that are not able to show continued learning in complicated situations, but complexity is the problem that we are not able to solve at this time. So, given that as a present day reality, I am asking what kind of limited tests could be used to show generality - at this time. There was once a time when it was hoped that a simple test could be designed for AGI programs by limiting the domain of the problems to be solved. However, this did not work because a complex use of knowledge is required for even limited domains. Instead, I am proposing that programmer design a test showing that his program could work with simple problems, and then other sympathetic programmers could challenge him with similar but different problems. If his program still worked then more demanding variations of problems could be proposed. Jim Bromer On Sat, Jun 9, 2012 at 6:47 AM, Bob Mottram <[email protected]> wrote: > On 09.06.2012 04:04, Jim Bromer wrote: > >> Because we would be testing primitive programs, they have to be >> defined by the programmers.� Other AGI programmers might then comment >> sympathetically to see if the -test- could be made a little more >> sophisticated for the programmer's program. The program would have to >> learn.� The learning could take place through direct instruction but >> the program would have to be able to figure some things out for >> itself.� The learning method would have to go beyond basic filling of >> variable types or of basic numerical computation. >> > > > Some common problems with this kind of thing: > > i) The environment is too simple. > > ii) The agent-environment system reaches an equilibrium state, and then > stays there. This is an "informational death" or "end of history" > situation. > > iii) The design of the agent-environment system intentionally or > unintentionally precludes virtual machine stratification. > > iv) The communications system between agents is not Turing complete, with > only simple signals or gestures being possible. > > > To create a good test environment for AGIs there needs to be no be no > single optimal or correct solution to the problem of existing in that > environment. Being intelligent involves continuous learning in a dynamic > environment where there are multiple possible trade-offs and strategies > (aka "complex goals in complex environments"). > > > > ------------------------------**------------- > AGI > Archives: > https://www.listbox.com/**member/archive/303/=now<https://www.listbox.com/member/archive/303/=now> > RSS Feed: https://www.listbox.com/**member/archive/rss/303/** > 10561250-164650b2<https://www.listbox.com/member/archive/rss/303/10561250-164650b2> > Modify Your Subscription: https://www.listbox.com/** > member/?&id_**secret=10561250-2cb1ec2f<https://www.listbox.com/member/?&;> > Powered by Listbox: http://www.listbox.com > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
