On Sunday, December 21, 2014 11:05:08 PM UTC, John Clark wrote: > > On Sun, Dec 21, 2014 at 2:45 PM, <[email protected] <javascript:>> > wrote: > > > Yesterday you said you had to conclude if the test detected >> consciousness well it must also detect intelligence. > > > That's not what I said, you've got it backward. Something can be conscious > but not intelligent, but if it's intelligent then it's conscious. > Consciousness is easy but intelligence >
John - take the amount new knowledge you assert I just the above sentence. From where or what do you acquire this position? I mean, I would say that's fundamental....can you show your workings? What due dilligence did y0u perform? it's a must to do it....because it is very easy to assemble some components....such as concepts....and feel like something was discovered. look again and that assembly was one of hundreds of variants. meaning NO VALUE! > The Turing Test does not detect either one > > > I am certain you have met people in your life that you wouldn't hesitate > to call brilliant, and you've met people you'd call complete morons, but if > you don't examine the same thing that the Turing Test does, behavior, how > do you make that determination? > does John, sure I would examine something empirically if necessary. It often isn't. The way I can know what the Turing test was set up to do, is because it is straightforward to set up the test with all three measures. * Intelligence* * the human* consciousness. . Not Direct - We can eliminate consciousness as a direct detection because there is no knowledge for that in the Test. - Now we entertain Intelligence. - So it comes down to the logic of the Test. If Turing proposed a test that was based on intelligence testing what is his reason for leaving intelligence testing, which was highly standardized in his time, and eminently automatable/systemisable - There's no evidence Turing was a luddite. He already had incredible automation experience and ability. - If it was intelligence testing, I would argue that part of the Test, or at least a large amount of it, would have been automated at that time. No need to wait for A.I, get humans to take the test. Of course al of this completely standard and automated today. - In summary, there's no way to make sense of his Test with intelligence testing. - I mean...it's not efficient getting a human to decide the way Turing says. It would work vastly better by double blind testing lots of humans and throwing in the A.I. at multiple places. There are thousands of tests that could be taken. Personality, tricks, lying, deliberately lying,, cohersion, pain - The right way to do that would be double scattering the A.I. everywhere, thousands of tests. And then at the end, the human interaction component would be best one of our top boys thinks, without peeping, something in those tests that should be there for every human interaction, and much less so, gone down and octave, for every interval it's the A.I. He makes his prediction, and another team test for it, The analysis is normalized and if he's right then the A.I. was detected. If he's wrong,his theory and all the tests, and the maths, and checking, is all fed into a copy of the A.!. That copy is going to version 2 but it's a long way off. The A.I. and its technicians spend a lot of time getting an algorithm to absorb that failed theory and so make the A.I. better at this game next time. Then staying with the version there's a queue outside, a lot of people want to take their go. Make 30 duplicates of version one, and get 30 at a time defining theory theory and getting it tested. Open up new varieties of theory. Now we've computer scientists and psychologists coming up with environments involve a lot of time making the setup look like it's for one thing, when actually it isn't. It could really be for something immensely simple...anything that a human - every human - can be spun and kept off balance and suddenly, it happens, the little voltage, of the white noise, or perhaps we use one of those tricks that harness human visual perception. On the other side The A.I. be massing large new knowledge of the sort of things humans will try. The A.I. would be duplicated several more times, now to be scattered among all 'roles'. Everything would be double blind. The A.I. might get ahead and reverse everything out and maybe bring about a ridiculous, disturbing and definitely very entertaining, "mirroring out" in which, say, particular theory test, is manipulatd such that the A.I. is identified! But it turns out to be the guys sister in law. Anyway, one could on. But the illustration point, is that testing for intelligence and psychology and s on....it could be harnessed if the process was more than just test but also an A.I. evolution density point. That's the way I'd do it for intelligence. It'd be a Turing Test, but not as we know it, captain. -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/d/optout.
