On Thu, Mar 3, 2011 at 8:00 AM, steve uurtamo <[email protected]> wrote:
> i find it quite interesting that all of this speculation is geared > toward saying that something unfair toward humans was happening. > It's not unfair, it's just not interesting. If this is pure Jeopardy test then the computers superior reaction time is a factor and completely fair but I don't care about that, I know that computers can do billions of discrete operations per second and I cannot. However, I assume that this test was supposed to highlight the artificial intelligence aspect of the software and not the reflex time of the computer. Presumably, there is still some non-trivial processing going on if the humans beat the computer to the buzzer on occasion. If the question is given to the computer AFTER it's been verbally given to the human (and Alex stops speaking) then it's probably unfair to the computer (as you say, the human can process the written version of the question before Alex stops speaking.) Another thing that might be considered unfair to the computer is that a human can press the buzzer as soon as he thinks he will be able to answer the question but before he knows for sure that he can answer the question. There is a lot of trivia that I KNOW the answer is in my head, but it might take me a couple of seconds or more to recall the information. But I assume the computer will not buzz until it actually has the answer. A fair test to me would be to build a Jeopardy playing robot - one that has to process the question using speech recognition and then press the buzzer with some kind of mechanical linkage. Something like Data the Android on Star Trek TNG. Don > in reality, the humans beat the machine to the buzzer more than once, > the machine had to actuate a physical buzzer, everyone had the same > amount of time to read and process the question (even i can read the > displayed question and finish thinking about whether i know the answer > before alex stops speaking, which is when they start checking the > buzzers). > > so just to be clear, here is a short list of the ways in which > everything was the same: > > the machine and humans got to see the question at the same time (it is > displayed for the humans, opened as a file by the machine). yes, alex > speaks the question. but even i can make a decision about whether or > not i know the answer before he finishes talking. > > nobody gets to hit the buzzer before they are signaled that it is > time. a human sends this signal. if anyone tries to buzz in before > this time, they get penalized with a delay. > > the machine actuated a real physical buzzer, just like the people did. > > here is a short list of the ways in which the machine had an advantage: > > there may have been a 50ms delay inbetween intending to hit the buzzer > and hitting it for the humans that was shorter for the machine. > > here is a short list of the ways in which the humans had an advantage: > > they are HUMAN and can UNDERSTAND the questions. > > jesus. > > s. > _______________________________________________ > Computer-go mailing list > [email protected] > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go >
_______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
