You are not engaging with what I am actually saying. Telmo
Am Sa, 18. Mär 2023, um 13:29, schrieb John Clark: > On Sat, Mar 18, 2023 at 5:28 AM Telmo Menezes <[email protected]> wrote: > >> *> Huge progresses is being made, but we are not at the human level of >> generality of intelligence and autonomy. Not even close.* > > Not even close? Don't be silly. > >> *> I fear that you are falling for the very human bias (I fall for it so >> many times myself) of seeing what you want to see.* > > And I fear you are whistling past the graveyard. > >> *> A machine learning system can only be objectively evaluated by applying >> it to data that was not used to train it.* > > I don't know what you mean by that, you're not falling for that old cliché > that computers can only do what they're told to do are you? GPT-4 was not > trained on the exact questions asked; I suppose you could make a case that > some of the training data GPT-4 was educated on was somewhat similar to some > of the questions it was asked, but the exact same thing is true for human > beings. When you ask questions to a human being some of those questions are > somewhat similar to data he was educated on. In fact if some of the data 2 > intelligences were educated on were not similar they would not be able to ask > each other questions because they wouldn't even be able to communicate. > > >> *>Again, it is important to understand what exactly GPT-4 is doing. It is >> certainly impressive, but it is not the same thing as a human being taking >> an IQ test,* > > So you must think the following fundamental axiom is true: > > *"If a human does something that is smart then the human is smart, but if a > computer does the exact same thing then the computer is NOT smart."* > > And from that axiom it's easy to derive the following Corollary: > > *"Computers, buy definition, can never be smart."* > > I think you need to be more careful in picking your fundamental axioms. > >> >> *> I do think that passing the Turing test is impressive,* > > Probably the greatest understatement of all time. > >> *> although it is true that most AI researchers never took it very >> seriously,* > > What?! I'm sure that in their daily lives AI researchers, like every other > human being on planet earth, have met people in their life that they > considered to be very intelligent, and people they considered to be very > stupid, but if they didn't use the Turing Test to make that determination > then what on earth did they use? All the Turing test is saying is that you > need to play fair, whatever criteria you used to judge the intelligence of > your fellow human beings you should also use on a computer to judge its > intelligence. > > It's always the same, I'm old enough to remember when respectable people were > saying a computer would never be able to do better than play a mediocre game > of chess and certainly never be able to beat a grandmaster at the game. But > when a computer did beat a grandmaster at Chess they switched gears and said > such an accomplishment means nothing and insisted a computer could never beat > a human champion at a game like GO because that really requires true > intelligence. Of course when a computer did beat the human champion at GO > they switched gears again and said that accomplishment means nothing because > a computer would never be able to pass the Turing Test because that really* > really* requires true intelligence. And now that a computer has passed the > Turing Test the human response to that accomplishment is utterly predictable. > As I said before, they're whistling past the graveyard. > > ... and so, just seconds before he was vaporized the last surviving human > being turned to Mr. Jupiter Brain and said "*I still think I'm more > intelligent than you*". > > >> *> GPT-4 and image generators are a type of intelligence that we had never >> seen before. Maybe the first time such a thing arises in this galaxy or even >> universe,* > > I agree, and I can't think of anything more important that happened in my > lifetime. > > >> > *They are probably also similar to stuff that happens in our brain. But >> what they are not is something you can be compare to a human mind with an IQ >> test in any meaningful way.* > > Not just *an* IQ test but 4 quite different types of IQ tests. And it was a > lobotomized version of GPT-4 that was tested that could not input graphs and > charts or diagrams so any question that contained them was automatically > marked as getting wrong, and yet it STILL got an IQ of 114. And the computer > completed those tests in seconds while it took humans hours to do the same > thing. Imagine what IQ score it will get in two years, or even two months. > And you say "not even close"? > >> *> That is just junk science.* > > Huh? Creating "a type of intelligence that we had never seen before, maybe > the first time such a thing arises in this galaxy or even the universe", is > junk science? > > John K Clark See what's on my new list at Extropolis > <https://groups.google.com/g/extropolis> > e4v > -- 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 view this discussion on the web visit https://groups.google.com/d/msgid/everything-list/09a49ea8-2d7e-49a1-9950-f28438d34d32%40app.fastmail.com.
