[ExI] Oxford scientists edge toward quantum PC with 10b qubits.

[John Clark]

On Jan 28, 2011, at 2:27 PM, Adrian Tymes wrote:
> 
>  Yep.  Because, in the process of solving them, we keep finding tricks and
> cheats that those who thought "X requires intelligence" didn't conceive of.

Did it ever occur to you that our own human intelligence is not some mystical thing but just such a collection of "tricks and cheats". Just because you have a very general understanding how it does what it does doesn't mean its not intelligent; you demand the "secret" of intelligence be specific enough to be encoded into the language of ones and zeros but at the same time it must be utterly mysterious, be purposeful but non deterministic, and simultaneously be exact and vague. So with all those contradictory requirements obviously you will never see a machine that you will call intelligent, but you will see a machine that uses "tricks and cheats" to perform any task you care to name better than you can, any task whatsoever. And if that's not "true intelligence" its good enough for me. 

>  Watson is not capable of doing anything but Jeopardy

But Jeopardy includes not just having an encyclopedic knowledge of everything from pulsars to pop culture and finding the one and only correctly wanted fact in a vast sea of facts based on a remark that  is elliptically phrased, but also in dealing with rhymes and riddles and even puns. To pretend that this is not impressive is silly.  

> it certainly didn't learn to do that on its own

And you didn't learn to do what you do on your own either, you had teachers, you read books written by others and you watched what other people did.

> Get me a computer that can learn to do things it was never programmed or
> designed to do.  

Like finding the question to a strangely worded answer neither it nor anybody else on this planet had ever heard before? I just have little patience with the "if a man does it then it's intelligence but if a machine does it then it's not" school of thought.

> Note that the Turing Test is a partial codification of this.

If Watson can find a good question (even if its not always the correct question) to any answer then that's not too far from the Turing Test.

  John K Clark




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20110128/aed5765d/attachment.html>