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

[Kelly Anderson]

2011/1/31 Dave Sill <sparge at gmail.com>:
> On Mon, Jan 31, 2011 at 1:08 PM, Kelly Anderson <kellycoinguy at gmail.com>
> wrote:
>>
>> The strongest Turing test is when someone who knows a lot about
>> natural language processing and it's weaknesses can't distinguish over
>> a long period of time the difference between a number of humans, and a
>> number of independently trained Turing computers.
>
> No, language processing is only one aspect of intelligence. The strongest
> Turing test would also measure the ability to learn, to learn from past
> experiences, to plan, to solve problems...all of the things the Wikipedia
> definition mentions, and maybe more.

You are right.

>> So perhaps I suggest a new test. If a computer is smart enough to get
>> admitted into Brigham Young University, then it has passed the
>> Anderson Test of artificial intelligence.
>
> You mean achieve an SAT score sufficient to get into BYU? Or do you mean
> that it has to go through school or take a GED, fill out an application to
> BYU, etc. like a human would have to do?

Passing the SAT would be only one part of the test, but it would also
have to pass some kind of high school, write an essay on why it
deserves to be admitted, fill out the forms and so forth. The idea is
to be intellectual enough to fool the admissions board. I picked this
test because you don't have to physically appear to be admitted to
most colleges. I would not require the robotics aspect... you could
put the paper in the printer for it... mail the forms, etc.

>> Is that harder or easier than the Turing test?
>
> Depends on the Turing test, I'd say.

Sure.

>> How about smart enough to graduate with a BS from BYU?
>
> How about it? It'd be an impressive achievement.

Would it be intelligent? I think so.

>> Another test... suppose that I subscribed an artificial intelligence
>> program to this list. How long would it take for you to figure out
>> that it wasn't human? That's a bit easier, since you don't have to do
>> the processing in real time as with a chat program.
>
> Depends how active it is, what it writes, and whether anyone is clued to the
> fact that there's a bot on the list. A Watson-like bot that answers
> questions occasionally could be pretty convincing. But it'd fall apart if
> anyone tried to engage it in a discussion.

I would assume that nobody would be clued in. If you have a suspicion
that something is amiss, you start looking for things. It's how I
watch CGI movies... I look for the mistakes. When I just relax and
enjoy the movie, I can believe what I see better. Benjamin Button
passed this "Graphical Reality" test for me, btw. Very impressive.

I don't think Watson could pass this test now, but I would not be
surprised if it could at some point in the not too distant future.

>> That's the difference between taking a picture, and telling you what
>> is in the picture. HUGE difference... this is not a "little" more
>> sophisticated.
>
> No, parsing a sentence into parts of speech is not hugely sophisticated.

But "understanding" the sentence is. Parsing an image into blobs is
not highly sophisticated either, but labeling the blob in the middle
as a "dog" is.

>> Once again, we run into another definition issue. What does it mean to
>> "understand"?
>
> http://en.wikipedia.org/wiki/Understanding

Quoting: "a psychological process related to an abstract or physical
object, such as a person, situation, or message whereby one is able to
think about it and use concepts to deal adequately with that object."

So contextually to Jeopardy, Watson understands the questions it
answers correctly. Right?

>> And if that form is such that I can
>> use it for future computation, to say answer a question, then Watson
>> does understand it. Yes. So by some definitions of "understand" yes,
>> Watson understands the text it has read.
>
>  Granted, at a trivial level Watson could be said to understand the data
> it's incorporated. But it doesn't have human-level understanding of it.

But by the Wikipedia definition, it only has to "deal adequately"...
Winning several thousand dollars on Jeopardy would certainly seem to
be adequate, IMHO.

-Kelly