1997 Loebner Prize $2000 and a bronze medal
Hugh Loebner
The 1997 Loebner Prize contest is scheduled for Tuesday,
April 29, 1997 at the Salmagundi Club, New York City.
Complete rules are at:
http://info.acm.org/~loebner/lpr97.html
Information and transcripts of the 1995 ad 1996 contests are at:
http://info.acm.org/~loebner/loebner-prize.htmlx
sig of the month---------------------------------------------------------
Voice Mail 1-500-EGO-TRIP Hugh Gene Loebner e-mail loe...@acm.org
http://info.acm.org/~loebner
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Parsimony says:"Minimalism to the max."
Bear
Hugh Loebner (loe...@ACM.ORG) wrote:
of many things
of dates and times and world wide web
of Babbages and Kings
Say Hughmeister,
Can't help but notice the title of this thread. Are you
saying that April 97's contest will be strictly for the bronze,
even if some some brilliant hack (including audio and visual)
should (at least in theory) walk with the Biggy?
If so, ok. Your contest, your rules.
If not, you wanna open the discussion on the net as
to exactly what it will take to grab the $100K?
I would guess:
1. convincing everyone gathered that it was a human
in the next room.
2. after the curtain was raised, further convincing
everyone there that it was the greatest program since the
invention of English Beer.
3. being able to converse via audio I/O, in human
manner (no "robotty" sounds)
4. being able to use "vision" to identify objects.
I would also assume that certain things that a given
(RANDOM) human would not be expected to do, would also
not be expected of the winning program. i.e. The problem
would not be put thusly:
"Write a 5-piece aria for Cello and Woodwinds. Make me cry."
I mean, in essence, you can ask anything of this program
right? I am assuming that the expected results lie within
the AVERAGE human plane.
Anyway, those are just starters, suggestions.
Any feedback/decisions are welcome. It's your game!
-Jimbo
ps. Please lemme know if you *are* only offering
the bronze in April. It would change my tack a smidge!
ss...@intranet.ca
To: Ecl...@access2.digex.net
Subject: Re: 1997 Loebner Prize $2
I'm not Loebner, but I'm replying anyway... suffah.
Ec> saying that April 97's contest will be strictly for the bronze,
Ec> even if some some brilliant hack (including audio and visual)
Ec> should (at least in theory) walk with the Biggy?
*cough* Yeah right. The fact that Loebner bumped the prize criteria up to
this even harder goal that'll probably never be realized in his lifetime
shouldn't bother you though.
[Actually, as far as I can tell, he's trying to humanize the criteria more,
which is both bad and good]
Ec> If so, ok. Your contest, your rules.
Ofc.
Ec> If not, you wanna open the discussion on the net as
Ec> to exactly what it will take to grab the $100K?
It's in the contest rules.
Ec> 1. convincing everyone gathered that it was a human
Ec> in the next room.
The average score of the AI would just have to be higher than the lowest
scoring human at that year's contest.
Ec> 2. after the curtain was raised, further convincing
Ec> everyone there that it was the greatest program since the
Ec> invention of English Beer.
I'm sure people would be suitably impressed. (Canadian beer is better! <G>)
Ec> 3. being able to converse via audio I/O, in human
Ec> manner (no "robotty" sounds)
The current solution to that problem has the same problems as natural language
processing. It can use the ANN abstraction layer to decipher the human's
speech but it still has to use its natural language processor to come up with
a response.
Ec> 4. being able to use "vision" to identify objects.
This requires knowledge about the universe on par with a human. Considering
humans have fully evolved wetware and a few decades to learn, I imagine this
goal will be realized sometime in the 24th century (hands up who understands
the joke!).
Ec> I would also assume that certain things that a given
Ec> (RANDOM) human would not be expected to do, would also
Ec> not be expected of the winning program. i.e. The problem
Only if the judges consider spontaneity important.
Ec> would not be put thusly:
Ec> "Write a 5-piece aria for Cello and Woodwinds. Make me cry."
What would you say to a request like that?
That's what the computer would say.
Ec> I mean, in essence, you can ask anything of this program
Yes, but it doesn't mean it has to answer.
Ec> right? I am assuming that the expected results lie within
Ec> the AVERAGE human plane.
No, the results have to lie above the lowest scoring human of the human
confederate at that year's contest.
Ec> Anyway, those are just starters, suggestions.
Ec> Any feedback/decisions are welcome. It's your game!
Reminds me of Finite ... we're learning about probabilities and, therefore,
'fairness' of games. The probability of anyone winning this game is equal to
the probability that Loebner's going to end it, which, the last I hear, was
0.00002. BTW, I don't mean that he should end it... just read the link
underneath Minsky Co-Sponsors Contest on the Loebner Contest homepage.
Ec> ps. Please lemme know if you *are* only offering
Ec> the bronze in April. It would change my tack a smidge!
I wouldn't even worry about it.
'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
' Sunir Shah (ss...@intranet.ca) '
' http://intranet.ca/~sshah ftp://ftp.intranet.ca/usr/synapsis '
' Fidonet: 1:241/11 BBS: The Open Fire BBS +1 (613) 584-1606 '
' '
' By the WEB: Vanity: http://intranet.ca/~sshah/ '
' The Programmers' Booklist booklist.html '
' ~`-,._.,-'~ Synapsis Entertainment synapsis.html '
' _.,-`~'-,._ WASTE (Warfare AI Contest) waste/waste.html '
' '
' comp.ai.games FAQ: ftp://ftp.intranet.ca/usr/synapsis/cagfaq?.txt '
' The Game Development Echo: Areafix GAMEDEV from Zone 1 (Fido) '
'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
>> Slipstream Jet - The QWK solution for Usenets #UNREGISTERED
Bernard DAUNAS
On 29 Apr 1996 03:05:20 GMT, ecl...@access2.digex.net (Bear) wrote:
> If not, you wanna open the discussion on the net as
>to exactly what it will take to grab the $100K?
>I would guess:
> 1. convincing everyone gathered that it was a human
>in the next room.
Why not just convincing every one that the thing in the other room is
intelligent ? Is this an intelligence contest, or a Oh gosh, see the
devil how I look and behave like a human ?
> 2. after the curtain was raised, further convincing
>everyone there that it was the greatest program since the
>invention of English Beer.
> 3. being able to converse via audio I/O, in human
>manner (no "robotty" sounds)
Obviously, deaf humans are not considered the be elligible to the
status of intelligent beings...
> 4. being able to use "vision" to identify objects.
Wahou... Blind people neither !
>I would also assume that certain things that a given
>(RANDOM) human would not be expected to do, would also
>not be expected of the winning program. i.e. The problem
>would not be put thusly:
>"Write a 5-piece aria for Cello and Woodwinds. Make me cry."
Let's be serious. Is this a contest to determine wether the goal of
machine intelligence has been achieved, or a human looking robot
contest ? If the second choice is Loeber's, the contest does not
qualifies as a Turing test. Let's rename it an Asimov test or
something... I wonder what is the rationalization behind this huge
increase in the difficulty of the contest, besides making sure the
Grand Prize is not awarded, or restricting the entrants to rich
multidisplinary teams who can afford very expensive hardware.
Why not sticking to the original Turing test ? You know... just a
computer terminal between judge and program. No audio, no video, no
fancy or expensive hardware required besides what is stictly necessary
to be qualified as intelligent, namely a pretty good PROGRAM ? Then
it's judges job to sort humans from computers.
Bernard Daunas.