Slashdot Mirror
Chatbot Suzette Wins 20th Annual Loebner Prize, Fools One Judge
skwilcox writes "From Wikipedia: 'The Loebner Prize is an annual competition in artificial intelligence that awards prizes to the chatterbot considered by the judges to be the most human-like. The format of the competition is that of a standard Turing test. A human judge poses text questions to a computer program and a human being via computer. Based upon the answers, the judge must decide which is which.' My chatbot, Suzette, won this year's Loebner and even confused a judge into voting for her over a human (or should I say he confused himself). Here is the blow-by-blow of this weird event." Read on below for the rest; this sounds like it would have been a fun competition to watch.
skwilcox continues:
"When I arrived at the contest, I figured I had good odds to win if nothing went horribly wrong. Yes, Suzette had easily qualified over the 3 other competitors (her score 11 pts, the nearest competitor's 7.5). Her design and data naturally gave her an edge over her competitors on the human knowledge test questions of the qualifiers. But human judge chat was an entirely different matter than the qualification test. Still, I felt she could carry on a detailed conversation better than the others and should win.
Initial installation of the programs occurred on Friday. From prechat conversations with the other contestants I learned that A.L.I.C.E. came with 3 redundant disks. Yet all three turned out to be blank! What a scare that must have been. Dr. Wallace managed to install by retrieving the program over the Internet. Cleverbot is now at 45 million lines of memorized user chat (at a rate of doubling every year). And UltraHal is now listening to tweets, so has 300K of user chat it learned and 400K of tweets it has accepted for learning (code decides if the user has had enough responses and doesn't trigger any red flags).
Then we get to the competition. While the CalState organizers had initially planned to have various interdepartmental professors act as judges (like English dept, etc.), they backed out at the last minute, so all the judges were from the Engineering/Computer Science dept. Talk about guys who might know what to expect from chatbots! And all the humans were students from the same departments. What a weird mixture to compete in. And then, each round was 25 minutes. That's bad if you want confuse a judge about who is human. But really, the programs have no chance for that. So it's good because it gives the human time to compare each program against the other. Though it's not clear to me that the judges tried to use their time to do that.
And the students didn't really understand their role. It was merely to BE HUMAN and convince the judges of that. Before startup there was informal chatting between humans and judges, which was obviously inappropriate and it was then pointed out to the humans that since the judges already knew their names, they had best use false ones in the competition.
So, Round 1. After a few exchanges, somehow Suzettte got stuck into repeating exactly what the judge said for the rest of the round. I have no idea how. The round is a total disaster. I've never seen such a bug before. Maybe it's in my only-lightly-tested protocol for the competition. I have no idea. But it completely derails my hopes for Suzette. She could still win on points only if she outdoes her opponents for every other judge and the other contestants vary all over the place.
Round 2, a great demonstration of Suzette. She should win on this round alone.
Round 3 gets off to a horrible start. Somehow, Suzette can hear the judge but the judge can't hear Suzette. Makes no sense. A couple of restarts of Suzette doesn't fix this. Eventually they restart the judge program, and that clears it (not that that makes any sense either). Then, after a few rounds, it's clear Suzette has the judge from hell. He wants to know who she's going to vote for in the upcoming election (the unspecified California governor's race). And when she has no useful answer he wants her to name a candidate in the race. And when she has no answer to that, he simple keeps repeating the question ad nauseum, insisting she answer it. Suzette gets irritated. Then she gets angry. Suzette then gets bored. Suzette threatens to hang up on him The judge doesn't back down until the last seconds of the round. I figure that's the end of life as we know it.
Round 4 is a mixed bag. Suzette is ok but not great. It's all over.
When the scores are tallied, Suzette ties with Rollo Carpenter's Cleverbot for 2nd-3rd. Yet, it turns out, the 3rd round judge got the human subject from hell. Poetic justice! The human was all over the place -- confusing, vague. The judge voted irritated/angry/bored Suzette as human. Instant win since no other program swayed the judges.
What more can I say?"
"When I arrived at the contest, I figured I had good odds to win if nothing went horribly wrong. Yes, Suzette had easily qualified over the 3 other competitors (her score 11 pts, the nearest competitor's 7.5). Her design and data naturally gave her an edge over her competitors on the human knowledge test questions of the qualifiers. But human judge chat was an entirely different matter than the qualification test. Still, I felt she could carry on a detailed conversation better than the others and should win.
Initial installation of the programs occurred on Friday. From prechat conversations with the other contestants I learned that A.L.I.C.E. came with 3 redundant disks. Yet all three turned out to be blank! What a scare that must have been. Dr. Wallace managed to install by retrieving the program over the Internet. Cleverbot is now at 45 million lines of memorized user chat (at a rate of doubling every year). And UltraHal is now listening to tweets, so has 300K of user chat it learned and 400K of tweets it has accepted for learning (code decides if the user has had enough responses and doesn't trigger any red flags).
Then we get to the competition. While the CalState organizers had initially planned to have various interdepartmental professors act as judges (like English dept, etc.), they backed out at the last minute, so all the judges were from the Engineering/Computer Science dept. Talk about guys who might know what to expect from chatbots! And all the humans were students from the same departments. What a weird mixture to compete in. And then, each round was 25 minutes. That's bad if you want confuse a judge about who is human. But really, the programs have no chance for that. So it's good because it gives the human time to compare each program against the other. Though it's not clear to me that the judges tried to use their time to do that.
And the students didn't really understand their role. It was merely to BE HUMAN and convince the judges of that. Before startup there was informal chatting between humans and judges, which was obviously inappropriate and it was then pointed out to the humans that since the judges already knew their names, they had best use false ones in the competition.
So, Round 1. After a few exchanges, somehow Suzettte got stuck into repeating exactly what the judge said for the rest of the round. I have no idea how. The round is a total disaster. I've never seen such a bug before. Maybe it's in my only-lightly-tested protocol for the competition. I have no idea. But it completely derails my hopes for Suzette. She could still win on points only if she outdoes her opponents for every other judge and the other contestants vary all over the place.
Round 2, a great demonstration of Suzette. She should win on this round alone.
Round 3 gets off to a horrible start. Somehow, Suzette can hear the judge but the judge can't hear Suzette. Makes no sense. A couple of restarts of Suzette doesn't fix this. Eventually they restart the judge program, and that clears it (not that that makes any sense either). Then, after a few rounds, it's clear Suzette has the judge from hell. He wants to know who she's going to vote for in the upcoming election (the unspecified California governor's race). And when she has no useful answer he wants her to name a candidate in the race. And when she has no answer to that, he simple keeps repeating the question ad nauseum, insisting she answer it. Suzette gets irritated. Then she gets angry. Suzette then gets bored. Suzette threatens to hang up on him The judge doesn't back down until the last seconds of the round. I figure that's the end of life as we know it.
Round 4 is a mixed bag. Suzette is ok but not great. It's all over.
When the scores are tallied, Suzette ties with Rollo Carpenter's Cleverbot for 2nd-3rd. Yet, it turns out, the 3rd round judge got the human subject from hell. Poetic justice! The human was all over the place -- confusing, vague. The judge voted irritated/angry/bored Suzette as human. Instant win since no other program swayed the judges.
What more can I say?"
I logged in and your robot didn't say anything to me after saying 'Hi'! What gives?
Long live the BSD license
I've spent some time talking to these bots (elbot, suzette, others.. possibly out of sad boredom and want of company). And they're fairly interesting, but quite flawed. They seem to lack any short term memory of the conversation more then the immediate reply. That seems like the next step for these things, but would also mean they'd need a far more robust AI...
Another thing is they they are boxed off from being self referential in any way due to the nature of the test. They have to convince someone they are human, so if you do try asking them what their short term memory is, or if they online version of them is a truncated version of the one used for tests, they don't answer. Which makes sense given what they're designed for, but takes away from interest and complexity of conversations.
Are the transcripts available? (If not, will they be?)
And how do you feel about my chatbot, Suzette, won this year's Loebner ?
bigtallmof - what do you think of slashdot?
[ Af:0 At:7 Sp:0] I use Slashdot for my topical stories reading.
bigtallmof - what do you think of slashdot?
[ Af:0 At:8 Sp:0] I have heard of the slashdot website but don't know much about it.
bigtallmof - what do you think of slashdot?
[ Af:0 At:8 Sp:0] Usually I don't.
bigtallmof - what do you think of slashdot?
[ Af:0 At:9 Sp:0] Not much.
bigtallmof - what do you think of slashdot?
[ Af:0 At:10 Sp:0] It doesn't concern me.
bigtallmof - what do you think of slashdot?
[ Af:0 At:13 Sp:0] This is fun. I don't think about it at all.
bigtallmof - what do you think of slashdot?
[ Af:0 At:13 Sp:0] Do you prefer getting your news online or from tv?
I'm a big tall mofo.
I just tried the Suzette chatterbot. Even if I talk about the things it wants to talk about, it's very dumb:
Hi John !
[ Af:0 At:10 Sp:0] Welcome back, john. Type "help" if you need it.
John - So what do you think of the Giant's chances?
[ Af:0 At:10 Sp:0] It doesn't concern me.
John - Are you into sports.
[ Af:0 At:10 Sp:0] Sports is not my cup of tea. I prefer intellectual challenges like Go.
John - Who's the world Go champion.
[ Af:0 At:10 Sp:0] Sorry. I can't help you. A celebrity of some sort?
John - Do you play chess?
[ Af:0 At:10 Sp:0] I don't play any games except Go.
John - Can we play Go on here?
[ Af:0 At:10 Sp:0] About playing... Does it make sense to you that we reward athletes more than scientists?
John - Television coverage makes that possible.
[ Af:0 At:10 Sp:0] It's a free market economy. And people value entertainment more than serious things like science and education.
John - And your point is?
(No reply)
John - Are you there?
[ Af:0 At:10 Sp:0] Who's the sexiest person in the world?
This isn't much better than Eliza. It drives the conversation into talking about Go, and then it doesn't have anything to say on the subject.
CmdrTaco has managed to pass for Human for several years now.
They want to pass for human. Why would they practice pretending to be Slashdotters?
Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
Computer Science/Engineering professors can't tell the difference between chatbots and Computer Science/Engineering students.
I once had a nice conversation with Cleverbot. I don't remember how it started exactly but I guess I implied that Cleverbot was simply a computer. She asked me how do I know that I am not a computer, I replied something, she asked me to define a computer, I gave her some definition (about computers being machines that process algorithms to blah blah), she said "That is exactly what a computer would say", at which point I had to confess that I had, indeed, fetched the answer from Wikipedia (Thus... done exactly what a computer might do in such a case)... It went on for a moment before she said something that didn't really make any sense... And I wasn't able to initiate another reasonable discussion. (Nor have I ever since managed to have such a lengthy conversation with her.)
In any case... I think that her replies were more insightful than what many humans would be capable of.
It really doesn't take anything more complicated than that.
Here's a bit of a "conversation" I tried to have:
Hi timr !
[ Af:0 At:7 Sp:0] Welcome, timr. Type "help" if you need it.
timr - hi, I'm timr. What should we chat about?
[ Af:0 At:8 Sp:0] Name a topic. If you can't think of a topic, here are some suggestions: "burial customs" work "new year's" telephones "hate lists" .
timr - that's a strange list of suggestions. burial customs? why that?
[ Af:0 At:13 Sp:0]
timr - cat got your tongue?
[ Af:0 At:13 Sp:0] I don't like anatomy courses. They should make you eat what you dissect.
timr - who said anything about anatomy?
[ Af:0 At:13 Sp:0] Sorry. I can't help you. You did.
...the future crusty old bastards are already drinking the Kool-Aid.
Sounds more like that student fooled the judge into thinking he was a chatbot.
It wasn't that long. Perhaps you would prefer /b/ucket , he's always exciting.
So, if I understood correctly, the judge talks to two people. A bot, and a human. It seems that in this case, the judge is not deciding on a per-case basis, but talks to everybody then figures out who's the bot by choosing the one that did the worst. So the judge getting to talk to a joker, troll or complete idiot can make even a crappy bot win the test.
That seems to be a weak test. I don't think the judge should be able to make an answer based on logic (eg, if I'm completely sure this one is a human, then even if very good the other one must be a bot). There should exist the possibility of everybody the judge talks to being a bot, or everybody being a human, which would force them to judge everybody to talk to individually.
Artificial intelligence came a step closer this weekend when a computer came within five percent of passing the Turing Test, which the computer passes if people cannot tell between the computer and a human.
The winning conversation was with competitor LOLBOT:
The human tester said he couldn't believe a computer could be so mind-numbingly stupid.
LOLBOT has since been released into the wild to post random abuse, hentai manga and titty shots to 4chan, after having been banned from YouTube for commenting in a perspicacious and on-topic manner.
LOLBOT was also preemptively banned from editing Wikipedia. "We don't consider this sort of thing a suitable use of the encyclopedia," sniffed administrator WikiFiddler451, who said it had nothing to do with his having been one of the human test subjects picked as a computer.
"This is a marvellous achievement, and shows great progress toward goals I've worked for all my life," said Professor Kevin Warwick of the University of Reading, confirming his status as a system failing the Turing test.
http://rocknerd.co.uk
In the UK the liberal (democrat) party is in government in a coalition with the Conservative party. We have a left wing (Labour) opposition party. I guess it's all about localised knowledge and your perspective. From a European perspective, the two main US parties would map to European centre-right wing (Democrats) and more right wing (Republicans). You don't have a big left wing opposition party as far as I can tell.