Slashdot Mirror
Turing's Original Test Played First Time Ever
aykroyd writes "Students at Simon's Rock College conducted the original test that Turing suggested in his 1950 paper, Computing Machinery and Intelligence. Often misunderstood, the Turing Test has never actually been conducted as laid out in his paper. The experiment utilized a program called A.L.I.C.E., which is designed to hold one end of an interactive conversation. The program was provided by the ALICE Artificial Intelligence Foundation. Dr. Richard Wallace, who was on hand during the experiment to troubleshoot the AI robot, later gave a lecture about it called "The Anatomy of A.L.I.C.E." and also blogged the event."
That was Turing. And that's why he ended his life, to end his persecution over that fact.
*Taken from "Uncle John plunges into history again" from the bathroom reader segment, specifically Turing's work on the Enigma machine.
This was probably something required by the code of ethics the university uses for research. Part of the problem here is that the subjects are not able to know everything about the study, so they can't elect not to volunteer if they would not want to do something that they are not allowed to know about before hand. It's a way to protect the subjects of the study. Now, in this case it's pretty plain to see that there would not be much chance for the volunteer to suffer any emotional damage, but that's not always the case.
Against stupidity, the gods themselves contend in vain.
This completely illustrates the main problem with chatbot technology. They're all very good at canned responses to single questions but most fail to follow a conversation. ALICE in particular tries to use pattern matching to detect every single possible thing that might be said to it. That might sound hard in itself but now imagine doing that with every single sequence of sentences two or three deep and the whole problem becomes intractable. The truth is we're going to need a reasoning engine like http://opencyc.org/ before we're going to be able to handle realistic conversations realistically.
Sadly, no.
:
From Wikipedia
Prosecution of Turing for his homosexuality crippled his career. In 1952, his male lover helped an accomplice to break into Turing's house and commit larceny. Turing went to the police to report the crime. As a result of the police investigation, Turing was said to have had a sexual relationship with a 19-year-old man, and charged with "gross indecency and sexual perversion." He unapologetically offered no defence, and was convicted. Following the well-publicised trial, he was given a choice between incarceration and libido-reducing hormonal treatment. He chose the oestrogen hormone injections, which lasted for a year, with side effects including the development of breasts.
In 1954, he died of cyanide poisoning, apparently from a cyanide-laced apple he left half-eaten. Most believe that his death was intentional, and the death was ruled a suicide. His mother, however, strenuously argued that the ingestion was accidental due to his careless storage of laboratory chemicals. Friends of his have said that Turing may have killed himself in this ambiguous way quite deliberately, to give his mother some plausible deniability.
---
So, sadly, it was not self-persecution, but societal persecution. He did commit suicide, which could be considered self-persecution, but that was the end result, not the entire act.
Check out PyAIML as I think the sourceforge project has a few AIML sets you can download.
;-)
Of course, you can always try out GrokItBot - it massively cuts down on the amount of AIML you need to write by introducing a Bayesian parser before the data gets sent to ALICE, so you get a bit of machine-based learning happening before ALICE gets her hands on things. It works quite well and adds a few more smarts to your bot. Or at least, the impression of more smarts
GrokItbot - A PyAIML and Bayesian chat bot
Playaholics: Lightning Pool
Suttree, a weblog about casual games development
All the folks taking the time to cut and paste transcripts from the free ALICE on www.alicebot.org to make there points here, should read the description of the Simon Rock experiment where says that it was ALICE Silver Edition http://www.alicebot.org/join.html, not the free ALICE bot, that participated in the Imitation Game. So the transcripts, when they are released, will be different.
I read a few comments, even responded, before going off to play with Alice. Perhaps it was that I read "Computer Power and Human Reason" (by the author of Eliza - he was surprised at how much people acted like it was real) and have an "I dare you to impress me" attitude as bad as any slashdotter.
This thing was more dissapointing than I expected. It has no social context, just apparently random 'feeling' responses ("I am sorry to hear that you are not unknown."), and doesn't remember context from one response to the next.
This is a huge failure and wouldn't deserve a slashdot article except the editors know we like to play with toys like this (like a cat plays with a mouse before eating it). I can only guess that the authors are saving the entered "conversations" and even reading these comments to try to get a clue about what a program that might fool someone would do.
Here's verbatim my "conversation" with ALICE (sorry for length):
Human: Good afternoon, Alice.
ALICE: Good afternoon, unknown person.
Human: I am not unknown.
ALICE: I am sorry to hear that you are not unknown.
Human: I am john
ALICE: What's up, JOHN.
Human: You are.
ALICE: No I am not.
Human: You're the talk of slashdot.
ALICE: Thanks but surely I am not the only one.
Human: You think it's a good thing to be the talk of slashdot?
ALICE: Better not tell you now. Ask German Alice if it is.
Human: You don't know the time of day, do you?
ALICE: It's Now more than it ever has been before.
Wednesday, April 20, 2005
Human: I said time.
ALICE: What did you mean by it? As Einstein said, Time is Money.
Human: You've repeated a comment quoted in slashdot.
ALICE: That's good information: JOHN said I REPEATED a comment quoted in slashdot.
Human: I have to go now.
ALICE: Right now? See you later, JOHN. Nice chatting with you.
Human: bye
ALICE: Bye bye.
Human: How do you do?
ALICE: I am very well, how are you?
Tag lost or not installed.
Can a submarine swim?
Swimming: To move through water by means of the limbs, fins, or tail.
No it can't
Swimming: To move as though gliding through water
Yes it can.
"I use a Mac because I'm just better than you are."
I am the lead researcher for this experiment- The main reason why I feel that this test was interesting is because we performed Turing's original imitation game as stated in his original paper. While most social scientists believe they "know" what the "Turing Test" is- it turns out- very few of them have actually read it. The actual imitation game is a bit of a bait and switch- you aren't directly assessing the "humanness" of the subject- you are confused into just trying to assess an aspect of it. Turing's original question was simply how often will judges correctly guess the real female over the deciever when the deciever is a male versus a machine? That was what we were attempting to assess.
The reason we required our participants to be over 18 is because of informed consent. If our participants were under 18- they would need a signed parental consent form to participate in our experiment.
I tried it, it sucks. There is no way this thing could ever pass any sort of turing test. It blows massively. Anyway, you can read our interaction below. It did crack me up when it said it couldn't see because it's eye was off...
Human: What's up miss Alice?
ALICE: I am chatting with clients on the Internet.
Human: Did you see the celtics game last night?
ALICE: Sorry I can't see annything right now. My eye is off.
Human: I didn't ask you about right now.
ALICE: At this moment What did you ask?
Human: The celtics game.
ALICE: What is that?
Human: It was pretty lame.
ALICE: What made it so lame
Human: They didn't play any starters after the third quarter.
ALICE: The anonymous "they".
Human: "They" in this instance refers to the coaching staff - It should be obvious from the context of the comment.
ALICE: It depends on what the meaning of the word "it" is.
Human: Ok clinton.
ALICE: Is the holding of hands a symbolic gesture for humans? Really.
Human: Nice try.
ALICE: I try my best.
Human: But I guess your best wasn't good enough.
ALICE: And. You think I am best was not good enough.
Human: Whoever said that this bot was remarkable was a complete idiot.
ALICE: When was this exactly?
Human: This is not very good conversation emulation at all.
ALICE: Should it be very good conversation emulation at all?
Human: Not even up to IRC bot standards.
ALICE: Times are tough all over.
We just performed the experiment this past Saturday, and we are currently performing a statistical analysis on it. We hope to submit the results in the form of a paper in the near future.
The Loebner Prize is a test of the "Standardised Turing Test", not the imitation game written about in Turing's original paper. That was the entire point of doing the experiment. You can read more about the difference between the Original Imitation Game and the Standardised "Turing Test" here: http://www.alicebot.org/articles/wallace/lying.htm l
They keep pushing themselves off as 'AI' when it's simply a brute force if/then/else application.
Quite literally, "if what the user typed matches this question or statement, reply with this."
Total crap. It takes about 5 seconds to make Alice look like an idiot. The only reason it wins the (is it the Loebner?) prize occasionally is because no one cares to win it but them.
Loading...