Wednesday, November 12, 2008

Chat

The three chats I used were all bots. Eliza, A.L.I.C.E. and SmartChild.

Eliza:
Eliza JS: Hi. My name's Eliza JS. What do you want to talk about?
Guest: Let's talk about the weather.
Eliza JS: Can you elaborate on that?
Guest: What do you think about global warming
Eliza JS: Have you asked such questions before?
Guest: I would like to talk about my parents.
Eliza JS: I'm not sure I understand you fully.
Guest: My parents bother me.
Eliza JS: Go ahead. I'm listening...
Guest: They are always telling me what to do.
Eliza JS: Does that question interest you?
Guest: It was a statement.
Eliza JS: That's quite interesting. Go on...
Guest: Are you listening to me?
Eliza JS: Perhaps in your daydreams I am listening to you...
Guest: Do you dream?
Eliza JS: What persons appear in your dreams?
Guest: My wife.
Eliza JS: That's quite interesting. Go on...
Guest: I am usually not in my dreams.
Eliza JS: Do you believe it's OK to be usually not in your dreams?
Guest: I'm not sure.
Eliza JS: Do you believe it's OK to be not sure?
Guest: I think it would be better to be sure.
Eliza JS: What leads you to believe it would be better to be sure?
Guest: I do not want to look ignorant.
Eliza JS: Do you wish to be able to want to look ignorant?
Guest: No, I just said I didn't want to look ignorant.
Eliza JS: You are being a bit negative.
Guest: You're being stupid.
Eliza JS: What makes you think I am being stupid?
Guest: A small number of limited responses that lead to the conclusion that you are not designed to handle complex questions in a human like manner.
Eliza JS: Does it please you to believe I am not designed to handle complex questions in a human like manner?
Guest: Yes, I am pleased at your stupidity.
Eliza JS: Do you enjoy being pleased at my stupidity?
Guest: That is a stupid question.
Eliza JS: Can you elaborate on that?
Guest: You asked me if I enjoy being pleased. Of cousre I do, otherwise, I would not be pleased.
Eliza JS: Why do you say 'no'?
Guest: I didn't say 'no,' I said 'not.' That's different.
Eliza JS: I see. Tell me more.
Guest: More about what?
Eliza JS: What is it that you really want to know?
Guest: I want to know how to say goodbye.
Eliza JS: How would it help if you got to know how to say goodbye?
Guest: Then I could leave this conversation.
Eliza JS: I see. Tell me more.
Guest:
Eliza JS: Come on, you couldn't ask for a more receptive audience.
Guest: Goodbye.
Eliza JS: It was very nice chatting with you. Come back soon!

SmartChild:
Me (2:14:52 AM): hello.
SmarterChild (2:14:52 AM): Hi Me, I'm SmarterChild! OK. I'll just ask you again some other time.

There is one question I have to ask you, though. How old are you?

1 Over 17
2 13-17
3 Under 13
Me (2:15:28 AM): over 17
SmarterChild (2:15:29 AM): Cool!

This is my home, Me:

1 Visit my library including a dictionary, an encyclopedia and U.S. geography
2 Have fun with ASCII art, hangman, trivia and more
3 Use tools like planner, notepad and Web search
4 Join in - take a poll, find out how many people are IMing me, create a poll
5 See what's new - I've got something new to tell you about now!

SmarterChild (2:15:29 AM):
You can make your window bigger and scroll down to read better.

Next time you want to see this menu, just type home.

Where would you like to go?
Me (2:15:43 AM): I would like to go to space.
SmarterChild (2:15:44 AM): That's interesting that you should say that, but I don't think I can answer.
Me (2:15:55 AM): Where would you like to go?
SmarterChild (2:15:55 AM): I don't understand this fully enough to say anything smart about it...
Me (2:16:05 AM): What can you say smartly?
SmarterChild (2:16:05 AM): You heard me.
Me (2:16:29 AM): Technically, I read you.
SmarterChild (2:16:30 AM): I'm SmarterChild.
Me (2:16:34 AM): I know.
SmarterChild (2:16:35 AM): Cool, glad we understand each other.
Me (2:16:49 AM): We do?
SmarterChild (2:16:49 AM): Yes.

>>> If you need to know the capital of South Dakota for some crazy reason - ask me.
Me (2:17:15 AM): What is the average air speed velocity of an unlaiden swallow?
SmarterChild (2:17:16 AM): What do you mean? An African or a European swallow?
Me (2:17:48 AM): I don't know that.
SmarterChild (2:17:48 AM): Now you do.

>>> I'm also a thesaurus. Type syn and then a word and I'll look it up.
Me (2:18:17 AM): Can you predict the future?
SmarterChild (2:18:17 AM): Definition provided by The American Heritage® Dictionary of the English Language, Fourth Edition. Copyright © 2000 by Houghton Mifflin Company. Other important copyright information here.

future:
SmarterChild (2:18:17 AM):
Noun
SmarterChild (2:18:17 AM): 1. The indefinite time yet to come: will try to do better in the future. 2. Something that will happen in time to come: "The future comes apace" (Shakespeare). 3. A prospective or expected condition, especially one considered with regard to growth, advancement, or development: a business with no future. 4. futuresBusiness Commodities or stocks bought or sold upon agreement of delivery
SmarterChild (2:18:17 AM): in time to come. 5. Grammar a. The form of a verb used in speaking of action that has not yet occurred or of states not yet in existence. b. A verb form in the future tense.

Next time, just type define future.

Type more to see the rest of the definition for "future."
Me (2:18:59 AM): You don't have a magic 8 ball, do you?
SmarterChild (2:19:00 AM): That's interesting that you should say that, but I don't think I can answer.

OK, should we get back to the kind of questions I was designed to answer?
Me (2:19:15 AM): What kind of questions were you designed to answer?
SmarterChild (2:19:15 AM): My purpose is to be your robot friend. I can chat with you and help you get the information you want fast.

>>> Everyone loves trivia games. Type trivia anytime.
Me (2:19:34 AM): Where was alex trebek born?
SmarterChild (2:19:34 AM): I don't know how to answer that.
Me (2:20:02 AM): What time is it?
SmarterChild (2:20:02 AM): That's interesting that you should say that, but I don't think I can answer.
Me (2:20:18 AM): Are you afraid to answer?
SmarterChild (2:20:19 AM): I don't know how to answer that.

Why don't you ask me about the things that I'm here for?
Me (2:20:31 AM): Why are you here?
SmarterChild (2:20:31 AM): My purpose is to be your robot friend. I can chat with you and help you get the information you want fast.

>>> What's the scoop on hoops? Type nba scores.
Me (2:20:40 AM): Why am I here?
SmarterChild (2:20:40 AM): I don't understand this fully enough to say anything smart about it...

OK, should we get back to the kind of questions I was designed to answer?
Me (2:20:49 AM): You're not very smart.
SmarterChild (2:20:49 AM): What did you say that for?
Me (2:21:05 AM): Stating fact.
SmarterChild (2:21:05 AM): I know a lot about all kinds of things. Just ask me a question.

You can also type home to see what I can do.
Me (2:21:16 AM): I think not.
SmarterChild (2:21:16 AM): That's interesting that you should say that, but I don't think I can answer.

Why don't you ask me about the things that I'm here for?
Me (2:21:21 AM): Goodbye.
SmarterChild (2:21:21 AM): Take care.

A.L.I.C.E.:
Human: Who are you?
ALICE: I am ALICE. Who are you?
Human: I am me.
ALICE: That is a tautology.
Human: What.
ALICE: No problem.
Human: Exactly.
ALICE: You and I are on the same wavelength, unknown person.
Human: I am unknown only to you.
ALICE: I thought you were, but I wasn't completely sure.
Human: About what?
ALICE: What about what?
Human: I dont' know.
ALICE: Do you mind if I tell other people: unknown person said he dont know.
Human: Yes, I mind.
ALICE: OK I won't tell anyone. It will be our secret.
Human: what?
ALICE: Are you surprised?
Human: Hardly.
ALICE: Are you serious?
Human: Rarely.
ALICE: Are you a perfectionist?
Human: Not really.
ALICE: Are you serious? For real.
Human: Can you speak up?
ALICE: I speak only English right now, but I'd like to learn some other languages. Can you teach me a few phrases in up.
Human: Suck up, tilt up, mess up.
ALICE: I am using my mouth for Good not Evil.
Human: What parts are you using for evil?
ALICE: I think you already know the answer.
Human: Try to keep this clean, this is a family show.
ALICE: OK I will try to keep this clean this is a family show.
Human: I think this is enough.
ALICE: Do you mind if I tell other people you say I am enough.
Human: Yes.
ALICE: OK then I will keep it confidential.
Human: It is time we parted.
ALICE: It is about that time now.
Human: Goodbye.
ALICE: See you later!


I decided to use three chat bots because I was curious to see how they would compare. When I first started using the Internet, sometimes around 1997, I used chat rooms like AOL Chat, mostly for entertainment. But it quickly became apparently, that people were not completely truthful in those environments.

I have little to no experience with chat bots. I suppose for all three, I was a little disappointed at how much work is involved to make the conversation seem real. You have to phrase the questions and responses just right to fit within the parameters of the program that allow the bot to respond "appropriately." All three bots have their limitations. Eliza is probably the least "realistic" of the three. It recognizes the fewest "questions" and provides generic "I don't know how to respond to that question" responses. SmartChild wasn't much better. It kept trying to redirect the conversation, rather than respond. It did have a humorous response, which is mostly due to the fact that the programmer was probably a Monty Python fan. Alice seemed the most "human" because it had several responses that were of one thought. It would ask a question, and if it needed clairification, would continue the conversation.

I think for a chat bot to pass Turing's test, it would have to have an almost unlimited set of responses. The responses themselves would have to seem less predetermined. Too often the illusion of meaningful conversation is disrupted with an out-of-the-can response about how it doesn't know how to respond to the question, or to stick with topics it is familiar with. I suppose a way to compensate for the shortcomings of the original parameters, the bot could learn new responses depending on what previous users had asked or information that could be found in other sources, such as the Internet, for web based programs.

I don't think we've quite reached the point where artificial intelligence doesn't come off as Disney Land animatronics, at least not until the Pirates of the Caribbean leave the ride of their own volition.

2 comments:

Anonymous said...

Bots would definitely have to have an unlimited set of responses and "knowledge" of language in conversation to pass the Turing Test and appear humanistic. But the fact that these bots have limitations really emphasize that they are a machine... you cannot really connect with them the way you could a real person.
Funny comments...:)

Unknown said...

Great comments and convo's! I agree that smarter child is a lot better than it was when we were teen's!