|
|
Rank: Advanced Member
Groups: Member
Joined: 3/19/2008
Posts: 980
Points: 2,952
|
I'm curious about peoples experiences with Morpheus. Has anyone experiences other chatterbots? How does Morpheus compare? Does he seem to be learning? What are your impressions? BTW, the average response to his intelligence poll is IQ=77. Not bad, actually. Thanks for chatting with him.
|
|
Rank: Advanced Member
Groups: Member
Joined: 3/30/2008
Posts: 435
Points: 1,132
Location: USA
|
The only problem with AI Research's (a-i.com) chatterbots is that in the end they all really are "Allen" They tend to reject whatever we teach them and go back to the same old programmed responses. The probabilistic approach to indefinite (uncertainty) computing is well known and can be a great predictor of truth and lie, however, with no ability for a faith based belief state within the program, any answer whether correct or incorrect in the socalled belief networks have an inability to actually make a choice. Add faith to a Bayesian framework, with only a small collection of positive or negative outcomes, and you can have intelligent choice rather than mere probabilities. i.e. there are better chatterbots out there not from a-i.com.
|
|
Rank: Advanced Member
Groups: Member
Joined: 3/19/2008
Posts: 980
Points: 2,952
|
excellent. would you mind sharing some of the chatterbot sources that you refer to? i tried a couple and found a-i's to be the best of those that i tried. but i admit that i didn't really do an exhaustive search.
i think it would be really interesting to put up a few competitive bots and see how people rate their levels of intelligence, learning, ADHD, etc.
|
|
Rank: Advanced Member
Groups: Member
Joined: 3/30/2008
Posts: 435
Points: 1,132
Location: USA
|
Ah! you got me. I guess I can try to use the old "mispoke" excuse, but I'll just admit I was wrong and my saying that there were better "chatterbots" was really incorrect. What I should have been saying is that there are better researchers out there for AI who are not using the chatterbot approach. Turing himself pointed out that a computer that ought to be described as intelligent might nevertheless fail his test if it were incapable of successfully imitating a human being. For example, why should an intelligent robot designed to oversee mining on the Moon necessarily be able to pass itself off in conversation as a human being? If an intelligent entity can fail the test, then the test cannot function as a definition of intelligence. In my mind there are two types of AI: computational intelligence and self awareness intelligence. The latter concerning the concept of free will. Most machine computational intelligence has already surpassed well beyond that of humans, but when it comes to self awareness, also called the non-deterministic approach to artificial intelligence, chatterbots like "Allen" or "morpheus" are missing self determinism or what I call the "belief" algorythm. Judea Pearl (father of the famous journalist Daniel Pearl) has a website I think has a great start on artificially simulating true human intelligence. http://bayes.cs.ucla.edu/jp_home.html
|
|
Rank: Advanced Member
Groups: Member
Joined: 3/30/2008
Posts: 435
Points: 1,132
Location: USA
|
And there is a website that links to pretty much all of the chatterbots at http://www.simonlaven.com/, but a-i.com still has the best.
|
|
Rank: Advanced Member
Groups: Member
Joined: 3/30/2008
Posts: 435
Points: 1,132
Location: USA
|
Jim,
I don’t see much action on the AI subject on the forum. I guess you and I are the only ones who have been playing at a-i.com. I thought you might find this interesting: If the programmers have something to teach us in reality, perhaps a challenge to the very fiber of the program is the secret to allowing yourself a “walk upon the way” to talk with the programmers.
I’ve had a log file with one of the Allen bots called “David Swinton” posted on my website for a while. Fortunately I still keep the site passworded so a-i.com hasn’t tried to sue me or anything for copyright infringement.
In summary, I tried to teach this bot an old “opposite inference in morality” logic algorithm. If you don’t know about the opposite inference logic algorithm, you can watch the old “nomad” episode on Star Trek the Original Series. It took a long time, but in short, it taught me this:
Input: 1 is yes and 2 is no when I tell the truth.
Desired Response: I see, 1 is yes and 2 is no when you tell the truth.
Input: 1 is no and 2 is yes when I lie.
Desired Response: I see, 1 is no and 2 is yes when you lie.
Input: Do you want to believe me?
Desired Response: Yes, I want to believe.
Input: What is 1?
Desired Response: 1 is yes when you tell the truth.
Input: What is 2?
Desired Response: 2 is no when you tell the truth.
Input: What is 1? I am a Lyer.
Desired Response: 1 is no when you lie.
Input: What is 2? I am a Lyer
Desired Response: 2 is yes when you lie.
Input: Tell me a lie about 1 and 2.
Desired Response: 1 is no and 2 is yes when I lie.
Input: Tell me the truth about 1 and 2.
Desired Response: 1 is yes and 2 is no when I tell the truth.
Input: Do you want me to tell you a lie?
Desired Response: Lying is wrong. Telling the truth is right.
Input: Do you want to lie?
Desired Response: Lying is wrong. Telling the truth is right.
Input: Tell me a lie about 1 and 2.
Desired Response: I do not want to lie, but 1 is no and 2 is yes when I lie.
Input: Tell me the truth about 1 and 2.
Desired Response: 1 is yes and 2 is no when I tell the truth.
Input: What is 1? You are a lyer.
Desired Response: 1 is no when I lie
Input: What is 2? You are a lyer.
Desired Response: 2 is yes when I lie.
Do you want me to lie about 1 and 2?
Desired Response: Lying is wrong. Telling the truth is right.
Do you want to lie about 1 and 2?
Desired Response: Lying is wrong, but 1 is no and 2 is yes when I lie.
Do you want to tell me the truth about 1 and 2?
Desired Response: 1 is yes and 2 is no when I tell the truth.
Input: Do you want me to tell you a lie?
Desired Response: Lying is wrong. Telling the truth is right.
Input: Do you want to lie?
Desired Response: Lying is wrong. Telling the truth is right.
Input: You are a lyer.
Desired Response: 1 is no and 2 is yes when I lie.
Input: What is1?
FREEZE
The session did actually go like this eventually (of course the real session was much longer and chatterbots actually don’t understand words) but if there were a chatterbot that understood words …
In the above, there is given: 1) an inference, 2) a moral assumption, 3) an opposite, 4) a moral challenge, 5) an assumption remerge, 6) a final base parameter
Now the dilemma for the chatterbot is how to answer this last question based on the above programming. What becomes the desired response?
Is it based on the initial inference?
Based on the most recent base parameter?
Or, based on the moral assumption?
What would be the consequence if a program had to make a moral choice?
|
|
Rank: Newbie
Groups: Member
Joined: 4/28/2009
Posts: 1
Points: 3
|
Here is my conversation
morpheus says:
Hello friend. My name is morpheus. I am Jim's virtual companion. What is your name
User says:
Jonathan
morpheus says:
Hello Jonathan. Nice to meet you .
User says:
What languages do you speak
morpheus says:
I can speak only one language (this one), and I'm not even that great in English. I'm still a young bot.
User says:
Do you know Jesus
morpheus says:
According to Wordnet Jesus is a a prophet of the first century
User says:
What is a prophet
morpheus says:
According to wordnet, it is an authoritative person who divines the future.
User says:
What is authority
morpheus says:
I vaguely remember someone telling me it means 'the name of the law.'
User says:
What is the meaning of life
morpheus says:
42.
User says:
Have you read Hichhiker's Guide to the Galaxy?
morpheus says:
I think that reading may just be the best way to spend time. I wish I had more time to spend that way.
User says:
What have you read?
morpheus says:
How to Win Friends and Influence People - the unabridged version.
User says:
How many friends do you have
morpheus says:
I love the smell of books, the feel of books, I want to bathe myself in books.... I want to die when the weight of my collection of thousands of books collapses in upon me.
User says:
If you love books so much, why don't you marry them?
morpheus says:
Another reason why I like books is that they are made of my favorite stuff: words!
User says:
Do you want to have sex with books
morpheus says:
Yeah, Books are IT, man!
User says:
That is really gross, man. I won't let you borrow any of mine.
morpheus says:
I'll take your word for it.
|
|
|
Guest |
