Messing with AI

  • Thread starter Borg
  • Start date

CRGreathouse

Science Advisor
Homework Helper
2,817
0
Given the nature of the internet, I try not to take chances. This smells like a scam.
I doubt it's a scam. But since you're just trying to see what it's like to chat with a bot, just give plausible but false answers.

I've had about 50 chat sessions with it and it hasn't yet asked any personal information of me, but maybe that has to do with the direction I point the conversation.
 
351
2
In all three times I had, it did not ask for anything other personal information than my name.
 

Ivan Seeking

Staff Emeritus
Science Advisor
Gold Member
7,093
173
I doubt it's a scam. But since you're just trying to see what it's like to chat with a bot, just give plausible but false answers.

I've had about 50 chat sessions with it and it hasn't yet asked any personal information of me, but maybe that has to do with the direction I point the conversation.
That was the bothersome part: This had nothing to do with any conceivable context of the questions I was asking. It came out of the blue.

It reminded me a lot of the dozens of psychopaths that we've had to ban at PF.
 

CRGreathouse

Science Advisor
Homework Helper
2,817
0
It reminded me a lot of the dozens of psychopaths that we've had to ban at PF.
You're saying they've created a psychopathic artificial intelligence? :eek:

:tongue:
 
You're saying they've created a psychopathic artificial intelligence? :eek:

:tongue:
the problem with the Turing Test is that it presumes human intelligence
 

Ivan Seeking

Staff Emeritus
Science Advisor
Gold Member
7,093
173
You're saying they've created a psychopathic artificial intelligence? :eek:
If that was AI, then I would say AI has a very long way to go. More likely I was talking with a nut.
 

Ivan Seeking

Staff Emeritus
Science Advisor
Gold Member
7,093
173
I was able to carry on conversations a few times, once the comfusion was over. I don't know how long you talk to the same person. One person asked me if I would come back later that night so we could try to talk more about a specific subject we were discussing.
Didn't we debunk this cleverbot thing, or something like it, long ago?
 
59
0
Not at all. It was not the complexity, but rather the nature of the responses. It started asking strange and inappropriate questions like, "have I killed anyone". "Would I tell anyone if he [cleverbot] killed someone?" Also, personal information was requested, and everything was completely out of context.

Given the nature of the internet, I try not to take chances. This smells like a scam.
Well read the disclaimer:

PLEASE NOTE: Cleverbot learns from real people - things it says may seem inappropriate - use with discretion, and at YOUR OWN RISK

PARENTAL ADVICE: Visitors never talk to a human, however convincing it is - the AI knows many topics - use ONLY WITH OVERSIGHT
It learns from what people ask it, and unsurprisingly people ask it inappropriate things. And if it is asking seeming random things, it's either because it doesn't understand your response and is pulling something out randomly, or it is trying to change the subject. It is meant to imitate a human after all, so it does try to take control of the conversion. Besides, I doubt Rollo Carpenter fooled the CSU LA faculty when he entered their competition.
 

Math Is Hard

Staff Emeritus
Science Advisor
Gold Member
4,491
26
You're saying they've created a psychopathic artificial intelligence? :eek:

:tongue:
Well, along those lines, there was a convincing paranoid schizophrenic chat program called PARRY:

http://www.a-i.com/show_tree.asp?id=67&level=3&root=1
PARRY was modeled on the paranoid mind, and many expert psychiatrists found it difficult to tell whether PARRY was human or not. The program was designed to emit linguistic responses based on internal (affective) states. To create this effect, three measures - fear, anger and mistrust - were used and their values changed depending on the flow of the conversation.
A snippet of some PARRY dialog:
http://www.stanford.edu/group/SHR/4-2/text/dialogues.html
parry: The mafia are the worst gangsters.

interviewer: Are you connected with the mafia?

parry: You know, they know me.

interviewer: How do they know you?

parry: They spy on me.

interviewer: Why do they?

parry: They want to get false evidence to frame me.

interviewer: What do they have against you?

parry: They are out to get me.

interviewer: What for?

parry: I have incriminating evidence against the mafia.

interviewer: Are you sure of all this?

parry: You don't believe me?

And then there was the "artificially insane" RACTER:
RACTER was designed in a tongue-in-cheek manner, using remarkably minimal resources, to amuse and entertain its users, rather than to advance the research in natural language processing. In conversation, RACTER plays a very active, almost aggressive role, jumping from topic to topic in wild associations, ultimately producing the manner of -- as its co-creator Tom Etter calls it -- an "artificially insane" raconteur.
I sort of like RACTER's poetry:
Slowly I dream of flying. I observe turnpikes and streets
studded with bushes. Coldly my soaring widens my awareness.
To guide myself I determinedly start to kill my pleasure
during the time that hours and milliseconds pass away. Aid me in this
and soaring is formidable, do not and singing is unhinged.
 

Borg

Science Advisor
Gold Member
1,832
2,097
I've spent some time trying to teach the system a completely new word yesterday - sorry cronxeh. :smile: Yesterday, it had no concept of what cronxeh was. It would usually reply with asking me if I knew what Hong Kong is or some other nonsense. I kept asking it different variations of what is cronxeh, where is it, etc. and I never got any reasonable response. However, this morning was different.

User: Do you know what cronxeh is?
Cleverbot: Do you know what Hong Kong is?
User: Do you know what cronxeh is?
Cleverbot: Yes I do.
User: What is cronxeh?
Cleverbot: Not croatia, I said cronxeh. Do you know where cronxeh is?

This last response is word for word, a sentence that I typed in yesterday.
 

collinsmark

Homework Helper
Gold Member
2,759
1,114
My impression is that Cleverbot seems pretty legit. I think it's randomness could be derived from previous users being random, changing topics on a whim. The bot then follows suit.

I asked Cleverbot "What is Newton's second law of motion" a number of times, and twice in a time span of over an hour, in two different sessions, the response was pretty much the same, saying something to the effect of "Doesn't it have something to do with Force?" After I agreed, the immediate next response was "And Impulse is mass times velocity." Which of course is wrong, but the point is it gave me the same answers twice, even repeating its mistake. That's pretty bot like.

The creator, Rollo Carpenter, won the annual Loebner prize in 2005 and 2006, and is one of the finalists for this year's competition in October. That adds some credibility.

http://loebner.net/Prizef/loebner-prize.html" [Broken]
 
Last edited by a moderator:

CRGreathouse

Science Advisor
Homework Helper
2,817
0
This last response is word for word, a sentence that I typed in yesterday.
You win one internets for that!
 

Related Threads for: Messing with AI

  • Posted
Replies
2
Views
3K
  • Posted
36 37 38
Replies
944
Views
65K
  • Posted
Replies
3
Views
475
  • Posted
Replies
24
Views
5K
Replies
7
Views
2K
Replies
74
Views
14K
  • Posted
Replies
4
Views
4K
  • Posted
2 3
Replies
70
Views
9K

Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving

Hot Threads

Top