What happens when two "artificial minds" talk? The results are eerie!

At first I thought this would go in the technology fourm, but I think ultimately it is more of a philosophical issue. Watch the video and see for yourself and I think you will agree.

The folks at the Cornell University Creative Machines Lab asked a simple yet profound question: What would happen if two "chatbots" (programs designed to interact in a human-like way with humans) were made to talk to each other? The results are startling (see video). Without giving too much away, the machines immediatley begin "sniffing each other out", suspecting that the other is a machine! They then launch into a rapid-fire discussion ranging over topics including the nature of God, whether it would be nice for them as machines to have bodies, and even displays of what seems like emotion.

Are these "philosophical machines" ? semi-minds? Something else?



For more in-depth info see the project's home page:
AI vs AI: from the Cornell Creative Machines Lab

Try interacting with a chatbot yourself online here:
cleverbot.com...

That is a classic, they are even arguing like Humans. So where is part two where they start an all out war on each other, then after that have, 'Make-up" Sex...lol

reply to post by silent thunder


Neat, I love stuff like this.

Good find,S&F.

SS

Mate

This was posted recently by boncho

This Thread

I'm not banging on about not using the search. I was on there the other day and to be honest it's pretty average. I've been meaning to port a thread to the board inquiring as to when this is going to be updated.

Neverth less as I said in the post above it's very reminiscent of some threads and posts in some of the forums.

Cheers

Ha,
Sounds like some of the arguments people have here on ATS.

reply to post by myselfaswell


Oh OK...how was I supposed to get that title in search?

Anyway, thanks for bringing it up. If the mods want to merge the threads I leave it to their discretion...

I don't understand why they quickly changed the subject to God from themselves. It seems more like a joke, rather than an actual conversation between AI - Nonetheless it was very entertaining, and very creepy.

"Within 2 minutes, they’ve argued about themselves, argued about theology, and they argued about the wanting of bodies. Humanity is screwed."

Originally posted by silent thunder
reply to post by myselfaswell

 

Oh OK...how was I supposed to get that title in search?

Anyway, thanks for bringing it up. If the mods want to merge the threads I leave it to their discretion...

You have my blessing with this thread.

So long as you acknowledge those two robots are looking to take over the world. They look innocent I know...

Holy Cow..creepy is right..thanks for the thread.

Peace

reply to post by boncho



Thank you kind sir. I for one do not welcome our new self-referential overlords...

They're a couple of textual databases with a (possibly weighted) randomisation algorithm, a voice synthesis program, and a graphical subsystem strung together. That's literally all.

I am prepared to state catagorically that strong (digital/mechanical, at least) artificial intelligence does not exist; and also that it most likely never will. If you're going to call this AI, you'll want to call literally any database lookup system that has ever been written AI, as well.

The responses are generated via pure text matching. Sometimes there are wildcards or regexes used, and because of that, the results can *look* a little emergent and like true intelligence, but that is all it is. A complete illusion. What you're essentially looking at here, is a couple of very basic implementations of something like Google, bouncing search results back and forth between each other. Each result is generated by an entirely dumb, static (or mildly fuzzy in the case of a regex, as stated) search pattern.

Nothing to see here, folks. Nothing at all. Move along.

well i've talked to Cleverbot and it simply argues like that, hooking Cleverbot up to itself basically tells me nothing
it is simply programed to make these nonsense reply s to anything said

I would like to two separate ai systems speak to each other

Cleverbot is far from clever...

Barley improves on the old AIM bots.

reply to post by silent thunder


Do you know the game 20 questions? Questions asked and then answered with either yes or no or I don't know?
You try to guess the item one person is thinking of in 20 trys? Well we have a little game ball that fits in the palm of your hand and costs about three bucks.
It asks you question after question and you punch in one of those three answers up to 20 and then it tells you what you are thinking of.

My first try was on a road trip and I saw a passing telephone pole. As it asked it's questions I could find no ryme or reason to it's questions and it seemed to be getting no closer to eliminating down the probabilities like one would expect when playing with a real person. However, when it came time to guess, the little ball in my hand guessed telephone pole.

This and the video you offer kinda makes one wonder just how much of human thought and behavior is really conscious after all.

Very interesting,, and obvious that they will soon find a way to get a 'body'.
doesn't sound good for humans,,, "gods",,,
I would like to hear this go on a bit longer,,, see where they really end up,,, this was way to short,, and especially since they went straight to theology then 'Don't you want a body",,

Thank god /dess that 'she' Isn't Christian....

scary
scary
scary,,

but

intriguing and I can't wait to see what happens when the first real AI robot appears on the scene

reply to post by silent thunder


That's got to be a hoax whereby their responses were preprogrammed, I can't believe that happened like that ah "naturally", no way.. ?

reply to post by petrus4


Phew! Because I was getting a little freaked out by it. Thanks for that technical analysis.

Our conversation went like this:

Q:What is the nature of personal reality?
A:Illusion. The nature of reality is illusion.
Q:Is illusion our only resource?
A:Right now it is warm. About 20 celsius.
Q:There is human warmth as well as heat.
A:No- we're both people?


What do you think?

This isn't REALLY a conversation at all -- it's more like two "expert systems" and there are slightly different tweaks.

The FEMALE "bot" is set to be slightly contrary -- argumentative. The MALE picks up parts of someone's phrase and spits it back.

The MALE expert system also looks for contrariness to it's responses and then searches earlier responses to say; "but you said X".

The question about God and the Body are likely just "random" topics that the expert system is loaded with -- as people will sometimes change the subject. Probably there is a setting that if a conversation is "in contradiction' more than a few times, the subject would be changed -- while OLD topics can be brought up again.

The INFERENCE that "you are a robot" -- might be something that is also pre-programmed response if the response has TOO MANY SIMILAR words -- a common automated response system trick is to merely re-arrange the sentence without understanding the subject at hand.

The "we are robots" is merely a response to a correlation of two sentences where "I am X" -- so "we are X".

However, there is SOME understanding of SUBJECTS and Grammar on these systems -- they are definitely NOT having a real conversation and there is no real understanding in this simulation.

>> This is just an amusing bit of programmed responses -- and it LOOKS more like today's speech because we've grown up watching debates where people respond while intentionally misconstruing context.

Does anybody else notice that this machine BLATANTLY lies?

Or maybe it is really a unicorn and I missed that Part?

This raises questions of ethics in my mind, should intelligent machines have the ability to lie? I say no way!! That monotone voice and lack of body language would give the machine an unfair advantage!