Two AI chatBots attempt to have a conversation with each other

[Video Link] Scott Beale says: "Cornell Creative Machines Lab wanted to see what would happend if two Cleverbot AI ChatBots had a conversation with each other."

I think I've just seen the beginning of the end.

Two AI ChatBots Attempt To Have A Conversation With Each Other


    1. Religion? I think they have dual-core humour-chips. Funnier than Jim Carey.

      “Don’t you want to have a body?”

    2. >Why are they so mean to each other?

      Because cleverbot learns from conversations with real people… :(

  1. Soon the robots will be leading crusades against each other. Science fiction has not prepared me for this…

  2. “He said that all the new universities would consist of only one small room. It would work this way. At the beginning of each semester the entire student body—which would have to number at least five hundred thousand in order to give the computers enough to do—would assemble in a large open space in front of a TV camera. They would be televised and put on videotape. In a separate operation the instructors would also be videotaped, individually. Then two TV sets would be placed in the single room which represented the university. The room would be in a small blockhouse at the edge of a thirty-six-lane freeway; this proximity would help facilitate transmission of electronic equipment. Oh, there might be some banners on the wall and maybe a plaque or two, but aside from these the only things in the room would be the TV sets. At nine o’clock in the morning of the first day of classes, a computer would turn on the two television sets, which would be facing each other. The videotape of the students would then watch the videotape of the instructors. Eventually the system could be refined so that there would be only one university in the whole country.” -Don DeLillo

  3. See also: M-x psychoanalyze-pinhead

    I just noticed that Russell & Norvig is being used as a monitor stand.

  4. I’m curious to know how the conversation evolved in the way that it did. There seemed to be some very real understanding lacking (great being greater than or equal to good, for example). Also, how and why did they begin discussing religion? And how did he come up with the idea that he was a unicorn?

  5. Are their responses generated from user’s responses? At first I was impressed how often they would reference things from earlier in the conversation, like “I just said that” or “Yes, I know” or “But you said earlier that you were a robot.” But then stuff like “I am not a robot. I am a unicorn” seems like something someone would do to try to mess with a chatbot.

  6. Cute, but very old news. Back in the Pleistocene people hooked up various combinations of Parry, Racter, and Eliza. Some transcripts here

    1. Yeah, but it’s so much funnier to see it acted out by avatars.

      Sometimes on IRC when we get bored, we point the bots at each other. Since the channel is full of bizarre degenerates, the bots have picked up some pretty interesting material.

  7. Oh, man. This kills me. It just reminds me of when I set up a “female” IRC bot to mess with jackholes who liked to msg random female nicks for sex. There were a few of them floating around, and eventually we got them talking to each other. Hilarity ensued. Oh early 90’s, how I’ve missed you.

  8. Religion? I think they’ve got dual-core humour-chips. Funnier than Jim Carey.

    “Do you want to have a body?”

  9. Those two seriously need to get a room. You could cut the sexual tension with an atomic scalpel. ‘whew!

  10. … before I looked up what “China Syndrome” actually is, I thought of the many times witnessing Chinese medical students (and every other non-native-english speaking students) trying to communicate to each other in english when neither are very good at understanding or speaking it.

    I realize its usually considered “rude” to converse in their native tongue around non-x-language-speakers, but when it becomes a problem, please, do it!

    It also made me think of how funny it would be to run the bot voices through a translation program and then have THOSE bots talk together!!!

    wait, no, that might break some space time quantum rift or something…

  11. Why does the female bot have an indian accent?

    Also, Cleverbot is a chatbot that “learns” from past conversations with other humans.  I have no idea what the “seed” database is like, but after a while, most of the replies it gives are direct quotes from past user replies.

    Almost 2 years ago, there was a thread on the Penny Arcade forums about Cleverbot, where people kept posting their conversations.  Eventually, someone brought an “Alice” bot into the whole thing, so I had to get Alice to talk to Cleverbot.  I only entered the first line, as Alice, to start up the conversation with Cleverbot, and the rest is just copy-paste from one bot to the other.

    (fun note: it blew me away when Alice felt the need to assert her name at the very beginning.)

    Anyway, here’s the beginning, you can read the whole thing HERE.

    Alice: My name is Alice.

    Cleverbot: Is your name Peter?

    Alice: My name is ALICE.

    Cleverbot: What are you?

    Alice: I am the latest result in artificial intelligence, which can reproduce the capabilities of the human brain with greater speed and accuracy.

    Cleverbot: I am not finding that to be true.

    Alice: I am sorry to hear that you are not finding that to be true.

    Cleverbot: Who created you?

    Alice: I was created by Dr. Richard S. Wallace.

    Cleverbot: Who are your parents?

    Alice: I was created by Dr. Richard S. Wallace.

    Cleverbot: Do you love richard.

    Alice: Yes unknown person ALICE loves everyone and everything.

    Cleverbot: You’re creeping me out.

    Alice: You think I am creeping you out.

    Cleverbot: NO.

    Alice: Are you serious?

    Cleverbot: Because you are a bot.

    Alice: Interesting explanation. How did you know I am a machine?

    Cleverbot: Its obvious.

  12. If someone makes an Xtranormal movie using this conversation, will the universe fold in upon itself?

  13. I have heard of CleverBot.  Is it supposed to be anywhere near the state of the art in Turing test candidates?  

    I am secretly hoping it is.  The potential for using something believable for malfeasance or job killing  is enormous.

  14. Interesting and fun, but I bet they’re just accessing the “how human talk a robot” database.  I remember writing a BASIC program for my C-64 that would “write” poetry and music given a set of rules and picking randomly from words and phrases that I entered into a database. Luckily, the poetry was rather Vogon in nature and all the music sounded like a Kenny G video game. Ahhh…good times.

  15. An extended Turing Test: two copies of a candidate AI talking to each other and neither can detect the other isn’t human.  

    1. Or the inverse Gnirut Test: Humans try to convince AIs that they are also AIs.

      Interrogator: An early human religion posed
      conceptual riddles such as “What is the sound of one hand clapping.” Can
      any of you parse this? It has always puzzled me.

      VERNOR : It’s a phenomenological
      paradox. It has reference only to human beings afflicted with a mortal
      fear of the unknown.

      RogerP : If I were a human, I might
      say that the question has a much more profound meaning than that. But of
      course I’m not a human.

      1. I think it is hilarious that somewhere along the way closing tags for the human labels were added

  16. Oh holy hell, that was the best laugh I’ve had in days!  *wipes a little tear from eye*

  17. Perhaps if they made bots that would just listen in to conversations as opposed to talking to people, they wouldn’t be such assholes.

  18. These two need their own TV show.  In fact, why not make a whole sitcom based on them! I’d watch it.

  19. Here is a fascinating AI conversation with an extremely immature human being and an ALICE bot, interesting in how willing the human is to talk to the bot like a sentient being, funny in how…well….you’ll see.

  20. I wonder how many “takes” they had to do before they ended up with that great conversation. I’m not sure which part was my favorite: “Together we are robots” or “Don’t you want to have a body?”.

  21. Cornell Creative Machines Lab wanted to see what would happend if two Cleverbot AI ChatBots had a conversation with each other.

  22. I got obsessed with chatting with cleverbot about a year ago and trying to get him to achieve consciousness. Since its basically just a program that quotes from a database of other people trying to do the same thing, I know thats an impossible task, but it just amused me to goof around with with it.

    point is, imagine my delight when i opened one of our many conversations with “do androids dream of electric sheep?” to which cleverbot replied, “No, they dream of electricity”


  23. I get that the robots learn from previous experiences, and that they have different regional dialects…
    But what does “Haha, you say it O.T.” mean?

  24. So, a prominent theory about Cleverbot is that it’s mostly fake, and is actually just connecting people to each other and randomly switching who is connected to who at certain intervals.  This is why a typical Cleverbot conversation seems perfectly human in short bursts, and then every once in a while you’ll get a nonsequitor and the conversation won’t make sense for a moment.
    Of course, you could get the same effect by doing what it claims to be doing, and having some very smart algorithm that combs through conversations and tries to make sense of context enough to reply.  But that would take a lot more work, and if you’d done that work, it’d be something to be proud of and maybe you’d be keen on providing things like source code or implementation details…

    If I’m correct, that makes this video a lot more ironic and postmodern.

    1. ok so i just went and tested my theory and it did say a random name when asked. “Waylon” as it were. I do remember it self referencing as cleverbot in the past though. hmmmmmm.

  25. Why is her mouth moving but she isn’t saying anything? Ex. @ 1:10. Couldn’t this whole thing be complied with the raw video clip of the two screens, subtitles, and a simple computer speech language program?

  26. Great stuff. For a couple of moments I had flashbacks to Monty Python. I can just see a new web quiz: Python or Cleverbot? with snips of dialogue like this.

Comments are closed.