Slashdot Mirror

← Back to Stories (view on slashdot.org)

ALICE vs. ALICE

Posted by ryuzaki0 on from the no-you-suck-no-YOU-suck dept.

new-black-hand writes "The A.L.I.C.E Bot-off has produced some interesting results. It basically involved taking two instances of the A.L.I.C.E conversation bot and pointing them at each other. The results show that the bot is not really that intelligent, and relies more on human input."

7 of 325 comments (clear)

  1. Try Alice Bot vs. Infobot in IRC by antdude · · Score: 3, Interesting

    One time (last year), I managed to get Alice (a very old version) -- is there a newer version for IRC?) to talk to an infobot in IRC. It was quite amusing, and eventually Infobot stopped the conversation after five lines or so.

    --
    Ant(Dude) @ Quality Foraged Links (AQFL.net) & The Ant Farm (antfarm.ma.cx / antfarm.home.dhs.org).
  2. The default AIML set is rather dumb by FooBarWidget · · Score: 5, Interesting

    I host a modified Alicebot with a modified AIML set. The default AIML set is not that smart and only answers 60% of all questions correctly (just an estimate, based on my own experience). Apart from that, it also contains lots of redundant French and German categories.
    (My modified AIML set is much better, and answers about 75-90% of all questions correctly, I think.)

  3. Re:So.. by Anonymous Coward · · Score: 1, Interesting

    http://www.mirrors.org/archived_software/www.techk night.com/esa/default.htm

  4. uh-huh by Xzzy · · Score: 5, Interesting

    > The results show that the bot is not really that
    > intelligent, and relies more on human input."

    so do we have any proof that if a human baby is given no "input", they will grow up to be intelligent?

    Seems like someone's expecting too much out of a computer program. You can't just go into it expecting intelligence to sprout into existence. It's something that takes time to develop.

  5. Turing Shruged. by Picass0 · · Score: 4, Interesting

    Has it ever occured to anyone to do this with Deep Blue? Perhaps the only reason Deep Blue appears to be good at chess is the skill of the opponent compliments the AI. If Kasparov looked on quietly as a flunky of the game played Deep Blue, the errors in gameply might become more obvious. Kasparov would then grade the game, rather than participate.

  6. I already did this by mikey573 · · Score: 4, Interesting

    ALICE vs ALICE seems pretty redundant. Its like asking you to talk to your clone. I previously did LeknorChat vs SmarterChild, where two differently developed chat bots talked to each other.

    My purpose of doing this was to see if two such scripts would ever stop talking to each other, or would be a never ending conversation? They stopped, but please read to see why.

    Also, interpreting ASCII art is the ultimate turning test. Also discussed on my page.

  7. Re:So.. by DarkZero · · Score: 3, Interesting

    Why don't we put two copies of the same human to talk to each other, to see if that human is really intelligent?

    Considering that the human and his identical copy would probably spend at least twenty minutes doubting that it's possible to copy a human being and angrily debating about which identical person is "the real one", A.L.I.C.E. might not be that stupid after all.