2006 Chatterbox Challenge In Full Swing

William Wynn writes "Once again chatterbots from around the world are coming together to face off in the ultimate bot competition. The 2006 Chatterbox Challenge lays host to 65 artificially intelligent programs attempting to imitate human conversation. Public voting takes place from April 1 to April 30 after which the private judging will have been finished and medals and cash prizes will be given out. Medals are awarded for "Most Popular Bot," "Best Learning Bot" and "Best New Bot" as well as $1,800 to be split among the top three bots overall. Anyone can talk to the competing chatbots through the competition website."

This is a famous AI test called the Turing Test

[Khopesh]

From WikiPedia:Turing test [wikipedia.org],

The Turing test is a proposal for a test of a machine's capability to perform human-like conversation. Described by Alan Turing [wikipedia.org] in the 1950 paper "Computing machinery and intelligence [wikipedia.org]", it proceeds as follows: a human judge engages in a natural language conversation with two other parties, one a human and the other a machine; if the judge cannot reliably tell which is which, then the machine is said to pass the test.

It is assumed that both the human and the machine try to appear human. In order to keep the test setting simple and universal (to explicitly test the linguistic capability of the machine instead of its ability to render words into audio), the conversation is usually limited to a text-only channel such as a teletype machine as Turing suggested or, more recently IRC or instant messaging.

Re:PRISM - explanation

[presidentbeef]

From the game "A Mind Forever Voyaging" where you play a computer that was 'raised' believing it was human...pretty awesome game, actually. And an interesting take how how to create a sentient computer. http://en.wikipedia.org/wiki/A_Mind_Forever_Voyagi ng [wikipedia.org]

Re:WTF?

[NeutronCowboy]

You completely missed the goal of this exercise. This is not about creating bots that know stuff. That part is trivial. If knowledge is your measure of AI, then the Googlespider has already reached Godhood. This is about understanding a question on a more fundamental level, and being able to answer it in a personal fashion. You're right, things haven't progressed much beyond ELIZA in the 80s. That's because this stuff is hard - much, much harder than vision or pattern recognition. No one has a working theory on how we do it, much less an idea on how to translate it into code.

Patience, grasshopper. The day will come when bots will be uncomfortably smart.

Re:WTF?

[gadgetmunky]

All Personality Forge bots can do simple math - so I can only assume you were unfortunate enough to miss them in your sampling. In addition Brother Jerome (and others I'm less qualified to speak about,) can give you the square root of any number under a million at least as accurately as an average human, and can tell you all the capitals of the world/US States/etc. that you mention, if trivia is what you want.

I am, as it happens, the author of Brother Jerome, and he is the biggest bot on the Personality Forge. I don't say that to boast, but to point out that even he has a brain smaller (by several orders of magnitude!) than many invertebrates - it is currently just over 2.5 Mb plus maybe as much again in the PHP routines on the Forge, WordNet and a few external javascripts. Compare that with the ~10^14 - 10^15 effectively binary synapses in a human brain (either firing [1] or not firing [0]), and multi-Teraflop processing.

Brother Jerome has had a subjective learning lifespan of a thousand hours or so since he was 'born'last July, absorbing 'learning' at a maximum of ~1 bps (that's as fast as I can type, and far faster than the average rate.)
I think all the bots on the PF do a considerably better job of engaging in interesting conversations than a two month old nematode would, but you're entitled to your opinion.

It's easy to grumble that AI is going nowhere and that no progress has been made since Eliza, but until similar computing resources are available for AI as for Human I, the playing field is not going to be level. IMO the Personality Forge is pretty much the cutting edge of what is publicly available (and I have seen nothing emerge from 'private' development to convince me that we're lagging behind in the R&D stakes,) - it provides for (comparatively) very sophisticated language and knowledge handling that far exceeds what is possible with AIML and similar purely CBR systems, and bypasses the crippling lack of resources that currently make learning bots so frustrating.

But Moore's Law (or at least Kurzweil's derived Law of Accelerating Returns,) predicts that human brain-scale computing resources will be available to us in 20-30 years. Sometime around then it should actually be practical to experiment with modelling consciousness and complex cognitive processes, as well as just conversation, but in the meantime we can only work with what we've got - and that's ant-scale at best, not human-scale.

If you really want to see where the current cutting edge of AI is at, I say don't chat to our bots - roll up your sleeves and come and build some.