Inspired by Weizenbaum’s ELIZA, Ben Weber has created a chatbot that converses with other players in online games of StarCraft. The bot’s name is EISBot. “Once every thirty seconds,” explains Weber, “EISBot randomly selects a message from a pool of 75 messages and sends it to the console.” As the following transcript reveals, the bot’s randomly chosen messages fit seamlessly into the conversation:
EISBot: What to ally victory?
Player: damn you
Player: im not stupid
EISBot: Wanna join my clan
Player: dunno
Player: im not that great
Player: lol
EISBot: So, this game is pretty much over
Player: yeah
Player: lol
Player: im playing off race
Player: lol
EISBot: Quit wasting time, come kill me
EISBot: What kind of hardware are you using? I’m on a 486 with 16M ram, pure luxory!!!
Player: i dunno
EISBot: PWN PWN PWN!
EISBot: I’m a huge nonyt fanboy
EISBot: Have you played SC 2 yet?
Player: yeah
Player: pretty sick
EISBot: Have you ever played against a C player?
Player: nope
Player: wait
Player: yeah
Player: i lost
Player: badly
Player: haa
EISBot: Did you see the past OSL?
Player: kinda
Note that the bot’s one major flaw is that its command of the English language, particularly the use of punctuation marks, is much too sophisticated in comparison with that of the human. The sure way to distinguish the computer’s messages from the human’s is to recognize that the computer has a rather sentimental attachment to the apostrophe and the comma.
I take this as another indication that I am correct in my suspicion that when computers finally pass the Turing test it won’t be because computers have become smarter; it will be because humans have become dumber.
lol
“…because computers have become smarter; it will be because humans have become dumber.”
It’s the other way around. Humans can get away writing without punctuation because we are smart. It will be a while before computers can understand text without punctuation. Sort of like how humans can understand warped text in a Captcha, while computers find bad handwriting difficult.
If anything, slang (and text abbreviations are slang, not just laziness) might be the last reveal for determining whether an intelligence is AI or human. (Slang is now often the last reveal for determining nationality or ethnicity among English speakers.)
When bots can understand “pretty sick” then that’s when Nick should really start to worry.
This story made the rounds last year, passing the reverse Turing test by being picked as real news by a couple of aggregators: http://newstechnica.com/2008/10/13/turing-test-won-with-artificial-stupidity/
We are still trapped in the Cartesian belief that ideas can be created and communicated in a “pure” manner without the body.
This test it just tells how poor online communication can be. Trite to say it, but without a real, embodied and energetic presence we miss most of the clues.
I wouldn’t be so optimistic about slang being much help in determining “humanity”, Kevin.
It is true that the fact we are smart allows us to extract meaning from an error-ridden transmission (and the typical Web missive tends to be error-ridden to the extreme, if we consider the accepted standards of written communication elsewhere), but this is a double-edged sword.
The problem is that the less actual verifiable information the message contains, the more we are required to supply the author’s meaning ourselves. Standards for doing so have evolved, but the consequence is that the author doesn’t actually have to make his meaning clear.
In terms of a Turing test, it used to be that we saw the implementation of aforementioned communication standards (grammar and punctuation, for instance) and relation to context as the key problems. These days, the former has completely ceased to be relevant (on the Web anyway) and the latter is diluted by the fact that it’s possible to say a lot without actually saying much and still be understood.
Since we’ve gotten used to the fact that we have to distill the meaning from poorly formulated communications, the construction of a dumb communicator – based on some kind of random or somewhat contextual generation algorithm – is that much more feasible. When faced with a communication that doesn’t actually make much sense of itself, we’re inclined to bend over backwards to extract meaning from it, since we’re used to the fact that many people on the Web actually do communicate like that.
In other words, Nick’s assertion may prove true in the sense that a thoroughly stupid program can be created that will generate communications indistinguishable from those sent by a multitude of (supposedly) intelligent human beings. It will pass the Turing test (for a given value of “pass”), but will in no way represent any form of Artificial Intelligence.
What Krzysztof said.
In addition, re: “It will be a while before computers can understand text without punctuation.”
If I enter, say, “im a little teapot” into Google, I receive the same results that I get when I enter “i’m a little teapot.” So the Google computers already “understand” text missing standard punctuation as well as they “understand” text containing it. The type of understanding involved (simple symbolic decoding) doesn’t, of course, indicate any form of consciousness, much less higher intelligence, but that fact doesn’t contradict my hypothesis. In fact, it supports it.
Hey Nick
The Turing Test has been passed! Check it out. Korean Parents mistake computer companion for their own child (who sadly starved to death while they nurtured the virtual one).
http://news.bbc.co.uk/1/hi/world/asia-pacific/8600517.stm
Maybe I’m missing something, but there doesn’t seem to be a real conversation taking place with the bot.
I don’t understand the context of the game, but it appears the bot is just asking questions and making comments but not really engaging in a back and forth dialog which is the hard part.
The bot seems to be stateless.
Can someone point out what I’m missing here?
OK Nick, I get it.
The bot isn’t doing anything that a bot couldn’t have done 20 years ago. It’s nowhere near passing the test.
It’s just the people are slipping.
Right.
The funny thing is that while Weber compares his bot to ELIZA, ELIZA, developed more than 40 years ago, was far more sophisticated. It actually showed a little bit of sense.
This is really sad. StarCraft is a real-time strategy game. Anyone taking the time to construct well-formed sentences is taking time away from game play.
This just shows you are willing to bash anything related to electronic communication without taking the time to investigate for 30 seconds.