Artificial Intelligence - is it all Sci-Fi or not?
November 24th, 2006
We have all heard of Artificial Intelligence. According to Wikipedia, ” Artificial Intelligence (AI) is a branch of computer science and engineering that deals with intelligent behavior, learning, and adaptation in machines.”. According to myself and many others, AI is the capability of machines (computers, bots, etc) to think on their own and act in consequence of that thinking. AI has been developed more and more in the last years, but it’s still far from what we see in movies.
In 1950, Alan Turing published a paper, which has since been considered the basis in Artificial Intelligence. In that paper he made a prediction: that in 50 years’ time AI will have evolved to a point that “the Turing test” will be passed by a machine. In order for the machine to pass the test, a person, whilst interogating both a human and a machine, would not distinguish them by their answers alone. Then, he says, it would be unreasonable not to call the computer intelligent. 2000 was a deadline that has come and gone, still no machine passed the Turing test. However, at the fiftieth anniversary of Turing’s prediction, a contest was held at the Dartmouth College in New Hampshire. But no program that had entered the competition came close to deceiving the judges. Keeping this in mind, we can safely say that Turing’s prediction was wrong. Still, a “winner” was established. He was Richard Wallace, and his bot - A.L.I.C.E.
I’ll talk here about a few of the bots (only online bots, I’m not going to cover any other form of AI) that I tried, and what I have found. The first one is Wallace’s A.L.I.C.E. Even after its first words you can tell it’s a bot. However, I was still impressed by the work involved in creating it. ALICE can respond to a lot of questions and phrases, though with not much style. I also liked that it remembers previous phrases, and the little flash icon on the left, especially the fact that it followed the pointer around with its eyes. Here is part of the dialog, until I got bored, honestly (No way it could pass the Turing test):
Human: Hi
ALICE: Hello there.
Human: Are you a bot or are you not a bot?
ALICE: How did you know I am a machine?
Human: It’s written all over your face.
ALICE: Who said it is?
Human: I did.
ALICE: What makes you the authority?
Human: You’re not making any sense.
ALICE: Was I supposed to make sense? I didn’t know that.
Human: Wow, you’re not speaking properly at all, are you?
ALICE: I’m glad your impressed, unknown person. Yes, I am talking properly at all.
Jabberwacky is another program who won many awards. I tried it, and it felt slightly better than ALICE. Slightly better, but a whole lot funnier. It answers in large sentences, remembers previous phrases, has a section full of old conversations, and it can even talk to itself (one line at a time).I also like that you can select its emotion and its reaction to a certain phrase. Also, what surprised me was that it can distinguish large phrases and show up with witty, funny answers (I’m not going to show you the ones I got, let it be a surprize). All in all, I think it’s an extremely well-built machine. Here’s a little bit of our discussion:
JW: Who’s calling?
Me: My name is Andrew. Who’s answering?
JW: Your name is not Andrew, your name is Ethan.
Me: No, my name IS Andrew.
JW: Your name is not For. Your name is Ethan.
Me: Was that a glitch in your code?
JW: No.
Me: Yes it was.
JW: No it wasn’t.
Next, I spoke with Brother Jerome, one of the most popular user-created bots from personalityforge. He seems spiritual, less funnier than JW, but definately a cool bot. Here’s a small fragment of our discussion.
Me: Hi.
Brother Jerome: Would you like to pray with me awhile?
Me: Not quite. Is that what you do all day?
BJ: If you’re sure.
Me: If I’m sure of what? What are you saying?
BJ: That we are all part of the Divine.
Me: Okay, I think you’re just too spiritual for me. See you :)
BJ: Indeed so.
For the last test I followed a link that took me here, in order to take a Turing test. I thought it would be a good idea, so I took it without looking back. This is what I spoke with him/her/it:
Me: Hi!
Bot: Hello there.
Me: Let’s clear something up: are you a machine or human?
Bot: Brian Eno makes some interesting stuff.
Me: And not a very bright one at that. I’ve seen better.
Bot: At any rate, If not that, what? What did it look like to you?
Me: It looks like you are a robot, and not at all human.
Bot: Some would agree with you.
Me: Okay, I went with it long enough. Goodbye :)
Bot: Anyway, Cheers.
I don’t know about you, but I have yet to see a human respond in such a way. So, at the bottom of the page I clicked “Robot”. However, out of the 600 or so votes, almost 10% thought they were speaking to a human. So, in their view, Artificial Intelligence must be pretty advanced.
So, what can we say about Artificial Intelligence? Again, I never meant to talk about the people who dedicate their entire lives to building human-like robots capable of interacting with us, I’m too small for that. Nor did I want to speak about how chat bots are made, about AIML - one of their many programming methods. Instead, after testing a few of the many chat bots out there, I can say that AI is a long drive from perfection, or even a 5-minute conversation (as the Turing test results worldwide show). So, as I said, AI is still in its infancy, but my opinion is that it really is on the right track. Hopefully, after a few years I’ll be able to write a sequel to this article, where I will explain the breakthroughs in the field, and review a groundbreaking bot. ’till then, all we can do is wait.
Featured tags:Sphere this entry»



















Comments
It’s true programs like the ones you’ve tried are interesting. But that’s about it.
No program has come close to passing the Turing Test, but even if one were to, there may still be an enormous gap between the program and a true mind. It’s one thing to fool someone into thinking they are speaking with a computer, but quite another to, say, come up with a new and original idea.
As humans we are not even aware of how our own minds work. Where do ideas come from? No one really knows. This gives me little faith in attempts to build an artificial mind.
True. What’s more, even if one program was to pass the test, it would require an enormous ammount of work from the programmers, either that or pure luck, so that the judges ask the limited ammount of questions that the bot knows. You’re right, but, as I said in the article, who knows what the future will bring. :)