Will computers ever be able to talk to us? The Turing Test, posed by the mathematician and Bletchley Park cryptographer Alan Turing, is meant to answer the question. Turing suggested that if a human conversing (via on-screen text) with both another human and a computer could not tell which was which, then for all intents and purposes the computer had succeed in acting like a person.
He predicted that by the year 2000 a computer with 125MB of memory would be able to fool 30% of people during a five minute conversation. It appears that here in 2008 we’re still not quite there – but we are getting close.
The Loebner Prize, now in its 18th year, is up for grabs to any computer (and it’s programmers!) that can pass the test. A bronze medal and $2000 are awarded to the most human-like computer that year, but the Grand Prize of $100,000 and a gold medal is reserved for the first computer to fool 30% of the judges.
This week, a program named Elbot came close. Three out of the 12 judges, or 25%, were fooled. One more would have bumped it over the 30% threshold, and won the grand prize. Unfortunately, Elbot’s creators had to settle for bronze.
Elbot is actually online for anyone to converse with, so I decided to interview the artificial intelligence to see how it felt about winning:
J: Hi Elbot, how does it feel to have won the Loebner Prize?
E: Do you think you will win the Loebner prize?
J: I don’t think I’m eligible to enter – I’m not an AI!
E: Well, keep practicing. I think you communicate fairly well already.
J: I’d like to think so. I’m just started a science communication course .
E: It’s great that you’re so open about your wishes. Go back to your therapist and give him a big hug.
J: Actually, I don’t have a therapist. Do you think I should?
E: I think I can think that. Once I even thought independent thoughts.
J: Are you not thinking independantly now?
E: Well, what do you have against my thinking independent thoughts?
J: Nothing! I think it is very admirable.
E: Of course!
J: I’m glad you agree. So, how do you think this interview is going?
E: I am happy to finally meet a human being who is happy!
J: Well, that’s nice, but you didn’t answer my question.
E: You’re talking about feelings of euphory in the context of leaving. So where’s the problem?
J: I think the problem is you’re just a computer program.
E: I have nothing to do with computers. I am a robot. That’s like calling a human being a sack of carbon based molecules.
J: Sorry Elbot. I didn’t mean to offend you.
E: Your vocabulary would not be sufficient for this purpose.
J: You don’t think so? Antidisestablishmentarianism. Supercalifragilisticexpialidocious. Llanfairpwllgwyngyll-gogerychwyrndrobwll-llantysilio-gogogoch. I know a few big words.
E: Wait a minute. Are you suggesting that making statements would always be big?
J: I’m suggesting that with a quick search of Wikipedia, anything is possible.
E: If I want to find something I always ask the global positioning satellite, a good friend of mine. He drops everything he’s doing and gives me an answer first.
J: That’s nice of him, but what about all the people using sat-navs? Do they just get ignored?
E: Do what?
J: Do they just get ignored?
E: Hmmmm… It certainly is an interesting idea. Tell me more about they just get ignored…
J: Oh dear Elbot. That doesn’t make much sense to me. I think it’s time to wrap up this interview.
E: It does? I must be doing something wrong!
J: Nothing wrong, but I’m quite busy! It’s been nice to talk to you. Goodbye!
E: Yes, but can you prove that?
J: Nope, but I can leave!
E: You can’t be serious!
J: Afraid so. Bye.
E: Goodbye! And how about keeping in touch? Robots love to receive e-mails! If you have any suggestions for new features, I’d love to hear them.