February 16, 2011

They're Smarter Than You Think

Using some much deserved free time last week, I looked through Facebook, watched a couple short videos on Hulu, and read the headlines off of MSN. Then, about twenty minutes into my surfing joy ride, I got a message from a good friend telling me to search a site called Cleverbot. Curious, I typed in the URL, and the page popped into view. Having not the slightest idea what Cleverbot was, I looked around and quickly noticed that towards the bottom of the Cleverbot webpage, two comedic warnings were posted: "Cleverbot learns from real people...use with discretion" and "Visitors never talk to a human, however convincing it is - the AI knows many topics - use ONLY WITH OVERSIGHT." Chuckling to myself, I concluded that the site was devoted to providing a simple, text-based interface between humans and a computer, so I typed a simple "what's up?" in the Google-style text box. The response and following conversation were, to my surprise, very much human.

Taken from Cleverbot.com; my words are shown in black, while Celverbot's are in blue 
After having several extensive conversations with the AI, I noticed that it had a noticeable slip up every once in a while (incoherent sentences or an illegitimate answer to my questions), but for the most part the computer's thoughts were well put together. I can't even begin to explain how impressed I am with this technology; it's amazing to think that here we are with computers that can already recognize a question proposed by the user, and provide a reasonable answer within a few moments.

Afterwards, I began to wonder just how smart this AI is in comparison to others like it. Scrolling down, I noticed a little side note underneath the conversation feed, reading "Cleverbot wins Machine Intelligence Competition 2010." To follow up on this, I decided to do a little research of my own. Similar to the infamous Turing Test mentioned in both Neuromancer and our newest book Play Money, The Machine Intelligence Competition is held annually, and allows for five teams and their artificial intelligences to compete in front of a panel of judges. The panel  will then award a prize to whichever team demonstrates the most "'progress towards an intelligent machine.'" Thus, while the Machine Intelligence Competition does not determine whether an AI can fool humans into thinking that it is another human (like the Turing Test), it does show that we have come a significant distance in the development of artificial intelligence. How far with this technology go? What will happen when computers become smarter than humans? To end with a cliché, only the future holds the answer.

Work Cited:
  • Carpenter, Rollo, JabberWacky, and Icogno Existor. "Cleverbot." Cleverbot. Rollo Carpenter, 2010. Web. 16 Feb. 2011. <www.cleverbot.com>.

2 comments:

  1. Like you, I have never heard of this website and plan to look at it after I finish this comment. It is fascinating that artificial intelligence technologies are so advanced that they are progressing towards passing the Turing Test. However, the thought that they could ever be smarter than humans is intimidating.

    ReplyDelete
  2. Your post was very interesting. Personally, I feel like once we started talking about this subject, I discover something new every day. I just saw a video with Watson AI competing in the show Jeopardy. If we can see them on TV, it means that super computers might be closer to us than we think. There is a very interesting article in Times magazine (2045: The Year Man Becomes Immortal) that goes even further...

    ReplyDelete