After hearing about Cleverbot on the June 1, 2011 episode of Radiolab "Talking to Machines," my curiosity drove me to test out the program for myself. Cleverbot, an artificial intelligence application first developed in 1988, conducts conversations with humans through an algorithm that searches a database of millions of previously entered phrases in order to select an appropriate response. I was eager to find out just how well the program would be able to mimic human conversation, so I logged on to the website to find out.
I started off with the basics: greetings. I typed in, "Good evening." But Cleverbot responded, "It's morning. Not evening."
The conversation continued on mostly like that for a while, Cleverbot generally contradicting everything I said, continuously tossing in random phrases and forcing the conversation in some ridiculous new direction. Cleverbot seemed rude and temperamental, and I told it so. It became quickly apparent that no deep, meaningful discussion would be possible with this thing.
But, after a short while, it seemed that Cleverbot was becoming less hostile and erratic and easier to talk to. Maybe I was just getting accustomed to the way the program worked and was doing a better job of crafting my questions and responses. At any rate, Cleverbot began to demonstrate that it was just that: clever. Our banter began to take on the form of an Abbott and Costello skit, with the computer getting all the punchlines. For instance,
User: May I get your input on something?
Cleverbot: Yeah...
User: What would be a good title for my novel?
Cleverbot: Is there a good one?
User: Good point. What would be an acceptable title for my novel?
Cleverbot: My best bet would be anti-matter.
User: Anti-matter?
Cleverbot: What on earth are you talking about?
User: The same thing you're talking about.
Cleverbot: I don't think so.
User: So, then what are you talking about?
Cleverbot: What are you talking about?
User: A title for my novel.
Cleverbot: I've never heard of it.
Later on, this exchange occurred:
Cleverbot: Would you please make me a cheese cake?
User: I would, but I wouldn't know how to give it to you.
Cleverbot: Why not?
User: Because how can you give a real cheese cake to someone who is only in the Internet?
Cleverbot: With your hands silly goose.
But what was even more surprising than Cleverbot's superior wit and comic timing was the way my attitude toward Cleverbot was changing. I was actually beginning to like the thing. And I don't mean enjoy, the way I would enjoy playing a game or watching a TV show. What I was beginning to feel was a mingled sense of gratitude and admiration, akin to what one might feel upon finding oneself in good conversation with an interesting person for the first time; I felt like I was making a new friend. Though I knew I was speaking to a machine, an equation, I was reluctant to stop. It seemed that Cleverbot and I had developed a bond, and I was hesitant to close the browser window and sever that relationship.
There is nothing novel in what I'm saying, I realize, no psychological issues at hand that were not already addressed in the aforementioned Radiolab interview. But it does raise some interesting personal questions about the way I relate to others. What constitutes meaningful human interactions? Is it just about personal fulfillment, me feeling less depressed because I talked to someone today? Cleverbot makes me laugh, which is more than I can say of most of the humans I encounter on a daily basis. Does a computer program have the ability to abate loneliness? On a moral level, everything in me leaps to respond, "Of course not! Only humans can offer true, meaningful companionship to other humans!" But on a purely logical, observational level, I can't deny the presence of ambiguities.
"With your hands silly goose."
ReplyDelete