BROOKE: Decades of research has not yet yielded a machine capable of delivering a tight ten minutes of observational humor, but computers can, under certain circumstances, grind out a joke. Recently, Joel Warner and Peter McGraw, authors of The Humor Code: A Global Search for What Makes Things Funny, pitted Comedian Myq Kaplan against Manatee, a computer created to tell jokes. Joel Warner is going to see if I can tell which joke is human and which isn't.
JOEL WARNER: Okay, you’re gonna guess now.
WARNER: I like thrills like I like flights, cheap.
BROOKE: Cheap thrills, cheap flights, I think that since “cheap” in this context is being used as kind of salacious sexy, that that was the human, Myq.
WARNER: You are incorrect.
BROOKE: Oh! [LAUGHS]
WARNER: That was Manatee.
BROOKE: All right, another.
WARNER: I like fingers like I like my notes, sticky.
BROOKE: That’s the computer.
BROOKE: Actually, artificial intelligence researchers have been studying computers and humor for a long time.
WARNER: Yeah. Among my favorites would be LIBJOB, which is the light bulb joke generator, the double entendre noun transfer program –
- which figures out where inside passages of text to insert the line, “That’s what she said.”
BROOKE: [LAUGHS] The current effort, the one that yielded Manatee, was a, a $700,000 government project?
WARNER: Yes. The project received stimulus funding, and when Republicans heard that a joke computer received $700,000 they were fuming.
BROOKE: How can you justify that?
WARNER: I think there’s a valid reason. We spend so much time with our gizmos, we want the interactions to be as smooth and fluid as possible. In some ways, we want these things to almost be a bit human. I mean, who wouldn't want a GPS device that, after you ignore it for the 20th time, it somehow sarcastically rolled its eyes at you?
BROOKE: [LAUGHS] I would say almost everybody wouldn't want that. But I'm sure there's another example. [LAUGHS]
WARNER: Well, think about when a webpage doesn't load now.
WARNER: Designers think we might be a little more okay with a stumble, if a computer almost makes humor about it.
BROOKE: I guess if it were real. Personifying a program can also run into a lot of problems - I mean, Clippy, may he rest in peace, was probably the most deeply unpopular thing ever produced by, what, Microsoft?
WARNER: If Clippy made fun of himself, said, hey, I know you don't like me –
- but you pressed the wrong button and here I am – I’m going to do my little Clippy dance, I mean, that – right, that might help make things a little more –
BROOKE: You are – you are so right.
WARNER: One of the examples of human intelligence is our ability to process all this information around us and find new and novel ways to make things funny, so it’s actually a really good challenge, a really good goal.
BROOKE: Do humans really understand what makes things funny?
WARNER: That's a very good question. There have been many attempts over the centuries to figure out what exactly makes things funny. You can go back to Plato and Aristotle, who said it was about superiority, laughing at the misfortune of other people. Then came the incongruity theory by folks like Kant, who said that humor arises when you’re lead to believe one thing and something else happens; setups and punch lines work well.
And then there was Freud, of course, who had his own theory.
BROOKE: [LAUGHS] Based on sexual frustration, no doubt.
WARNER: Of course.
BROOKE: And do you have a favorite?
WARNER: I have to say I’m partial to the one my co-author and I looked at in the book, The Humor Code. His name is Peter McGraw and his theory is something called the Benign Violation Theory, the idea that humor arises when something seems benign, while simultaneously appears to be a violation or somehow wrong.
BROOKE: Of course, all of these can be right.
WARNER: Yeah. There haven't been enough hard and fast research to say this is it, this is the end-all, be-all of humor research.
BROOKE: Okay. So a guy is going down the street and he sees his friend. He hasn't seen him for I don't know how long. And he has this big orange head. And, and he goes up to him and he goes, Hey, what’s with the big orange head? And he goes, you know, it’s a funny story. I was, I was in an antique shop and I found this lamp and I, I rubbed it and a genie came out, it gave me three wishes. And so, I wished for a gorgeous house and, and you see behind me this huge mansion. Yeah, it’s really nice. That’s the house. And then I wished for a beautiful wife, and, and you see that really lovely blonde coming down the street. That’s my wife! And, and then here’s where I think I went wrong. I wished for a big orange head.
WARNER: [LAUGHS] The amount of kind of human intelligence that you and I have to have to 1) tell that joke and create that joke and then appreciate it, I mean, can you imagine?
BROOKE: The smartest person I know did not get that joke.
WARNER: I think it’s less about pure intelligence but I mean, how you process the world, how you see the world. I mean, how are you gonna teach a computer to do that?
BROOKE: [LAUGHS] But if we were to do that, would that be the apotheosis of artificial intelligence?
WARNER: Some people have suggested that one of the ultimate tests is to get a computer to tell jokes and appreciate jokes in all their forms. I think we’re still a ways off on that.
BROOKE: What was the best joke told in the contest?
WARNER: Okay. I’d have to say the one that sums the whole contest up, “I like jokes like I like robots, efficient and killer.”
BROOKE: And whose joke was that?
WARNER: Thankfully, for the human race, that was Myq Kaplan.
BROOKE: [LAUGHS] Joel, thank you so much.
WARNER: Thank you for having me.
BROOKE: Joel Warner wrote, with Peter McGraw, The Humor Code: A Global Search for What Makes Things Funny.