February 12, 2015

If you're a little clueless when it comes to flirting, you're not alone — at least among other humans. But computers — especially one designed at Stanford — may have flirting down to a science.

Researchers there listened to almost 1,000 speed-dating sessions to understand how men and women talk when they’re trying to be flirtatious, and picked up on the patterns.
 
“Guys, when they are flirting, tend to speak faster and at a higher pitch, than when [they] are not flirting," explains Melissa Dahl, who covered the subject for New York Magazine. "Women, like men, tend to laugh more when they’re trying to flirt, and they also use more sexual words."
 
But neither men nor women are particularly good at picking up on each other's cues. Women “were able to judge with about a 62 percent accuracy when a man was flirting. Men were worse — they accurately guessed that a woman was flirting with them about 56 percent of the time.”
 
After identifying how people talked during the dating sessions, the scientists then taught the language of dating to a computer using their “flirtation-detection system.”
 
So how well did the computer detect flirting?

Much better than your average bachelor or bachelorette — the software correctly judged flirting with 71 percent accuracy.

Dahl has one theory about why the technology has an edge: “We project our own behavior or our own intentions on the person that we’re with — so if I’m trying to flirt, then I’m more likely to say ‘oh, this person was flirting with me.’”
 
That’s right — computers are better listeners than your new crush. Or at least Stanford’s “flirtation-detection system” is.

Melissa Dahl, Sci and Tech, Stanford University, dating, communication

Previous Post

The Science Evangelist

Next Post

The People Powering AI Decisions

comments powered by Disqus