A typically bizarre result from What Would I Say. (screenshot )
What WOULD you say?
A new Facebook app developed by Princeton students in a 36-hour hackathon is making the rounds, and it’s eerily accurate at copying how users talk on the social networking site. What Would I Say asks users to login using their Facebook, and using fancy terminology like bigram and unigram probabilities, trains a “bot” to mimic any given users way of speaking.
Past comments and status messages are mashed up, and the results range from dead-on accurate to hilariously bizarre. Here’s some examples, using some guy named Mick Swasko’s previous statuses and comments:
“Ignoring the obvious of what's wrong, here's a position at RedEye as beat drops on iPod.”
“Years later, somehow, Aladdin's on.”
“Had a coworkers mom call into a temporary coma and maybe the ass. Their daughter, Neelu, was born at the very beginning when he knows you guys.”
“Riot Fest tonight is the third large cup of coffee machine.”
The Los Angeles Times reports the developers of the app were inspired by old AOL Instant Messenger “bots” that would respond automatically to chats. Users on Twitter Monday were using the hashtag #wwis to post their favorite results.
Try your own, and tag @redeyechicago with #wwis to share your best results. Warning, it gets really addicting.
Want more? Discuss this article and others on RedEye's Facebook page.