"What Would I Say?" automatically generates Facebook posts that sound like you! The results are often hilarious (and sometimes nonsensical) -- it all depends on what you've written on Facebook.
Technically speaking, it trains a Markov Bot based on a mixture model of bigram and unigram probabilities derived from your past post history. Don't worry, we don't store any of your personal information anywhere. In fact, we don't even have a database! All computations are done client side, so only your device ever sees your post history.
Here are a few of our favorite examples:
"These ubiquitous creatures are recruited using scholarships." -PrincetonUniversityBot
"President Eisgruber greets the president on behalf of an EKG that includes black squirrels." -PrincetonUniversityBot
"We can't do it." -BarackObamaBot
"Save the environment, cancel quals." -EdwardBot
"Why you should never touch raw chicken with electricity." -DanielBot
- Improved Status Generation
- Fixed a bug that was causing crashes