What if you fed a machine-learning algorithm a list of ice cream flavor names and tasked it with creating new ones? What if you fed it a bunch of pick-up lines and got it to generate new ones? What would the results be like?
In episode 40 of their podcast, Lingthusiasm hosts Gretchen McCulloch and Lauren Gawne explore these questions with artificial intelligence specialist Dr. Janelle Shane. Shane’s new book is “You Look Like a Thing, and I Love You” (one of the bot’s pick-up lines). She explains how the off-sounding or whimsical results the computer produces shed light on how artificial intelligence works. It’s all about pattern matching, trial and error, and guess-and-check.
Shane’s bot also generated some surprisingly effective results, as in the case of craft beer names – do you agree? Some examples: Farm Fight; Dang River; and Devil’s Chard. The podcast also treats us to a visual chatbot’s mistakes in giraffe counting, and how the ethical problem of bias exaggeration can happen in automated resume sorting. For 44 minutes of fun, listen to their podcast here.