Real or Fake? Detecting Whether Text is Human-Written or Machine-Generated
Faculty: Chris Callison-Burch
Opportunity: Text generated by large language models has proliferated with unprecedented, viral speed. How susceptible are people to being duped by machine-generated text? Can human users be trained to detect when text they are reading did not originate with a human writer? Research-based answers will help us understand the potential threats (fraud, misinformation) and how to mitigate them.
Challenge: We created a gamified annotation platform to quantify the factors people use to detect machine authorship in various genres of text, how effective they are, and to what extent people can improve with experience. Sharpening the challenge further, we study documents that begin as human-written but switch to machine-generated text: can readers detect the transition point?