A faster, better way to prevent an AI chatbot from giving to

A new technique can more effectively perform a safety check on an AI chatbot. MIT researchers enabled their model to prompt a chatbot to generate toxic responses, which are used to prevent the chatbot from giving hateful or harmful answers when deployed.

Related Keywords

Idan Shenfield , Aldo Pareja , Tsun Hsuan Wang , Zhang Wei Hong , Pulkit Agrawal , Prakash Srivastava , Us Army Research Office , Improbable Ai Lab , Watson Ai Lab , International Conference On Learning Representations , Amazon Web Services , Us Air Force Research Laboratory , Us Office Of Naval Research , Quanta Computer Inc , Artificial Intelligence Laboratory , Language Systems Group , Us Air Force Artificial Intelligence Accelerator , Hyundai Motor Company , Yung Sung Chuang , James Glass , Spoken Language Systems Group , Computer Science , International Conference , Research Office , Naval Research , Force Research Laboratory , Force Artificial Intelligence ,