Abstract
The worldwide proliferation of hate speech in digital spaces leads to users constantly absorbing toxic and hateful content. This can give harmful behaviors the semblance of normality, while reducing prosocial behaviors like empathy and compassion. We seek to study the effectiveness of AI generated counterspeech to scalably encourage prosociality across cultures. We intend to conduct a global longitudinal study of how various AI counterspeech methods affect participants’ perception of identity groups targeted by hate speech online and overall prosociality.
